From 49b7b9e21b0dd896764d9d12abb61983bd7896e9 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Thu, 5 Feb 2026 23:12:30 -0600
Subject: [PATCH 01/58] Fix: Connection capacity leak - failed stream
 initialization permanently consumes connection slot

Fixed four leak paths in the TS proxy streaming initialization:

1. url_utils.py - generate_stream_url(): Wrapped post-get_stream() code in try/except that calls release_stream() on any failure

2. views.py - stream_ts() retry loop: Added release_stream() before returning 503 to clean up the error-checking get_stream() call at line 181 which could INCR without a matching release when the retry loop times out

3. views.py - stream_ts() init failure: Added guarded release_stream() when initialize_channel() returns False, using a connection_allocated flag to prevent incorrect decrements when joining an existing channel

4. views.py - stream_ts() outer except: Added guarded release_stream() as a safety net for unexpected exceptions, only releasing when the flag confirms the slot was allocated and the channel hasn't been initialized yet

The connection_allocated flag prevents double-decrements by tracking whether request flow actually INCR'd the counter (fresh initialization) vs
returned cached results (joining an existing channel on another worker)
---
 apps/proxy/ts_proxy/url_utils.py | 60 +++++++++++++++++---------------
 apps/proxy/ts_proxy/views.py     | 23 ++++++++++--
 2 files changed, 51 insertions(+), 32 deletions(-)

diff --git a/apps/proxy/ts_proxy/url_utils.py b/apps/proxy/ts_proxy/url_utils.py
index 8b467b7f..9daaf4ef 100644
--- a/apps/proxy/ts_proxy/url_utils.py
+++ b/apps/proxy/ts_proxy/url_utils.py
@@ -129,39 +129,41 @@ def generate_stream_url(channel_id: str) -> Tuple[str, str, bool, Optional[int]]
             logger.error(f"No stream available for channel {channel_id}: {error_reason}")
             return None, None, False, None
 
-        # Look up the Stream and Profile objects
+        # get_stream() allocated a connection slot - ensure it's released on any error
         try:
+            # Look up the Stream and Profile objects
             stream = Stream.objects.get(id=stream_id)
             profile = M3UAccountProfile.objects.get(id=profile_id)
-        except (Stream.DoesNotExist, M3UAccountProfile.DoesNotExist) as e:
-            logger.error(f"Error getting stream or profile: {e}")
+
+            # Get the M3U account profile for URL pattern
+            m3u_profile = profile
+
+            # Get the appropriate user agent
+            m3u_account = M3UAccount.objects.get(id=m3u_profile.m3u_account.id)
+            stream_user_agent = m3u_account.get_user_agent().user_agent
+
+            if stream_user_agent is None:
+                stream_user_agent = UserAgent.objects.get(id=CoreSettings.get_default_user_agent_id())
+                logger.debug(f"No user agent found for account, using default: {stream_user_agent}")
+
+            # Generate stream URL based on the selected profile
+            input_url = stream.url
+            stream_url = transform_url(input_url, m3u_profile.search_pattern, m3u_profile.replace_pattern)
+
+            # Check if transcoding is needed
+            stream_profile = channel.get_stream_profile()
+            if stream_profile.is_proxy() or stream_profile is None:
+                transcode = False
+            else:
+                transcode = True
+
+            stream_profile_id = stream_profile.id
+
+            return stream_url, stream_user_agent, transcode, stream_profile_id
+        except Exception as e:
+            logger.error(f"Error generating stream URL for channel {channel_id}: {e}")
+            channel.release_stream()
             return None, None, False, None
-
-        # Get the M3U account profile for URL pattern
-        m3u_profile = profile
-
-        # Get the appropriate user agent
-        m3u_account = M3UAccount.objects.get(id=m3u_profile.m3u_account.id)
-        stream_user_agent = m3u_account.get_user_agent().user_agent
-
-        if stream_user_agent is None:
-            stream_user_agent = UserAgent.objects.get(id=CoreSettings.get_default_user_agent_id())
-            logger.debug(f"No user agent found for account, using default: {stream_user_agent}")
-
-        # Generate stream URL based on the selected profile
-        input_url = stream.url
-        stream_url = transform_url(input_url, m3u_profile.search_pattern, m3u_profile.replace_pattern)
-
-        # Check if transcoding is needed
-        stream_profile = channel.get_stream_profile()
-        if stream_profile.is_proxy() or stream_profile is None:
-            transcode = False
-        else:
-            transcode = True
-
-        stream_profile_id = stream_profile.id
-
-        return stream_url, stream_user_agent, transcode, stream_profile_id
     except Exception as e:
         logger.error(f"Error generating stream URL: {e}")
         return None, None, False, None
diff --git a/apps/proxy/ts_proxy/views.py b/apps/proxy/ts_proxy/views.py
index 91f254a7..6980f957 100644
--- a/apps/proxy/ts_proxy/views.py
+++ b/apps/proxy/ts_proxy/views.py
@@ -54,6 +54,7 @@ def stream_ts(request, channel_id):
 
     client_user_agent = None
     proxy_server = ProxyServer.get_instance()
+    connection_allocated = False  # Track if connection slot was allocated via get_stream()
 
     try:
         # Generate a unique client ID
@@ -219,9 +220,9 @@ def stream_ts(request, channel_id):
                     )
 
             if stream_url is None:
-                # Release the channel's stream lock if one was acquired
-                # Note: Only call this if get_stream() actually assigned a stream
-                # In our case, if stream_url is None, no stream was ever assigned, so don't release
+                # Release any connection slot that may have been allocated
+                # by the error-checking get_stream() call during retries
+                channel.release_stream()
 
                 # Get the specific error message if available
                 wait_duration = f"{int(time.time() - wait_start_time)}s"
@@ -237,6 +238,11 @@ def stream_ts(request, channel_id):
                     {"error": error_msg, "waited": wait_duration}, status=503
                 )  # 503 Service Unavailable is appropriate here
 
+            # generate_stream_url() called get_stream() which allocated a connection
+            # slot (INCR'd profile_connections) - track this for cleanup on error
+            if needs_initialization:
+                connection_allocated = True
+
             # Get the stream ID from the channel
             stream_id, m3u_profile_id, _ = channel.get_stream()
             logger.info(
@@ -344,10 +350,16 @@ def stream_ts(request, channel_id):
             )
 
             if not success:
+                if connection_allocated:
+                    channel.release_stream()
+                    connection_allocated = False
                 return JsonResponse(
                     {"error": "Failed to initialize channel"}, status=500
                 )
 
+            # Channel initialized - cleanup lifecycle now owns the connection release
+            connection_allocated = False
+
             # If we're the owner, wait for connection to establish
             if proxy_server.am_i_owner(channel_id):
                 manager = proxy_server.stream_managers.get(channel_id)
@@ -507,6 +519,11 @@ def stream_ts(request, channel_id):
 
     except Exception as e:
         logger.error(f"Error in stream_ts: {e}", exc_info=True)
+        if connection_allocated:
+            try:
+                channel.release_stream()
+            except Exception:
+                pass
         return JsonResponse({"error": str(e)}, status=500)
 
 

From 16309cd9b455daf98fd1bc5af213c83531111ea9 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Tue, 10 Feb 2026 18:15:48 -0600
Subject: [PATCH 02/58] Remove if logic from scheduled shutdown

- stream_generator.py: _schedule_channel_shutdown_if_needed() already guards itself, the if logic created conflict between release_stream() and stop_channel(), so if the stream was already released the shutdown would not be scheduled
---
 apps/proxy/ts_proxy/stream_generator.py | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/apps/proxy/ts_proxy/stream_generator.py b/apps/proxy/ts_proxy/stream_generator.py
index 50404f1d..771b0aa9 100644
--- a/apps/proxy/ts_proxy/stream_generator.py
+++ b/apps/proxy/ts_proxy/stream_generator.py
@@ -472,8 +472,7 @@ class StreamGenerator:
                 logger.error(f"Could not log client disconnect event: {e}")
 
             # Schedule channel shutdown if no clients left
-            if not stream_released:  # Only if we haven't already released the stream
-                self._schedule_channel_shutdown_if_needed(local_clients)
+            self._schedule_channel_shutdown_if_needed(local_clients)
 
     def _schedule_channel_shutdown_if_needed(self, local_clients):
         """

From 15d67c7f6518cd599823c57d070562da792f23d2 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Tue, 10 Feb 2026 18:30:43 -0600
Subject: [PATCH 03/58] Atomize profile counter swap in update_stream_profile
 to prevent counter drift

DECR/SET/INCR during stream profile switches are now executed in a Redis
pipeline, preventing partial updates if an exception occurs mid-operation
---
 apps/channels/models.py | 20 ++++++++++----------
 1 file changed, 10 insertions(+), 10 deletions(-)

diff --git a/apps/channels/models.py b/apps/channels/models.py
index ec5e135b..dd67095c 100644
--- a/apps/channels/models.py
+++ b/apps/channels/models.py
@@ -566,18 +566,18 @@ class Channel(models.Model):
         if current_profile_id == new_profile_id:
             return True
 
-        # Decrement connection count for old profile
+        # Use pipeline for atomic profile switch to prevent counter drift
+        # if an exception occurs between DECR and INCR
         old_profile_connections_key = f"profile_connections:{current_profile_id}"
-        old_count = int(redis_client.get(old_profile_connections_key) or 0)
-        if old_count > 0:
-            redis_client.decr(old_profile_connections_key)
-
-        # Update the profile mapping
-        redis_client.set(f"stream_profile:{stream_id}", new_profile_id)
-
-        # Increment connection count for new profile
         new_profile_connections_key = f"profile_connections:{new_profile_id}"
-        redis_client.incr(new_profile_connections_key)
+        old_count = int(redis_client.get(old_profile_connections_key) or 0)
+
+        pipe = redis_client.pipeline()
+        if old_count > 0:
+            pipe.decr(old_profile_connections_key)
+        pipe.set(f"stream_profile:{stream_id}", new_profile_id)
+        pipe.incr(new_profile_connections_key)
+        pipe.execute()
         logger.info(
             f"Updated stream {stream_id} profile from {current_profile_id} to {new_profile_id}"
         )

From 9e8f227f494eb0d1aa27b1359d8b470221a56e3a Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Tue, 10 Feb 2026 18:58:17 -0600
Subject: [PATCH 04/58] Fix: VOD proxy double-DECR of profile connection
 counter during cleanup

- multi_worker_connection_manager.py: Error handler at line 1042: removed connection_manager=self from cleanup() call since the DECR was already done at line 1036. Prevents double-DECR.

- connection_manager.py: Added connection_key_exists check before calling remove_connection(). The fallback DECR from session data now only fires when the connection tracking key has already expired. Prevents double-DECR.
---
 apps/proxy/vod_proxy/connection_manager.py    | 22 +++++++++++++------
 .../multi_worker_connection_manager.py        |  3 ++-
 2 files changed, 17 insertions(+), 8 deletions(-)

diff --git a/apps/proxy/vod_proxy/connection_manager.py b/apps/proxy/vod_proxy/connection_manager.py
index aafc75cd..42d69a11 100644
--- a/apps/proxy/vod_proxy/connection_manager.py
+++ b/apps/proxy/vod_proxy/connection_manager.py
@@ -1363,17 +1363,25 @@ class VODConnectionManager:
                         client_id = session_id
 
                         # Remove individual connection tracking keys created during streaming
+                        # Check if connection key exists first - remove_connection()
+                        # handles the profile DECR when the key exists, but skips it
+                        # if the key already expired by TTL
+                        connection_key_exists = False
                         if content_type and content_uuid:
+                            connection_key = self._get_connection_key(content_type, content_uuid, client_id)
+                            connection_key_exists = bool(self.redis_client.exists(connection_key))
                             logger.info(f"[{session_id}] Cleaning up connection tracking keys")
                             self.remove_connection(content_type, content_uuid, client_id)
 
-                        # Remove from profile connections if counted (additional safety check)
-                        if session_data.get(b'connection_counted') == b'True' and profile_id:
-                            profile_key = self._get_profile_connections_key(int(profile_id.decode('utf-8')))
-                            current_count = int(self.redis_client.get(profile_key) or 0)
-                            if current_count > 0:
-                                self.redis_client.decr(profile_key)
-                                logger.info(f"[{session_id}] Decremented profile {profile_id.decode('utf-8')} connections")
+                        # Fallback DECR: only if the connection tracking key had already
+                        # expired (remove_connection skips DECR in that case)
+                        if not connection_key_exists:
+                            if session_data.get(b'connection_counted') == b'True' and profile_id:
+                                profile_key = self._get_profile_connections_key(int(profile_id.decode('utf-8')))
+                                current_count = int(self.redis_client.get(profile_key) or 0)
+                                if current_count > 0:
+                                    self.redis_client.decr(profile_key)
+                                    logger.info(f"[{session_id}] Decremented profile {profile_id.decode('utf-8')} connections (fallback)")
 
                     # Remove session tracking key
                     self.redis_client.delete(session_key)
diff --git a/apps/proxy/vod_proxy/multi_worker_connection_manager.py b/apps/proxy/vod_proxy/multi_worker_connection_manager.py
index 1534f761..58d6bd04 100644
--- a/apps/proxy/vod_proxy/multi_worker_connection_manager.py
+++ b/apps/proxy/vod_proxy/multi_worker_connection_manager.py
@@ -1036,9 +1036,10 @@ class MultiWorkerVODConnectionManager:
                 self._decrement_profile_connections(m3u_profile.id)
 
                 # Also clean up the Redis connection state since we won't be using it
+                # Pass connection_manager=None since we already decremented above
                 if redis_connection:
                     try:
-                        redis_connection.cleanup(connection_manager=self, current_worker_id=self.worker_id)
+                        redis_connection.cleanup(current_worker_id=self.worker_id)
                     except Exception as cleanup_error:
                         logger.error(f"[{client_id}] Error during cleanup after connection failure: {cleanup_error}")
 

From cf96287c90a32120ad22aa7a17973ddae96a627a Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Tue, 10 Feb 2026 19:06:51 -0600
Subject: [PATCH 05/58] Add guard to remove_connection()

- multi_worker_connection_manager.py: Add missing count > 0 guard to multi-worker remove_connection()

This method doesn't appear to be called anywhere but added the guard in case it is put back into service
---
 apps/proxy/vod_proxy/multi_worker_connection_manager.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/apps/proxy/vod_proxy/multi_worker_connection_manager.py b/apps/proxy/vod_proxy/multi_worker_connection_manager.py
index 58d6bd04..63eeae00 100644
--- a/apps/proxy/vod_proxy/multi_worker_connection_manager.py
+++ b/apps/proxy/vod_proxy/multi_worker_connection_manager.py
@@ -1263,12 +1263,14 @@ class MultiWorkerVODConnectionManager:
                 profile_id = connection_data.get('m3u_profile_id')
                 if profile_id:
                     profile_connections_key = f"profile_connections:{profile_id}"
+                    current_count = int(self.redis_client.get(profile_connections_key) or 0)
 
                     # Use pipeline for atomic operations
                     pipe = self.redis_client.pipeline()
                     pipe.delete(connection_key)
                     pipe.srem(content_connections_key, client_id)
-                    pipe.decr(profile_connections_key)
+                    if current_count > 0:
+                        pipe.decr(profile_connections_key)
                     pipe.execute()
 
                     logger.info(f"Removed Redis-backed connection: {client_id}")

From 5a4b51b60a90d64d11a56df273607cc0be6ac72d Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Wed, 11 Feb 2026 18:08:24 -0600
Subject: [PATCH 06/58] Fix TOCTOU race condition in Channel.get_stream() with
 atomic slot reservation

Add _check_and_reserve_profile_slot() method that uses an INCR-first-then-check pattern to atomically reserve connection slots. This eliminates the TOCTOU race condition where separate GET > check > INCR operations allowed concurrent requests to both pass the capacity check, exceeding max_streams.

For profiles with max_streams=0 (unlimited), INCR is skipped entirely. If over capacity, the increment is rolled back with DECR.

Based on PR #838 by patchy8736, adapted from Lua script to native Redis commands for consistency with project style and easier debugging.
---
 apps/channels/models.py | 53 ++++++++++++++++++++++++++++++-----------
 1 file changed, 39 insertions(+), 14 deletions(-)

diff --git a/apps/channels/models.py b/apps/channels/models.py
index dd67095c..50538238 100644
--- a/apps/channels/models.py
+++ b/apps/channels/models.py
@@ -391,6 +391,38 @@ class Channel(models.Model):
 
         return stream_profile
 
+    def _check_and_reserve_profile_slot(self, profile, redis_client):
+        """
+        Atomically check and reserve a connection slot for the given profile.
+
+        Uses an INCR-first-then-check pattern to eliminate the TOCTOU race
+        condition where separate GET + check + INCR operations could allow
+        concurrent requests to both pass the capacity check.
+
+        For profiles with max_streams=0 (unlimited), no reservation is needed.
+
+        Args:
+            profile: M3UAccountProfile instance
+            redis_client: Redis client instance
+
+        Returns:
+            tuple: (reserved: bool, current_count: int)
+        """
+        if profile.max_streams == 0:
+            return (True, 0)
+
+        profile_connections_key = f"profile_connections:{profile.id}"
+
+        # Atomically increment first — this is a single Redis command
+        new_count = redis_client.incr(profile_connections_key)
+
+        if new_count <= profile.max_streams:
+            return (True, new_count)
+
+        # Over capacity — roll back the increment
+        redis_client.decr(profile_connections_key)
+        return (False, new_count - 1)
+
     def get_stream(self):
         """
         Finds an available stream for the requested channel and returns the selected stream and profile.
@@ -456,24 +488,16 @@ class Channel(models.Model):
             for profile in profiles:
                 has_active_profiles = True
 
-                profile_connections_key = f"profile_connections:{profile.id}"
-                current_connections = int(
-                    redis_client.get(profile_connections_key) or 0
+                # Atomically check and reserve a slot (INCR-first pattern)
+                reserved, current_count = self._check_and_reserve_profile_slot(
+                    profile, redis_client
                 )
 
-                # Check if profile has available slots (or unlimited connections)
-                if (
-                    profile.max_streams == 0
-                    or current_connections < profile.max_streams
-                ):
-                    # Start a new stream
+                if reserved:
+                    # Slot reserved — assign stream to this channel
                     redis_client.set(f"channel_stream:{self.id}", stream.id)
                     redis_client.set(f"stream_profile:{stream.id}", profile.id)
 
-                    # Increment connection count for profiles with limits
-                    if profile.max_streams > 0:
-                        redis_client.incr(profile_connections_key)
-
                     return (
                         stream.id,
                         profile.id,
@@ -483,7 +507,8 @@ class Channel(models.Model):
                     # This profile is at max connections
                     has_streams_but_maxed_out = True
                     logger.debug(
-                        f"Profile {profile.id} at max connections: {current_connections}/{profile.max_streams}"
+                        f"Profile {profile.id} at max connections: "
+                        f"{current_count}/{profile.max_streams}"
                     )
 
         # No available streams - determine specific reason

From 33f68a9808a1a29eed16afb6d85ab8b04ff58c85 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Wed, 11 Feb 2026 19:11:28 -0600
Subject: [PATCH 07/58] Remove unused Stream.get_stream() with TOCTOU
 vulnerability

Remove the Stream.get_stream() method which had the same TOCTOU race condition as Channel.get_stream()

All stream acquisition goes through
Channel.get_stream(), making this dead code. Removing it eliminates
a potential source of connection capacity leaks

Based on PR #838 by patchy8736
---
 apps/channels/models.py | 48 -----------------------------------------
 1 file changed, 48 deletions(-)

diff --git a/apps/channels/models.py b/apps/channels/models.py
index 50538238..998b7aca 100644
--- a/apps/channels/models.py
+++ b/apps/channels/models.py
@@ -202,54 +202,6 @@ class Stream(models.Model):
 
         return stream_profile
 
-    def get_stream(self):
-        """
-        Finds an available stream for the requested channel and returns the selected stream and profile.
-        """
-        redis_client = RedisClient.get_client()
-        profile_id = redis_client.get(f"stream_profile:{self.id}")
-        if profile_id:
-            profile_id = int(profile_id)
-            return self.id, profile_id, None
-
-        # Retrieve the M3U account associated with the stream.
-        m3u_account = self.m3u_account
-        m3u_profiles = m3u_account.profiles.all()
-        default_profile = next((obj for obj in m3u_profiles if obj.is_default), None)
-        profiles = [default_profile] + [
-            obj for obj in m3u_profiles if not obj.is_default
-        ]
-
-        for profile in profiles:
-            logger.info(profile)
-            # Skip inactive profiles
-            if profile.is_active == False:
-                continue
-
-            profile_connections_key = f"profile_connections:{profile.id}"
-            current_connections = int(redis_client.get(profile_connections_key) or 0)
-
-            # Check if profile has available slots (or unlimited connections)
-            if profile.max_streams == 0 or current_connections < profile.max_streams:
-                # Start a new stream
-                redis_client.set(f"channel_stream:{self.id}", self.id)
-                redis_client.set(
-                    f"stream_profile:{self.id}", profile.id
-                )  # Store only the matched profile
-
-                # Increment connection count for profiles with limits
-                if profile.max_streams > 0:
-                    redis_client.incr(profile_connections_key)
-
-                return (
-                    self.id,
-                    profile.id,
-                    None,
-                )  # Return newly assigned stream and matched profile
-
-        # 4. No available streams
-        return None, None, None
-
     def release_stream(self):
         """
         Called when a stream is finished to release the lock.

From 56719d93fe53445226a8c2dd812e9a23287f2ced Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Wed, 11 Feb 2026 19:16:01 -0600
Subject: [PATCH 08/58] Add boolean return and metadata fallback to
 Channel.release_stream()

Channel.release_stream() now returns True on success and False when no
stream/profile info could be found for cleanup. This allows callers to
detect and handle failed releases.

When the primary Redis keys (channel_stream/stream_profile) are already cleaned up by the proxy, the method falls back to the channel metadata hash (ts_proxy:channel:{uuid}:metadata) to recover stream_id and profile_id for proper connection counter cleanup.

Uses str(self.uuid) for metadata key lookup - fixes a bug in PR #838
which used self.id (integer PK), but metadata is keyed by UUID.

Enhanced logging now includes channel UUID for better traceability.

Based on PR #838 by patchy8736
---
 apps/channels/models.py | 58 ++++++++++++++++++++++++++++++++++++-----
 1 file changed, 52 insertions(+), 6 deletions(-)

diff --git a/apps/channels/models.py b/apps/channels/models.py
index 998b7aca..9e466cc7 100644
--- a/apps/channels/models.py
+++ b/apps/channels/models.py
@@ -3,6 +3,8 @@ from django.core.exceptions import ValidationError
 from django.conf import settings
 from core.models import StreamProfile, CoreSettings
 from core.utils import RedisClient
+from apps.proxy.ts_proxy.redis_keys import RedisKeys
+from apps.proxy.ts_proxy.constants import ChannelMetadataField
 import logging
 import uuid
 from datetime import datetime
@@ -476,32 +478,74 @@ class Channel(models.Model):
     def release_stream(self):
         """
         Called when a stream is finished to release the lock.
+
+        Returns:
+            bool: True if stream was successfully released, False if
+                  no stream/profile info could be found for cleanup.
         """
         redis_client = RedisClient.get_client()
 
         stream_id = redis_client.get(f"channel_stream:{self.id}")
         if not stream_id:
-            logger.debug("Invalid stream ID pulled from channel index")
-            return
+            # Primary key missing — try metadata hash fallback.
+            # The proxy may have already cleaned up channel_stream/stream_profile
+            # keys, but the metadata hash can still have the stream_id and profile.
+            metadata_key = RedisKeys.channel_metadata(str(self.uuid))
+            meta_stream_id = redis_client.hget(
+                metadata_key, ChannelMetadataField.STREAM_ID
+            )
+            meta_profile_id = redis_client.hget(
+                metadata_key, ChannelMetadataField.M3U_PROFILE
+            )
+
+            if meta_stream_id and meta_profile_id:
+                stream_id = int(meta_stream_id)
+                profile_id = int(meta_profile_id)
+                logger.debug(
+                    f"Channel {self.uuid}: recovered stream_id={stream_id}, "
+                    f"profile_id={profile_id} from metadata fallback"
+                )
+                # Clean up any remaining keys
+                redis_client.delete(f"channel_stream:{self.id}")
+                redis_client.delete(f"stream_profile:{stream_id}")
+
+                profile_connections_key = f"profile_connections:{profile_id}"
+                current_count = int(
+                    redis_client.get(profile_connections_key) or 0
+                )
+                if current_count > 0:
+                    redis_client.decr(profile_connections_key)
+                return True
+
+            logger.debug(
+                f"Channel {self.uuid}: no stream info found in primary keys "
+                f"or metadata fallback"
+            )
+            return False
 
         redis_client.delete(f"channel_stream:{self.id}")  # Remove active stream
 
         stream_id = int(stream_id)
         logger.debug(
-            f"Found stream ID {stream_id} associated with channel stream {self.id}"
+            f"Channel {self.uuid}: found stream_id={stream_id} for "
+            f"channel_stream:{self.id}"
         )
 
         # Get the matched profile for cleanup
         profile_id = redis_client.get(f"stream_profile:{stream_id}")
         if not profile_id:
-            logger.debug("Invalid profile ID pulled from stream index")
-            return
+            logger.debug(
+                f"Channel {self.uuid}: no profile found for "
+                f"stream_profile:{stream_id}"
+            )
+            return False
 
         redis_client.delete(f"stream_profile:{stream_id}")  # Remove profile association
 
         profile_id = int(profile_id)
         logger.debug(
-            f"Found profile ID {profile_id} associated with stream {stream_id}"
+            f"Channel {self.uuid}: found profile_id={profile_id} for "
+            f"stream {stream_id}"
         )
 
         profile_connections_key = f"profile_connections:{profile_id}"
@@ -511,6 +555,8 @@ class Channel(models.Model):
         if current_count > 0:
             redis_client.decr(profile_connections_key)
 
+        return True
+
     def update_stream_profile(self, new_profile_id):
         """
         Updates the profile for the current stream and adjusts connection counts.

From 6826e94921b0d7fa49a032d161059c014ce2ddde Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Wed, 11 Feb 2026 19:27:45 -0600
Subject: [PATCH 09/58] Add boolean return to Stream.release_stream()

Stream.release_stream() now returns True on success and False when no
profile info could be found in stream_profile:{stream_id}. This allows
callers to detect and handle failed releases.

Enhanced logging to include stream ID for better traceability.

Based on PR #838 by patchy8736
---
 apps/channels/models.py | 15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)

diff --git a/apps/channels/models.py b/apps/channels/models.py
index 9e466cc7..b6cb6ab0 100644
--- a/apps/channels/models.py
+++ b/apps/channels/models.py
@@ -207,6 +207,10 @@ class Stream(models.Model):
     def release_stream(self):
         """
         Called when a stream is finished to release the lock.
+
+        Returns:
+            bool: True if stream was successfully released, False if
+                  no profile info could be found for cleanup.
         """
         redis_client = RedisClient.get_client()
 
@@ -214,14 +218,17 @@ class Stream(models.Model):
         # Get the matched profile for cleanup
         profile_id = redis_client.get(f"stream_profile:{stream_id}")
         if not profile_id:
-            logger.debug("Invalid profile ID pulled from stream index")
-            return
+            logger.debug(
+                f"Stream {stream_id}: no profile found in "
+                f"stream_profile:{stream_id}"
+            )
+            return False
 
         redis_client.delete(f"stream_profile:{stream_id}")  # Remove profile association
 
         profile_id = int(profile_id)
         logger.debug(
-            f"Found profile ID {profile_id} associated with stream {stream_id}"
+            f"Stream {stream_id}: found profile_id={profile_id}"
         )
 
         profile_connections_key = f"profile_connections:{profile_id}"
@@ -231,6 +238,8 @@ class Stream(models.Model):
         if current_count > 0:
             redis_client.decr(profile_connections_key)
 
+        return True
+
 
 class ChannelManager(models.Manager):
     def active(self):

From fe17d0d5850c0b58e2cea150d97202e8e713afdb Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Wed, 11 Feb 2026 19:49:11 -0600
Subject: [PATCH 10/58] Check release_stream() boolean return at all call sites

All 12 call sites for Channel.release_stream() and Stream.release_stream() now check the boolean return value and log warnings when release fails.

Files updated:
- server.py: 4 call sites (zombie cleanup + _clean_redis_keys)
- channel_service.py: 2 call sites (stop_channel)
- stream_generator.py: 1 call site (last client disconnect)
- views.py: 4 call sites (failed init, redirect, init failure, exception)
- url_utils.py: 1 call site (URL generation error)

Also fixed bare except clause in _clean_redis_keys and uncaught
Stream.DoesNotExist in channel_service.stop_channel.

Based on PR #838 by patchy8736
---
 apps/proxy/ts_proxy/server.py                 | 25 +++++++++++++------
 .../ts_proxy/services/channel_service.py      | 22 ++++++++++------
 apps/proxy/ts_proxy/stream_generator.py       |  8 +++---
 apps/proxy/ts_proxy/url_utils.py              |  3 ++-
 apps/proxy/ts_proxy/views.py                  | 12 ++++++---
 5 files changed, 47 insertions(+), 23 deletions(-)

diff --git a/apps/proxy/ts_proxy/server.py b/apps/proxy/ts_proxy/server.py
index df51744d..feef4957 100644
--- a/apps/proxy/ts_proxy/server.py
+++ b/apps/proxy/ts_proxy/server.py
@@ -788,13 +788,17 @@ class ProxyServer:
             # Force release resources in the Channel model
             try:
                 channel = Channel.objects.get(uuid=channel_id)
-                channel.release_stream()
-                logger.info(f"Released stream allocation for zombie channel {channel_id}")
+                if not channel.release_stream():
+                    logger.warning(f"Failed to release stream for zombie channel {channel_id}")
+                else:
+                    logger.info(f"Released stream allocation for zombie channel {channel_id}")
             except Exception as e:
                 try:
                     stream = Stream.objects.get(stream_hash=channel_id)
-                    stream.release_stream()
-                    logger.info(f"Released stream allocation for zombie channel {channel_id}")
+                    if not stream.release_stream():
+                        logger.warning(f"Failed to release stream for zombie channel {channel_id}")
+                    else:
+                        logger.info(f"Released stream allocation for zombie channel {channel_id}")
                 except Exception as e:
                     logger.error(f"Error releasing stream for zombie channel {channel_id}: {e}")
 
@@ -1339,10 +1343,15 @@ class ProxyServer:
         # Release the channel, stream, and profile keys from the channel
         try:
             channel = Channel.objects.get(uuid=channel_id)
-            channel.release_stream()
-        except:
-            stream = Stream.objects.get(stream_hash=channel_id)
-            stream.release_stream()
+            if not channel.release_stream():
+                logger.debug(f"Channel {channel_id}: release_stream found no keys to clean")
+        except Channel.DoesNotExist:
+            try:
+                stream = Stream.objects.get(stream_hash=channel_id)
+                if not stream.release_stream():
+                    logger.debug(f"Stream {channel_id}: release_stream found no keys to clean")
+            except Stream.DoesNotExist:
+                logger.debug(f"No Channel or Stream found for {channel_id}")
 
         if not self.redis_client:
             return 0
diff --git a/apps/proxy/ts_proxy/services/channel_service.py b/apps/proxy/ts_proxy/services/channel_service.py
index 4c4a73ac..efe07c7e 100644
--- a/apps/proxy/ts_proxy/services/channel_service.py
+++ b/apps/proxy/ts_proxy/services/channel_service.py
@@ -265,15 +265,23 @@ class ChannelService:
         # Release the channel in the channel model if applicable
         try:
             channel = Channel.objects.get(uuid=channel_id)
-            channel.release_stream()
-            logger.info(f"Released channel {channel_id} stream allocation")
-            model_released = True
+            model_released = channel.release_stream()
+            if model_released:
+                logger.info(f"Released channel {channel_id} stream allocation")
+            else:
+                logger.warning(f"Channel {channel_id}: release_stream found no keys to clean")
         except Channel.DoesNotExist:
             logger.warning(f"Could not find Channel model for UUID {channel_id}, attempting stream hash")
-            stream = Stream.objects.get(stream_hash=channel_id)
-            stream.release_stream()
-            logger.info(f"Released stream {channel_id} stream allocation")
-            model_released = True
+            try:
+                stream = Stream.objects.get(stream_hash=channel_id)
+                model_released = stream.release_stream()
+                if model_released:
+                    logger.info(f"Released stream {channel_id} stream allocation")
+                else:
+                    logger.warning(f"Stream {channel_id}: release_stream found no keys to clean")
+            except Stream.DoesNotExist:
+                logger.error(f"No Channel or Stream found for {channel_id}")
+                model_released = False
         except Exception as e:
             logger.error(f"Error releasing channel stream: {e}")
             model_released = False
diff --git a/apps/proxy/ts_proxy/stream_generator.py b/apps/proxy/ts_proxy/stream_generator.py
index 771b0aa9..3c2a66fe 100644
--- a/apps/proxy/ts_proxy/stream_generator.py
+++ b/apps/proxy/ts_proxy/stream_generator.py
@@ -441,9 +441,11 @@ class StreamGenerator:
                                 try:
                                     # Get the channel by UUID
                                     channel = Channel.objects.get(uuid=self.channel_id)
-                                    channel.release_stream()
-                                    stream_released = True
-                                    logger.debug(f"[{self.client_id}] Released stream for channel {self.channel_id}")
+                                    stream_released = channel.release_stream()
+                                    if stream_released:
+                                        logger.debug(f"[{self.client_id}] Released stream for channel {self.channel_id}")
+                                    else:
+                                        logger.warning(f"[{self.client_id}] release_stream found no keys for channel {self.channel_id}")
                                 except Exception as e:
                                     logger.error(f"[{self.client_id}] Error releasing stream for channel {self.channel_id}: {e}")
             except Exception as e:
diff --git a/apps/proxy/ts_proxy/url_utils.py b/apps/proxy/ts_proxy/url_utils.py
index 9daaf4ef..fe06ad48 100644
--- a/apps/proxy/ts_proxy/url_utils.py
+++ b/apps/proxy/ts_proxy/url_utils.py
@@ -162,7 +162,8 @@ def generate_stream_url(channel_id: str) -> Tuple[str, str, bool, Optional[int]]
             return stream_url, stream_user_agent, transcode, stream_profile_id
         except Exception as e:
             logger.error(f"Error generating stream URL for channel {channel_id}: {e}")
-            channel.release_stream()
+            if not channel.release_stream():
+                logger.warning(f"Failed to release stream for channel {channel_id} after URL generation error")
             return None, None, False, None
     except Exception as e:
         logger.error(f"Error generating stream URL: {e}")
diff --git a/apps/proxy/ts_proxy/views.py b/apps/proxy/ts_proxy/views.py
index 6980f957..5060a3a7 100644
--- a/apps/proxy/ts_proxy/views.py
+++ b/apps/proxy/ts_proxy/views.py
@@ -222,7 +222,8 @@ def stream_ts(request, channel_id):
             if stream_url is None:
                 # Release any connection slot that may have been allocated
                 # by the error-checking get_stream() call during retries
-                channel.release_stream()
+                if not channel.release_stream():
+                    logger.debug(f"[{client_id}] release_stream found no keys during failed init cleanup")
 
                 # Get the specific error message if available
                 wait_duration = f"{int(time.time() - wait_start_time)}s"
@@ -314,7 +315,8 @@ def stream_ts(request, channel_id):
                                 f"[{client_id}] Alternate stream #{alt['stream_id']} failed validation: {message}"
                             )
                 # Release stream lock before redirecting
-                channel.release_stream()
+                if not channel.release_stream():
+                    logger.warning(f"[{client_id}] Failed to release stream before redirect")
                 # Final decision based on validation results
                 if is_valid:
                     logger.info(
@@ -351,7 +353,8 @@ def stream_ts(request, channel_id):
 
             if not success:
                 if connection_allocated:
-                    channel.release_stream()
+                    if not channel.release_stream():
+                        logger.warning(f"[{client_id}] Failed to release stream after init failure")
                     connection_allocated = False
                 return JsonResponse(
                     {"error": "Failed to initialize channel"}, status=500
@@ -521,7 +524,8 @@ def stream_ts(request, channel_id):
         logger.error(f"Error in stream_ts: {e}", exc_info=True)
         if connection_allocated:
             try:
-                channel.release_stream()
+                if not channel.release_stream():
+                    logger.warning(f"[{client_id}] Failed to release stream in exception handler")
             except Exception:
                 pass
         return JsonResponse({"error": str(e)}, status=500)

From 6360271e2830831fbfca9bf5c938e29dd25d7ef5 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Wed, 11 Feb 2026 20:09:17 -0600
Subject: [PATCH 11/58] Fix unprotected _clean_redis_keys() call in
 check_if_channel_exists

Wrap _clean_redis_keys() call in check_if_channel_exists() with
try/except to prevent exception propagation when Channel/Stream
lookup or release_stream() encounters DB or Redis errors.
---
 apps/proxy/ts_proxy/server.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/apps/proxy/ts_proxy/server.py b/apps/proxy/ts_proxy/server.py
index feef4957..3a6e8ae3 100644
--- a/apps/proxy/ts_proxy/server.py
+++ b/apps/proxy/ts_proxy/server.py
@@ -766,7 +766,10 @@ class ProxyServer:
                 if self.redis_client.exists(key):
                     # Found orphaned keys without metadata - clean them up
                     logger.warning(f"Found orphaned keys for channel {channel_id} without metadata - cleaning up")
-                    self._clean_redis_keys(channel_id)
+                    try:
+                        self._clean_redis_keys(channel_id)
+                    except Exception as e:
+                        logger.error(f"Error cleaning redis keys for channel {channel_id}: {e}")
                     return False
 
         return False

From 8bb8a0fca904ccfc3c554ac6702e88ab94d3f094 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Wed, 11 Feb 2026 20:52:30 -0600
Subject: [PATCH 12/58] fix(ts-proxy): Remove redundant local import causing
 UnboundLocalError

The local `from apps.channels.models import Channel` inside a
conditional block in _handle_client_disconnect() shadowed the top-level import. When the condition was False, Python's function-level scoping caused line 462's Channel.objects.get() to raise UnboundLocalError. Channel is already imported at module level (line 11), making the local import unnecessary.

Discovered when testing proxy changes
---
 apps/proxy/ts_proxy/stream_generator.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/apps/proxy/ts_proxy/stream_generator.py b/apps/proxy/ts_proxy/stream_generator.py
index 3c2a66fe..875d9d0e 100644
--- a/apps/proxy/ts_proxy/stream_generator.py
+++ b/apps/proxy/ts_proxy/stream_generator.py
@@ -437,7 +437,6 @@ class StreamGenerator:
                             client_count = proxy_server.client_managers[self.channel_id].get_total_client_count()
                             # Only the last client or owner should release the stream
                             if client_count <= 1 and proxy_server.am_i_owner(self.channel_id):
-                                from apps.channels.models import Channel
                                 try:
                                     # Get the channel by UUID
                                     channel = Channel.objects.get(uuid=self.channel_id)

From 837ba847685bb93f6d9351720e68c3ca19c6d04e Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Wed, 11 Feb 2026 23:57:10 -0600
Subject: [PATCH 13/58] fix(ts-proxy): Restore Stream.get_stream() with atomic
 slot reservation

Stream.get_stream() was incorrectly removed as dead code in commit
33f68a98. It is actually called get_stream_object()
in views.py when a stream hash (not channel UUID) is used, such as
during stream preview. The variable is named channel but holds a
Stream instance, which masked the dependency during analysis.

Restored with the INCR-first-then-check pattern applied, fixing the
same TOCTOU vulnerability the original method had.
---
 apps/channels/models.py | 46 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 46 insertions(+)

diff --git a/apps/channels/models.py b/apps/channels/models.py
index b6cb6ab0..b9c1dde1 100644
--- a/apps/channels/models.py
+++ b/apps/channels/models.py
@@ -204,6 +204,52 @@ class Stream(models.Model):
 
         return stream_profile
 
+    def get_stream(self):
+        """
+        Finds an available profile for this stream and reserves a connection slot.
+
+        Returns:
+            Tuple[Optional[int], Optional[int], Optional[str]]: (stream_id, profile_id, error_reason)
+        """
+        redis_client = RedisClient.get_client()
+        profile_id = redis_client.get(f"stream_profile:{self.id}")
+        if profile_id:
+            profile_id = int(profile_id)
+            return self.id, profile_id, None
+
+        # Retrieve the M3U account associated with the stream.
+        m3u_account = self.m3u_account
+        m3u_profiles = m3u_account.profiles.all()
+        default_profile = next((obj for obj in m3u_profiles if obj.is_default), None)
+        profiles = [default_profile] + [
+            obj for obj in m3u_profiles if not obj.is_default
+        ]
+
+        for profile in profiles:
+            logger.info(profile)
+            # Skip inactive profiles
+            if profile.is_active == False:
+                continue
+
+            # Atomic slot reservation: INCR first, check, rollback if over capacity
+            if profile.max_streams == 0:
+                reserved = True
+            else:
+                profile_connections_key = f"profile_connections:{profile.id}"
+                new_count = redis_client.incr(profile_connections_key)
+                if new_count <= profile.max_streams:
+                    reserved = True
+                else:
+                    redis_client.decr(profile_connections_key)
+                    reserved = False
+
+            if reserved:
+                redis_client.set(f"channel_stream:{self.id}", self.id)
+                redis_client.set(f"stream_profile:{self.id}", profile.id)
+                return self.id, profile.id, None
+
+        return None, None, "All active M3U profiles have reached maximum connection limits"
+
     def release_stream(self):
         """
         Called when a stream is finished to release the lock.

From 20ce45f7adfd7a0f959207f09125accec2e93806 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Thu, 12 Feb 2026 00:02:40 -0600
Subject: [PATCH 14/58] fix(ts-proxy): Add Stream fallback in cleanup path to
 prevent counter leak on stream preview

When previewing a stream directly (via stream hash), the cleanup in stream_generator.py tried Channel.objects.get(uuid=stream_hash) which failed with ValidationError since the hash is not a UUID. This
prevented release_stream() from running, permanently leaking the
connection counter. With max_streams=1, a single preview permanently blocked all subsequent streams.

Added Stream.objects.get(stream_hash=...) fallback so the cleanup
path works for both Channel (UUID) and Stream (hash) identifiers.
Also added Stream import to stream_generator.py.
---
 apps/proxy/ts_proxy/stream_generator.py | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/apps/proxy/ts_proxy/stream_generator.py b/apps/proxy/ts_proxy/stream_generator.py
index 875d9d0e..f43eb528 100644
--- a/apps/proxy/ts_proxy/stream_generator.py
+++ b/apps/proxy/ts_proxy/stream_generator.py
@@ -8,7 +8,7 @@ import logging
 import threading
 import gevent  # Add this import at the top of your file
 from apps.proxy.config import TSConfig as Config
-from apps.channels.models import Channel
+from apps.channels.models import Channel, Stream
 from core.utils import log_system_event
 from .server import ProxyServer
 from .utils import create_ts_packet, get_logger
@@ -438,9 +438,12 @@ class StreamGenerator:
                             # Only the last client or owner should release the stream
                             if client_count <= 1 and proxy_server.am_i_owner(self.channel_id):
                                 try:
-                                    # Get the channel by UUID
-                                    channel = Channel.objects.get(uuid=self.channel_id)
-                                    stream_released = channel.release_stream()
+                                    # Try Channel first (normal flow), fall back to Stream (preview flow)
+                                    try:
+                                        obj = Channel.objects.get(uuid=self.channel_id)
+                                    except (Channel.DoesNotExist, Exception):
+                                        obj = Stream.objects.get(stream_hash=self.channel_id)
+                                    stream_released = obj.release_stream()
                                     if stream_released:
                                         logger.debug(f"[{self.client_id}] Released stream for channel {self.channel_id}")
                                     else:

From 51b58247378a63de9adfa0dd075adb85f8cb5e44 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Thu, 12 Feb 2026 00:09:29 -0600
Subject: [PATCH 15/58] Fix cleanup paths for stream preview

Fix exception handling in _clean_redis_keys() and stop_channel()
   where `except Channel.DoesNotExist` didn't catch ValidationError from invalid UUID format, bypassing the Stream fallback path.
---
 apps/proxy/ts_proxy/server.py                   | 4 ++--
 apps/proxy/ts_proxy/services/channel_service.py | 9 +++------
 2 files changed, 5 insertions(+), 8 deletions(-)

diff --git a/apps/proxy/ts_proxy/server.py b/apps/proxy/ts_proxy/server.py
index 3a6e8ae3..d036bc2e 100644
--- a/apps/proxy/ts_proxy/server.py
+++ b/apps/proxy/ts_proxy/server.py
@@ -1348,12 +1348,12 @@ class ProxyServer:
             channel = Channel.objects.get(uuid=channel_id)
             if not channel.release_stream():
                 logger.debug(f"Channel {channel_id}: release_stream found no keys to clean")
-        except Channel.DoesNotExist:
+        except (Channel.DoesNotExist, Exception):
             try:
                 stream = Stream.objects.get(stream_hash=channel_id)
                 if not stream.release_stream():
                     logger.debug(f"Stream {channel_id}: release_stream found no keys to clean")
-            except Stream.DoesNotExist:
+            except (Stream.DoesNotExist, Exception):
                 logger.debug(f"No Channel or Stream found for {channel_id}")
 
         if not self.redis_client:
diff --git a/apps/proxy/ts_proxy/services/channel_service.py b/apps/proxy/ts_proxy/services/channel_service.py
index efe07c7e..e765ece3 100644
--- a/apps/proxy/ts_proxy/services/channel_service.py
+++ b/apps/proxy/ts_proxy/services/channel_service.py
@@ -270,7 +270,7 @@ class ChannelService:
                 logger.info(f"Released channel {channel_id} stream allocation")
             else:
                 logger.warning(f"Channel {channel_id}: release_stream found no keys to clean")
-        except Channel.DoesNotExist:
+        except (Channel.DoesNotExist, Exception):
             logger.warning(f"Could not find Channel model for UUID {channel_id}, attempting stream hash")
             try:
                 stream = Stream.objects.get(stream_hash=channel_id)
@@ -279,12 +279,9 @@ class ChannelService:
                     logger.info(f"Released stream {channel_id} stream allocation")
                 else:
                     logger.warning(f"Stream {channel_id}: release_stream found no keys to clean")
-            except Stream.DoesNotExist:
-                logger.error(f"No Channel or Stream found for {channel_id}")
+            except (Stream.DoesNotExist, Exception) as e:
+                logger.error(f"No Channel or Stream found for {channel_id}: {e}")
                 model_released = False
-        except Exception as e:
-            logger.error(f"Error releasing channel stream: {e}")
-            model_released = False
 
         return {
             'status': 'success',

From c69718b55987362403d468913dbe6cd7dd040954 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Tue, 17 Feb 2026 21:18:20 -0600
Subject: [PATCH 16/58] Fix: Make release_stream() idempotent to prevent
 double-DECR on rapid channel switching

Multiple code paths call release_stream() during channel shutdown (stream generator cleanup, _clean_redis_keys, ChannelService.stop_channel), causing
the profile_connections counter to be decremented multiple times for the same stream. The metadata hash fallback persisted longer than the primary keys, allowing subsequent calls to find stale stream_id/m3u_profile and DECR again.

Changes:
- Clear STREAM_ID and M3U_PROFILE metadata fields after decrementing in both the primary and fallback paths, making release_stream() safe to call multiple times
- Add metadata fallback when stream_profile:{id} key is missing but channel_stream:{id} exists, preventing a counter leak in that edge case
- Replace redundant get_stream() call in views.py with direct Redis reads to avoid side-effect INCR from a method that can allocate connection slots
---
 apps/channels/models.py      | 47 +++++++++++++++++++++++++++++-------
 apps/proxy/ts_proxy/views.py | 14 +++++++++--
 2 files changed, 50 insertions(+), 11 deletions(-)

diff --git a/apps/channels/models.py b/apps/channels/models.py
index b9c1dde1..217a92c0 100644
--- a/apps/channels/models.py
+++ b/apps/channels/models.py
@@ -564,6 +564,14 @@ class Channel(models.Model):
                 redis_client.delete(f"channel_stream:{self.id}")
                 redis_client.delete(f"stream_profile:{stream_id}")
 
+                # Clear metadata fields so duplicate release_stream() calls
+                # won't find them and DECR again
+                redis_client.hdel(
+                    metadata_key,
+                    ChannelMetadataField.STREAM_ID,
+                    ChannelMetadataField.M3U_PROFILE,
+                )
+
                 profile_connections_key = f"profile_connections:{profile_id}"
                 current_count = int(
                     redis_client.get(profile_connections_key) or 0
@@ -588,16 +596,27 @@ class Channel(models.Model):
 
         # Get the matched profile for cleanup
         profile_id = redis_client.get(f"stream_profile:{stream_id}")
-        if not profile_id:
-            logger.debug(
-                f"Channel {self.uuid}: no profile found for "
-                f"stream_profile:{stream_id}"
+        if profile_id:
+            redis_client.delete(f"stream_profile:{stream_id}")  # Remove profile association
+            profile_id = int(profile_id)
+        else:
+            # stream_profile key missing — try metadata hash fallback
+            metadata_key = RedisKeys.channel_metadata(str(self.uuid))
+            meta_profile_id = redis_client.hget(
+                metadata_key, ChannelMetadataField.M3U_PROFILE
             )
-            return False
-
-        redis_client.delete(f"stream_profile:{stream_id}")  # Remove profile association
-
-        profile_id = int(profile_id)
+            if meta_profile_id:
+                profile_id = int(meta_profile_id)
+                logger.debug(
+                    f"Channel {self.uuid}: recovered profile_id={profile_id} "
+                    f"from metadata fallback (stream_profile:{stream_id} was missing)"
+                )
+            else:
+                logger.warning(
+                    f"Channel {self.uuid}: no profile found for "
+                    f"stream_profile:{stream_id} or in metadata fallback"
+                )
+                return False
         logger.debug(
             f"Channel {self.uuid}: found profile_id={profile_id} for "
             f"stream {stream_id}"
@@ -610,6 +629,16 @@ class Channel(models.Model):
         if current_count > 0:
             redis_client.decr(profile_connections_key)
 
+        # Clear metadata fields so duplicate release_stream() calls
+        # (e.g. from _clean_redis_keys or ChannelService.stop_channel)
+        # won't find them via fallback and DECR again
+        metadata_key = RedisKeys.channel_metadata(str(self.uuid))
+        redis_client.hdel(
+            metadata_key,
+            ChannelMetadataField.STREAM_ID,
+            ChannelMetadataField.M3U_PROFILE,
+        )
+
         return True
 
     def update_stream_profile(self, new_profile_id):
diff --git a/apps/proxy/ts_proxy/views.py b/apps/proxy/ts_proxy/views.py
index 5060a3a7..90ac2673 100644
--- a/apps/proxy/ts_proxy/views.py
+++ b/apps/proxy/ts_proxy/views.py
@@ -244,8 +244,18 @@ def stream_ts(request, channel_id):
             if needs_initialization:
                 connection_allocated = True
 
-            # Get the stream ID from the channel
-            stream_id, m3u_profile_id, _ = channel.get_stream()
+            # Read stream assignment from Redis (already set by generate_stream_url → get_stream).
+            # Avoid calling get_stream() again — (INCR profile counter)
+            # It could double-allocate if the keys were cleared by a concurrent release.
+            stream_id = None
+            m3u_profile_id = None
+            if proxy_server.redis_client:
+                stream_id_bytes = proxy_server.redis_client.get(f"channel_stream:{channel.id}")
+                if stream_id_bytes:
+                    stream_id = int(stream_id_bytes)
+                    profile_id_bytes = proxy_server.redis_client.get(f"stream_profile:{stream_id}")
+                    if profile_id_bytes:
+                        m3u_profile_id = int(profile_id_bytes)
             logger.info(
                 f"Channel {channel_id} using stream ID {stream_id}, m3u account profile ID {m3u_profile_id}"
             )

From 9b6c793b1186da86c2860f74006ee8b42d23ad1c Mon Sep 17 00:00:00 2001
From: Endoze <endoze@users.noreply.github.com>
Date: Thu, 26 Feb 2026 09:41:57 -0500
Subject: [PATCH 17/58] fix(wsgi): ensure gevent monkey-patching before Django
 import chain

Celery's Redis connection pool was silently failing in modular
deployment mode because Django's setup imports celery.py before uWSGI's
gevent plugin reliably patches stdlib sockets. Moving patch_all() to the
top of wsgi.py guarantees all broker connections are gevent-aware
regardless of uWSGI initialization order.
---
 dispatcharr/tests/__init__.py  |  0
 dispatcharr/tests/test_wsgi.py | 39 ++++++++++++++++++++++++++++++++++
 dispatcharr/wsgi.py            | 15 +++++++++++++
 3 files changed, 54 insertions(+)
 create mode 100644 dispatcharr/tests/__init__.py
 create mode 100644 dispatcharr/tests/test_wsgi.py

diff --git a/dispatcharr/tests/__init__.py b/dispatcharr/tests/__init__.py
new file mode 100644
index 00000000..e69de29b
diff --git a/dispatcharr/tests/test_wsgi.py b/dispatcharr/tests/test_wsgi.py
new file mode 100644
index 00000000..cdf7a2e9
--- /dev/null
+++ b/dispatcharr/tests/test_wsgi.py
@@ -0,0 +1,39 @@
+"""Tests for dispatcharr/wsgi.py gevent monkey-patching."""
+
+import os
+import subprocess
+import sys
+import unittest
+
+
+class TestWSGIGeventPatching(unittest.TestCase):
+    """Verify that wsgi.py applies gevent monkey-patching before Django imports.
+
+    These tests run in subprocesses because monkey.patch_all() must execute
+    before any other imports.  By the time Django's test runner loads this
+    file, ssl/socket are already imported, so calling patch_all() in-process
+    would fail.  A subprocess mirrors how uWSGI actually loads wsgi.py.
+    """
+
+    def test_socket_is_patched_after_wsgi_import(self):
+        """Loading wsgi.py first (as uWSGI does) should patch sockets."""
+        result = subprocess.run(
+            [
+                sys.executable, "-c",
+                "import dispatcharr.wsgi; "
+                "from gevent import monkey; "
+                "assert monkey.is_module_patched('socket'), "
+                "'socket module was not patched by wsgi.py'; "
+                "print('PASS')",
+            ],
+            capture_output=True,
+            text=True,
+            env={**os.environ, "DJANGO_SETTINGS_MODULE": "dispatcharr.settings"},
+            timeout=60,
+            cwd="/app",
+        )
+        self.assertEqual(
+            result.returncode,
+            0,
+            f"monkey-patching check failed:\n{result.stderr}",
+        )
diff --git a/dispatcharr/wsgi.py b/dispatcharr/wsgi.py
index 29c2ba01..6e8860af 100644
--- a/dispatcharr/wsgi.py
+++ b/dispatcharr/wsgi.py
@@ -1,6 +1,21 @@
 """
 WSGI config for dispatcharr project.
 """
+# When running under uWSGI with gevent, ensure monkey-patching is fully
+# applied before any other imports.  Django's setup triggers the import of
+# dispatcharr/__init__.py → celery.py, which initialises the Celery broker
+# transport.  If those imports happen before sockets are patched, Celery's
+# Redis connection pool holds unpatched sockets that silently fail under
+# gevent's cooperative scheduling.  Calling patch_all() here—before the
+# Django import chain—guarantees every socket (including pooled broker
+# connections) is gevent-aware.  The call is idempotent, so it's harmless
+# if uWSGI's gevent plugin has already patched.
+try:
+    from gevent import monkey
+    monkey.patch_all()
+except ImportError:
+    pass
+
 import os
 from django.core.wsgi import get_wsgi_application
 

From 8bd38ad71c455317d9a0753887b223bd3321fee5 Mon Sep 17 00:00:00 2001
From: PFalko <falko.poeritz@gmail.com>
Date: Fri, 27 Feb 2026 16:40:28 +0100
Subject: [PATCH 18/58] Fix stream ownership bugs causing streams to die after
 30-200s
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Three interrelated bugs cause TS proxy streams to terminate prematurely
in multi-worker uWSGI deployments:

1. Double ProxyServer instantiation: ProxyConfig.ready() in apps/proxy/apps.py
   calls TSProxyServer() directly, bypassing get_instance(). The subsequent
   TSProxyConfig.ready() call to get_instance() creates a second instance.
   Each instance starts its own cleanup thread, but only one holds channel
   data — the orphaned cleanup thread cannot extend ownership.

2. Redis flushdb() on every client init: RedisClient.get_client() in
   core/utils.py calls flushdb() whenever a new connection is created.
   Celery autoscale workers spawning mid-stream nuke all Redis keys
   including ownership, client records, and channel metadata.

3. No recovery from expired ownership: get_channel_owner() has a TOCTOU
   bug (two separate GET calls in a lambda). extend_ownership() silently
   fails when keys expire. Non-owner cleanup unconditionally kills streams
   even when the worker holds the stream_manager.

Fixes:
- Use TSProxyServer.get_instance() in ProxyConfig.ready()
- Remove flushdb() from Redis client initialization
- Use sentinel pattern for gevent-safe singleton (threading.Lock does not
  work with gevent greenlets)
- Single GET in get_channel_owner() to avoid TOCTOU race
- Re-acquire expired ownership keys in extend_ownership()
- Attempt re-acquisition before cleanup in non-owner path

Relates to #992, #980
---
 apps/proxy/apps.py            |  4 +--
 apps/proxy/ts_proxy/server.py | 66 +++++++++++++++++++++++++++++------
 core/utils.py                 |  1 -
 3 files changed, 58 insertions(+), 13 deletions(-)

diff --git a/apps/proxy/apps.py b/apps/proxy/apps.py
index d1c8b966..b33d46cc 100644
--- a/apps/proxy/apps.py
+++ b/apps/proxy/apps.py
@@ -12,6 +12,6 @@ class ProxyConfig(AppConfig):
             from .hls_proxy.server import ProxyServer as HLSProxyServer
             from .ts_proxy.server import ProxyServer as TSProxyServer
 
-            # Initialize proxy servers
+            # Initialize proxy servers (TS uses singleton to prevent duplicate instances)
             self.hls_proxy = HLSProxyServer()
-            self.ts_proxy = TSProxyServer()
+            self.ts_proxy = TSProxyServer.get_instance()
diff --git a/apps/proxy/ts_proxy/server.py b/apps/proxy/ts_proxy/server.py
index df51744d..b28154b3 100644
--- a/apps/proxy/ts_proxy/server.py
+++ b/apps/proxy/ts_proxy/server.py
@@ -34,18 +34,29 @@ logger = get_logger()
 class ProxyServer:
     """Manages TS proxy server instance with worker coordination"""
     _instance = None
+    _INITIALIZING = object()  # sentinel for gevent-safe singleton
 
     @classmethod
     def get_instance(cls):
-        if cls._instance is None:
+        inst = cls._instance
+        if inst is not None and inst is not cls._INITIALIZING:
+            return inst
+        if inst is None:
+            cls._instance = cls._INITIALIZING
             from .server import ProxyServer
             from .stream_manager import StreamManager
             from .stream_buffer import StreamBuffer
             from .client_manager import ClientManager
-
-            cls._instance = ProxyServer()
-
-        return cls._instance
+            real_instance = ProxyServer()
+            cls._instance = real_instance
+            return real_instance
+        # Another greenlet is initializing — wait for completion
+        import time
+        while True:
+            inst = cls._instance
+            if inst is not None and inst is not cls._INITIALIZING:
+                return inst
+            time.sleep(0.05)
 
     def __init__(self):
         """Initialize proxy server with worker identification"""
@@ -353,9 +364,16 @@ class ProxyServer:
 
         try:
             lock_key = RedisKeys.channel_owner(channel_id)
-            return self._execute_redis_command(
-                lambda: self.redis_client.get(lock_key).decode('utf-8') if self.redis_client.get(lock_key) else None
+            result = self._execute_redis_command(
+                lambda: self.redis_client.get(lock_key)
             )
+            if result is None:
+                return None
+            try:
+                return result.decode('utf-8')
+            except (AttributeError, UnicodeDecodeError) as e:
+                logger.error(f"Error decoding channel owner for {channel_id}: {e}, raw={result!r}")
+                return None
         except Exception as e:
             logger.error(f"Error getting channel owner: {e}")
             return None
@@ -433,7 +451,7 @@ class ProxyServer:
             logger.error(f"Error releasing channel ownership: {e}")
 
     def extend_ownership(self, channel_id, ttl=30):
-        """Extend ownership lease with grace period"""
+        """Extend ownership lease, re-acquiring if key expired"""
         if not self.redis_client:
             return False
 
@@ -441,10 +459,24 @@ class ProxyServer:
             lock_key = RedisKeys.channel_owner(channel_id)
             current = self.redis_client.get(lock_key)
 
-            # Only extend if we're still the owner
-            if current and current.decode('utf-8') == self.worker_id:
+            if current is None:
+                # Key expired — re-acquire if we have the stream_manager
+                if channel_id in self.stream_managers:
+                    acquired = self.redis_client.setnx(lock_key, self.worker_id)
+                    if acquired:
+                        self.redis_client.expire(lock_key, ttl)
+                        logger.warning(f"Re-acquired expired ownership for channel {channel_id}")
+                        return True
+                    else:
+                        new_owner = self.redis_client.get(lock_key)
+                        logger.warning(f"Could not re-acquire ownership for {channel_id}, new owner: {new_owner}")
+                        return False
+                return False
+
+            if current.decode('utf-8') == self.worker_id:
                 self.redis_client.expire(lock_key, ttl)
                 return True
+
             return False
         except Exception as e:
             logger.error(f"Error extending ownership: {e}")
@@ -1173,6 +1205,20 @@ class ProxyServer:
 
                         else:
                             # === NON-OWNER CHANNEL HANDLING ===
+                            # Safety: if we have a stream_manager, we ARE the real owner
+                            # but the Redis key may have expired. Try to re-acquire.
+                            if channel_id in self.stream_managers:
+                                logger.warning(
+                                    f"Ownership gap for {channel_id}: this worker has stream_manager "
+                                    f"but am_i_owner returned False. Attempting re-acquisition."
+                                )
+                                reacquired = self.extend_ownership(channel_id)
+                                if reacquired:
+                                    logger.info(f"Successfully re-acquired ownership for {channel_id}")
+                                    continue
+                                else:
+                                    logger.error(f"Failed to re-acquire ownership for {channel_id}, will clean up")
+
                             # For channels we don't own, check if they've been stopped/cleaned up in Redis
                             if self.redis_client:
                                 # Method 1: Check for stopping key
diff --git a/core/utils.py b/core/utils.py
index a5ea41df..f0e02bb4 100644
--- a/core/utils.py
+++ b/core/utils.py
@@ -81,7 +81,6 @@ class RedisClient:
 
                     # Validate connection with ping
                     client.ping()
-                    client.flushdb()
 
                     # Disable persistence on first connection - improves performance
                     # Only try to disable if not in a read-only environment

From a318d919d8263110fbd088222edc6f69ce1d6b17 Mon Sep 17 00:00:00 2001
From: Jonathan Caicedo <jonathan@jcaicedo.com>
Date: Thu, 26 Feb 2026 09:19:19 -0500
Subject: [PATCH 19/58] feat: filter channels that contain stale streams

closes: #1025
---
 apps/channels/api_views.py                    |  4 +++
 .../src/components/tables/ChannelsTable.jsx   | 24 +++++++++++++----
 .../ChannelsTable/ChannelTableHeader.jsx      | 27 ++++++++++++++++++-
 frontend/src/components/tables/table.css      |  2 +-
 4 files changed, 50 insertions(+), 7 deletions(-)

diff --git a/apps/channels/api_views.py b/apps/channels/api_views.py
index d0fb3d97..b77cf60d 100644
--- a/apps/channels/api_views.py
+++ b/apps/channels/api_views.py
@@ -571,6 +571,7 @@ class ChannelViewSet(viewsets.ModelViewSet):
         channel_profile_id = self.request.query_params.get("channel_profile_id")
         show_disabled_param = self.request.query_params.get("show_disabled", None)
         only_streamless = self.request.query_params.get("only_streamless", None)
+        only_stale = self.request.query_params.get("only_stale", None)
 
         if channel_profile_id:
             try:
@@ -590,6 +591,9 @@ class ChannelViewSet(viewsets.ModelViewSet):
 
         if only_streamless:
             q_filters &= Q(streams__isnull=True)
+        if only_stale:
+            # Filter channels that have at least one related stream marked as stale
+            q_filters &= Q(streams__is_stale=True)
 
         if self.request.user.user_level < 10:
             filters["user_level__lte"] = self.request.user.user_level
diff --git a/frontend/src/components/tables/ChannelsTable.jsx b/frontend/src/components/tables/ChannelsTable.jsx
index 06fe652a..75bf0e33 100644
--- a/frontend/src/components/tables/ChannelsTable.jsx
+++ b/frontend/src/components/tables/ChannelsTable.jsx
@@ -321,6 +321,7 @@ const ChannelsTable = ({ onReady }) => {
   const [showDisabled, setShowDisabled] = useState(true);
   const [showOnlyStreamlessChannels, setShowOnlyStreamlessChannels] =
     useState(false);
+  const [showOnlyStaleChannels, setShowOnlyStaleChannels] = useState(false);
 
   const [paginationString, setPaginationString] = useState('');
   const [filters, setFilters] = useState({
@@ -424,6 +425,9 @@ const ChannelsTable = ({ onReady }) => {
     if (showOnlyStreamlessChannels === true) {
       params.append('only_streamless', true);
     }
+    if (showOnlyStaleChannels === true) {
+      params.append('only_stale', true);
+    }
 
     // Apply sorting
     if (sorting.length > 0) {
@@ -511,6 +515,7 @@ const ChannelsTable = ({ onReady }) => {
     showDisabled,
     selectedProfileId,
     showOnlyStreamlessChannels,
+    showOnlyStaleChannels,
   ]);
 
   const stopPropagation = useCallback((e) => {
@@ -1128,11 +1133,18 @@ const ChannelsTable = ({ onReady }) => {
     getRowStyles: (row) => {
       const hasStreams =
         row.original.streams && row.original.streams.length > 0;
-      return hasStreams
-        ? {} // Default style for channels with streams
-        : {
-            className: 'no-streams-row', // Add a class instead of background color
-          };
+      const hasStaleStreams =
+        row.original.streams && row.original.streams.some((s) => s.is_stale);
+
+      if (!hasStreams) {
+        return { className: 'no-streams-row' };
+      }
+
+      if (hasStaleStreams) {
+        return { className: 'stale-streams-row' };
+      }
+
+      return {};
     },
   });
 
@@ -1435,6 +1447,8 @@ const ChannelsTable = ({ onReady }) => {
             setShowDisabled={setShowDisabled}
             showOnlyStreamlessChannels={showOnlyStreamlessChannels}
             setShowOnlyStreamlessChannels={setShowOnlyStreamlessChannels}
+            showOnlyStaleChannels={showOnlyStaleChannels}
+            setShowOnlyStaleChannels={setShowOnlyStaleChannels}
           />
 
           {/* Table or ghost empty state inside Paper */}
diff --git a/frontend/src/components/tables/ChannelsTable/ChannelTableHeader.jsx b/frontend/src/components/tables/ChannelsTable/ChannelTableHeader.jsx
index ac2f4d8d..c8023283 100644
--- a/frontend/src/components/tables/ChannelsTable/ChannelTableHeader.jsx
+++ b/frontend/src/components/tables/ChannelsTable/ChannelTableHeader.jsx
@@ -116,6 +116,8 @@ const ChannelTableHeader = ({
   setShowDisabled,
   showOnlyStreamlessChannels,
   setShowOnlyStreamlessChannels,
+  showOnlyStaleChannels,
+  setShowOnlyStaleChannels,
 }) => {
   const theme = useMantineTheme();
 
@@ -218,7 +220,21 @@ const ChannelTableHeader = ({
   };
 
   const toggleShowOnlyStreamlessChannels = () => {
-    setShowOnlyStreamlessChannels(!showOnlyStreamlessChannels);
+    const newVal = !showOnlyStreamlessChannels;
+    setShowOnlyStreamlessChannels(newVal);
+    if (newVal) {
+      // Ensure stale toggle is cleared when enabling streamless-only
+      setShowOnlyStaleChannels(false);
+    }
+  };
+
+  const toggleShowOnlyStaleChannels = () => {
+    const newVal = !showOnlyStaleChannels;
+    setShowOnlyStaleChannels(newVal);
+    if (newVal) {
+      // Ensure streamless toggle is cleared when enabling stale-only
+      setShowOnlyStreamlessChannels(false);
+    }
   };
 
   const toggleHeaderPinned = () => {
@@ -307,6 +323,15 @@ const ChannelTableHeader = ({
               >
                 <Text size="xs">Only Empty Channels</Text>
               </Menu.Item>
+
+              <Menu.Item
+                onClick={toggleShowOnlyStaleChannels}
+                leftSection={
+                  showOnlyStaleChannels ? <SquareCheck size={18} /> : <Square size={18} />
+                }
+              >
+                <Text size="xs">Only Containing Stale Streams</Text>
+              </Menu.Item>
             </Menu.Dropdown>
           </Menu>
 
diff --git a/frontend/src/components/tables/table.css b/frontend/src/components/tables/table.css
index b957d741..688cf486 100644
--- a/frontend/src/components/tables/table.css
+++ b/frontend/src/components/tables/table.css
@@ -92,7 +92,7 @@ html {
     opacity: 0;
   }
 
-  *:hover > .resizer {
+  *:hover>.resizer {
     opacity: 1;
   }
 }

From 31fb27dffc279b2cc34875c16b313b8a4c0424b5 Mon Sep 17 00:00:00 2001
From: Jonathan Caicedo <jonathan@jcaicedo.com>
Date: Thu, 26 Feb 2026 11:26:39 -0500
Subject: [PATCH 20/58] perf: avoid double-calling row.original.streams - reuse
 hasStreams

This is a micro-optimization, but prevents checking row.original.streams twice and also guards checking for stale streams unless we know we have streams. The performance hit before would have been tiny, but every bit counts.
---
 frontend/src/components/tables/ChannelsTable.jsx | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/frontend/src/components/tables/ChannelsTable.jsx b/frontend/src/components/tables/ChannelsTable.jsx
index 75bf0e33..25bea550 100644
--- a/frontend/src/components/tables/ChannelsTable.jsx
+++ b/frontend/src/components/tables/ChannelsTable.jsx
@@ -1134,7 +1134,7 @@ const ChannelsTable = ({ onReady }) => {
       const hasStreams =
         row.original.streams && row.original.streams.length > 0;
       const hasStaleStreams =
-        row.original.streams && row.original.streams.some((s) => s.is_stale);
+        hasStreams && row.original.streams.some((s) => s.is_stale);
 
       if (!hasStreams) {
         return { className: 'no-streams-row' };

From 4dfe172fc4dfdea911ea2abee2ba0897963b3fd8 Mon Sep 17 00:00:00 2001
From: Jonathan Caicedo <jonathan@jcaicedo.com>
Date: Sat, 28 Feb 2026 12:20:48 -0500
Subject: [PATCH 21/58] tweak: label name + CSS class name

---
 frontend/src/components/tables/ChannelsTable.jsx      |  2 +-
 .../tables/ChannelsTable/ChannelTableHeader.jsx       |  2 +-
 frontend/src/components/tables/table.css              | 11 +++++++++++
 3 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/frontend/src/components/tables/ChannelsTable.jsx b/frontend/src/components/tables/ChannelsTable.jsx
index 25bea550..833bdb4f 100644
--- a/frontend/src/components/tables/ChannelsTable.jsx
+++ b/frontend/src/components/tables/ChannelsTable.jsx
@@ -1141,7 +1141,7 @@ const ChannelsTable = ({ onReady }) => {
       }
 
       if (hasStaleStreams) {
-        return { className: 'stale-streams-row' };
+        return { className: 'partially-stale-streams-row' };
       }
 
       return {};
diff --git a/frontend/src/components/tables/ChannelsTable/ChannelTableHeader.jsx b/frontend/src/components/tables/ChannelsTable/ChannelTableHeader.jsx
index c8023283..cda9b496 100644
--- a/frontend/src/components/tables/ChannelsTable/ChannelTableHeader.jsx
+++ b/frontend/src/components/tables/ChannelsTable/ChannelTableHeader.jsx
@@ -330,7 +330,7 @@ const ChannelTableHeader = ({
                   showOnlyStaleChannels ? <SquareCheck size={18} /> : <Square size={18} />
                 }
               >
-                <Text size="xs">Only Containing Stale Streams</Text>
+                <Text size="xs">Has Stale Streams</Text>
               </Menu.Item>
             </Menu.Dropdown>
           </Menu>
diff --git a/frontend/src/components/tables/table.css b/frontend/src/components/tables/table.css
index 688cf486..0e02dc1e 100644
--- a/frontend/src/components/tables/table.css
+++ b/frontend/src/components/tables/table.css
@@ -140,6 +140,17 @@ html {
   background-color: color-mix(in srgb, rgb(239, 68, 68) 30%, #27272a) !important;
 }
 
+/* Rows that contain some stale streams (may be mixed with active) */
+.partially-stale-streams-row {
+  background-color: rgba(239, 154, 68, 0.15) !important;
+  /* orange-500 @ 12% */
+  box-shadow: inset 0 0 0 1px rgba(239, 154, 68, 0.3);
+}
+
+.table-striped .tbody .tr.partially-stale-streams-row:hover {
+  background-color: rgba(239, 154, 68, 0.3) !important;
+}
+
 /* Prevent text selection when shift key is pressed */
 .shift-key-active {
   cursor: pointer !important;

From 7fed49a3340e70e739ee7d72954be1ea6629d7cb Mon Sep 17 00:00:00 2001
From: Claude Code <claude@ha-docker>
Date: Mon, 2 Mar 2026 13:23:45 +0000
Subject: [PATCH 22/58] Address review feedback from CodeBormen

- Use atomic SET NX EX instead of separate SETNX + EXPIRE to prevent
  zombie locks if the process crashes between the two calls
- Replace time.sleep() with gevent.sleep() in get_instance() spin-wait
  to avoid blocking the greenlet hub
- Defer cleanup when ownership is lost but clients are still connected,
  giving the new owner time to establish its stream before we tear down

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 apps/proxy/ts_proxy/server.py | 19 +++++++++++++++----
 1 file changed, 15 insertions(+), 4 deletions(-)

diff --git a/apps/proxy/ts_proxy/server.py b/apps/proxy/ts_proxy/server.py
index b28154b3..b0b1a08e 100644
--- a/apps/proxy/ts_proxy/server.py
+++ b/apps/proxy/ts_proxy/server.py
@@ -51,12 +51,11 @@ class ProxyServer:
             cls._instance = real_instance
             return real_instance
         # Another greenlet is initializing — wait for completion
-        import time
         while True:
             inst = cls._instance
             if inst is not None and inst is not cls._INITIALIZING:
                 return inst
-            time.sleep(0.05)
+            gevent.sleep(0.05)
 
     def __init__(self):
         """Initialize proxy server with worker identification"""
@@ -462,9 +461,8 @@ class ProxyServer:
             if current is None:
                 # Key expired — re-acquire if we have the stream_manager
                 if channel_id in self.stream_managers:
-                    acquired = self.redis_client.setnx(lock_key, self.worker_id)
+                    acquired = self.redis_client.set(lock_key, self.worker_id, nx=True, ex=ttl)
                     if acquired:
-                        self.redis_client.expire(lock_key, ttl)
                         logger.warning(f"Re-acquired expired ownership for channel {channel_id}")
                         return True
                     else:
@@ -1217,6 +1215,19 @@ class ProxyServer:
                                     logger.info(f"Successfully re-acquired ownership for {channel_id}")
                                     continue
                                 else:
+                                    # Defer cleanup if we still have active clients — give the
+                                    # new owner time to spin up its own stream before we tear
+                                    # ours down, so viewers don't get disconnected.
+                                    has_clients = (
+                                        channel_id in self.client_managers
+                                        and self.client_managers[channel_id].get_client_count() > 0
+                                    )
+                                    if has_clients:
+                                        logger.warning(
+                                            f"Ownership lost for {channel_id} but {self.client_managers[channel_id].get_client_count()} "
+                                            f"client(s) still connected — deferring cleanup to next cycle"
+                                        )
+                                        continue
                                     logger.error(f"Failed to re-acquire ownership for {channel_id}, will clean up")
 
                             # For channels we don't own, check if they've been stopped/cleaned up in Redis

From 1e629d57ecc8c9d2cadd4e0ccf163b28ab9de0ef Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Tue, 3 Mar 2026 10:11:45 -0600
Subject: [PATCH 23/58] Refactor ChannelsTable row styling logic to use
 memoized class mapping for improved performance and readability.

---
 .../src/components/tables/ChannelsTable.jsx   | 29 ++++++++++---------
 1 file changed, 15 insertions(+), 14 deletions(-)

diff --git a/frontend/src/components/tables/ChannelsTable.jsx b/frontend/src/components/tables/ChannelsTable.jsx
index 9e7782be..c627495a 100644
--- a/frontend/src/components/tables/ChannelsTable.jsx
+++ b/frontend/src/components/tables/ChannelsTable.jsx
@@ -269,6 +269,19 @@ const ChannelsTable = ({ onReady }) => {
   // store/channelsTable
   const data = useChannelsTableStore((s) => s.channels);
   const pageCount = useChannelsTableStore((s) => s.pageCount);
+
+  const rowClassMap = useMemo(() => {
+    const map = {};
+    for (const channel of data) {
+      const hasStreams = channel.streams?.length > 0;
+      if (!hasStreams) {
+        map[channel.id] = 'no-streams-row';
+      } else if (channel.streams.some((s) => s.is_stale)) {
+        map[channel.id] = 'partially-stale-streams-row';
+      }
+    }
+    return map;
+  }, [data]);
   const setSelectedChannelIds = useChannelsTableStore(
     (s) => s.setSelectedChannelIds
   );
@@ -1130,20 +1143,8 @@ const ChannelsTable = ({ onReady }) => {
       epg: renderHeaderCell,
     },
     getRowStyles: (row) => {
-      const hasStreams =
-        row.original.streams && row.original.streams.length > 0;
-      const hasStaleStreams =
-        hasStreams && row.original.streams.some((s) => s.is_stale);
-
-      if (!hasStreams) {
-        return { className: 'no-streams-row' };
-      }
-
-      if (hasStaleStreams) {
-        return { className: 'partially-stale-streams-row' };
-      }
-
-      return {};
+      const cls = rowClassMap[row.original.id];
+      return cls ? { className: cls } : {};
     },
   });
 

From eb020724311797c4d4f821ea5c759d921424415a Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Tue, 3 Mar 2026 10:20:37 -0600
Subject: [PATCH 24/58] Renamed "partially-stale-streams-row" to
 "has-stale-streams-row" for accuracy. Also updated the changelog.

---
 CHANGELOG.md                                     | 4 ++++
 frontend/src/components/tables/ChannelsTable.jsx | 2 +-
 frontend/src/components/tables/table.css         | 6 +++---
 3 files changed, 8 insertions(+), 4 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index af516928..c6606c9b 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -7,6 +7,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Added
+
+- Channel table filter for channels that have stale streams: A new "Has Stale Streams" filter option in the channel table header menu highlights and filters channels containing at least one stale stream. Channels with stale streams are visually distinguished with an orange tint. The filter is mutually exclusive with "Only Empty Channels". - Thanks [@JCBird1012](https://github.com/JCBird1012)
+
 ## [0.20.2] - 2026-03-03
 
 ### Security
diff --git a/frontend/src/components/tables/ChannelsTable.jsx b/frontend/src/components/tables/ChannelsTable.jsx
index c627495a..b6b93f43 100644
--- a/frontend/src/components/tables/ChannelsTable.jsx
+++ b/frontend/src/components/tables/ChannelsTable.jsx
@@ -277,7 +277,7 @@ const ChannelsTable = ({ onReady }) => {
       if (!hasStreams) {
         map[channel.id] = 'no-streams-row';
       } else if (channel.streams.some((s) => s.is_stale)) {
-        map[channel.id] = 'partially-stale-streams-row';
+        map[channel.id] = 'has-stale-streams-row';
       }
     }
     return map;
diff --git a/frontend/src/components/tables/table.css b/frontend/src/components/tables/table.css
index 0e02dc1e..1e21540a 100644
--- a/frontend/src/components/tables/table.css
+++ b/frontend/src/components/tables/table.css
@@ -140,14 +140,14 @@ html {
   background-color: color-mix(in srgb, rgb(239, 68, 68) 30%, #27272a) !important;
 }
 
-/* Rows that contain some stale streams (may be mixed with active) */
-.partially-stale-streams-row {
+/* Rows that contain at least 1 stale stream (may be mixed with active) */
+.has-stale-streams-row {
   background-color: rgba(239, 154, 68, 0.15) !important;
   /* orange-500 @ 12% */
   box-shadow: inset 0 0 0 1px rgba(239, 154, 68, 0.3);
 }
 
-.table-striped .tbody .tr.partially-stale-streams-row:hover {
+.table-striped .tbody .tr.has-stale-streams-row:hover {
   background-color: rgba(239, 154, 68, 0.3) !important;
 }
 

From 6d7130d71e4d2f70df8999dbb29ff6cbc4c607a0 Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Tue, 3 Mar 2026 11:06:52 -0600
Subject: [PATCH 25/58] Bug Fix: fix get_instance deadlock and non-atomic
 ownership acquisition

If ProxyServer() raises during singleton construction, _instance was left
as _INITIALIZING permanently, causing all subsequent get_instance() callers
to spin in an infinite gevent.sleep() loop. Wrap construction in try/except
and reset _instance to None on failure so callers can retry.

Also replace the non-atomic setnx() + expire() pair in try_acquire_ownership()
with a single atomic SET NX EX call, consistent with the approach already used
in extend_ownership() and eliminating the race window where a crash between the
two calls could leave a key with no TTL (permanent ownership lock).
---
 apps/proxy/ts_proxy/server.py | 26 +++++++++++++-------------
 1 file changed, 13 insertions(+), 13 deletions(-)

diff --git a/apps/proxy/ts_proxy/server.py b/apps/proxy/ts_proxy/server.py
index b0b1a08e..609a3000 100644
--- a/apps/proxy/ts_proxy/server.py
+++ b/apps/proxy/ts_proxy/server.py
@@ -43,13 +43,17 @@ class ProxyServer:
             return inst
         if inst is None:
             cls._instance = cls._INITIALIZING
-            from .server import ProxyServer
-            from .stream_manager import StreamManager
-            from .stream_buffer import StreamBuffer
-            from .client_manager import ClientManager
-            real_instance = ProxyServer()
-            cls._instance = real_instance
-            return real_instance
+            try:
+                from .server import ProxyServer
+                from .stream_manager import StreamManager
+                from .stream_buffer import StreamBuffer
+                from .client_manager import ClientManager
+                real_instance = ProxyServer()
+                cls._instance = real_instance
+                return real_instance
+            except Exception:
+                cls._instance = None  # Reset so next call can retry
+                raise
         # Another greenlet is initializing — wait for completion
         while True:
             inst = cls._instance
@@ -391,20 +395,16 @@ class ProxyServer:
             # Create a lock key with proper namespace
             lock_key = RedisKeys.channel_owner(channel_id)
 
-            # Use Redis SETNX for atomic locking with error handling
+            # Use atomic SET NX EX for locking with error handling
             acquired = self._execute_redis_command(
-                lambda: self.redis_client.setnx(lock_key, self.worker_id)
+                lambda: self.redis_client.set(lock_key, self.worker_id, nx=True, ex=ttl)
             )
 
             if acquired is None:  # Redis command failed
                 logger.warning(f"Redis command failed during ownership acquisition - assuming ownership")
                 return True
 
-            # If acquired, set expiry to prevent orphaned locks
             if acquired:
-                self._execute_redis_command(
-                    lambda: self.redis_client.expire(lock_key, ttl)
-                )
                 logger.info(f"Worker {self.worker_id} acquired ownership of channel {channel_id}")
                 return True
 

From d6e4a34f3b78f14a480db00203576b3e826931fb Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Tue, 3 Mar 2026 11:11:50 -0600
Subject: [PATCH 26/58] changelog: Update changelog for proxy server pr.

---
 CHANGELOG.md | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index c6606c9b..fadc44e8 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -11,6 +11,15 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 - Channel table filter for channels that have stale streams: A new "Has Stale Streams" filter option in the channel table header menu highlights and filters channels containing at least one stale stream. Channels with stale streams are visually distinguished with an orange tint. The filter is mutually exclusive with "Only Empty Channels". - Thanks [@JCBird1012](https://github.com/JCBird1012)
 
+### Fixed
+
+- TS proxy streams dying after 30–200 seconds in multi-worker uWSGI/Celery deployments, caused by three interrelated bugs. (Fixes #992, #980) - Thanks [@PFalko](https://github.com/PFalko)
+  - **Double ProxyServer instantiation**: `ProxyConfig.ready()` called `TSProxyServer()` directly while `TSProxyConfig.ready()` also called `TSProxyServer.get_instance()`, creating two instances per worker — each with its own cleanup thread. The orphaned thread could not extend ownership because it had no entries in `stream_managers`. Fixed by using `TSProxyServer.get_instance()` in `ProxyConfig.ready()`.
+  - **`flushdb()` on every Redis client init**: `RedisClient.get_client()` called `client.flushdb()` whenever `_client` was `None`. Celery autoscale (`--autoscale=6,1`) spawning new workers mid-stream triggered this path, nuking all Redis keys including active ownership keys, client records, and channel metadata. Removed the `flushdb()` call entirely.
+  - **No recovery from expired ownership**: `get_channel_owner()` called `redis.get()` twice inside a lambda (TOCTOU race — key could expire between calls); `extend_ownership()` silently returned `False` on expiry with no re-acquisition; and the non-owner cleanup path unconditionally killed streams even when the worker held the `stream_manager`. Fixed with a single `GET` in `get_channel_owner()`, re-acquisition via atomic `SET NX EX` in `extend_ownership()`, and a re-acquisition attempt with client-aware cleanup deferral in the cleanup thread.
+- `get_instance()` deadlock: if `ProxyServer()` raised an exception during singleton construction, `_instance` was left permanently as the `_INITIALIZING` sentinel, causing all subsequent `get_instance()` callers to spin in an infinite `gevent.sleep()` loop. Construction is now wrapped in `try/except`; on failure `_instance` resets to `None` so the next call can retry.
+- Non-atomic ownership acquisition in `try_acquire_ownership()`: replaced the separate `setnx()` + `expire()` calls with a single atomic `SET NX EX`, eliminating the race window where a process crash between the two calls could leave an ownership key with no TTL (permanent ownership lock).
+
 ## [0.20.2] - 2026-03-03
 
 ### Security

From dcb3cf8ea866a9e4928e476bbb5398b2e979bb6f Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Tue, 3 Mar 2026 11:56:06 -0600
Subject: [PATCH 27/58] changelog: Update changelog for monkey patch PR.

---
 CHANGELOG.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index fadc44e8..55fac12c 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -13,6 +13,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- Celery Redis connections using unpatched sockets under gevent: `dispatcharr/__init__.py` imports `celery.py` as part of `django.setup()`, initialising the Redis connection pool before uWSGI's gevent plugin patches stdlib sockets. Added `gevent.monkey.patch_all()` at the top of `wsgi.py` to guarantee all broker connections are gevent-aware. - Thanks [@endoze](https://github.com/endoze)
 - TS proxy streams dying after 30–200 seconds in multi-worker uWSGI/Celery deployments, caused by three interrelated bugs. (Fixes #992, #980) - Thanks [@PFalko](https://github.com/PFalko)
   - **Double ProxyServer instantiation**: `ProxyConfig.ready()` called `TSProxyServer()` directly while `TSProxyConfig.ready()` also called `TSProxyServer.get_instance()`, creating two instances per worker — each with its own cleanup thread. The orphaned thread could not extend ownership because it had no entries in `stream_managers`. Fixed by using `TSProxyServer.get_instance()` in `ProxyConfig.ready()`.
   - **`flushdb()` on every Redis client init**: `RedisClient.get_client()` called `client.flushdb()` whenever `_client` was `None`. Celery autoscale (`--autoscale=6,1`) spawning new workers mid-stream triggered this path, nuking all Redis keys including active ownership keys, client records, and channel metadata. Removed the `flushdb()` call entirely.

From 7653859a9a268ba0d9d5aac56152d79800516f16 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Tue, 3 Mar 2026 14:22:45 -0600
Subject: [PATCH 28/58] Reset connection_allocated flag after redirect stream
 release

The redirect path in stream_ts() released the connection slot via release_stream() but did not reset the connection_allocated flag. Could have could caused a redundant no-op call in the exception handler.
---
 apps/proxy/ts_proxy/views.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/apps/proxy/ts_proxy/views.py b/apps/proxy/ts_proxy/views.py
index 90ac2673..ccbb947d 100644
--- a/apps/proxy/ts_proxy/views.py
+++ b/apps/proxy/ts_proxy/views.py
@@ -327,6 +327,7 @@ def stream_ts(request, channel_id):
                 # Release stream lock before redirecting
                 if not channel.release_stream():
                     logger.warning(f"[{client_id}] Failed to release stream before redirect")
+                connection_allocated = False
                 # Final decision based on validation results
                 if is_valid:
                     logger.info(

From e78c24aad735952eb6f8939dc2935aa4fe20546d Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Tue, 3 Mar 2026 15:13:46 -0600
Subject: [PATCH 29/58] Enhance Redis chunk management in StreamBuffer and
 StreamGenerator

- Register Lua scripts for optimized chunk retrieval in StreamBuffer.
- Implement atomic Lua binary search to find the oldest available chunk in Redis.
- Update StreamGenerator to handle expired chunks more efficiently, reducing TOCTOU race conditions.
---
 apps/proxy/ts_proxy/stream_buffer.py    | 92 +++++++++++++++++++++++++
 apps/proxy/ts_proxy/stream_generator.py | 36 +++++++---
 2 files changed, 119 insertions(+), 9 deletions(-)

diff --git a/apps/proxy/ts_proxy/stream_buffer.py b/apps/proxy/ts_proxy/stream_buffer.py
index 85feb5dd..2bf99527 100644
--- a/apps/proxy/ts_proxy/stream_buffer.py
+++ b/apps/proxy/ts_proxy/stream_buffer.py
@@ -45,6 +45,15 @@ class StreamBuffer:
         self._write_buffer = bytearray()
         self.target_chunk_size = ConfigHelper.get('BUFFER_CHUNK_SIZE', TS_PACKET_SIZE * 5644)  # ~1MB default
 
+        # Register Lua scripts once — subsequent calls use EVALSHA (just the
+        # SHA hash) instead of sending the full script text on every invocation.
+        if self.redis_client:
+            self._find_oldest_chunk_sha = self.redis_client.register_script(
+                self._FIND_OLDEST_CHUNK_LUA
+            )
+        else:
+            self._find_oldest_chunk_sha = None
+
         # Track timers for proper cleanup
         self.stopping = False
         self.fill_timers = []
@@ -328,6 +337,89 @@ class StreamBuffer:
 
         return chunks, client_index + chunk_count
 
+    # Lua script that runs an atomic binary search on the Redis server.
+    # Chunks expire in FIFO order (same TTL, sequential writes), so the
+    # alive range is contiguous: [oldest_surviving .. buffer_head].
+    # Binary search finds the boundary in O(log N) EXISTS calls with zero
+    # round-trips between steps and no TOCTOU races (Lua scripts are atomic).
+    #
+    # ARGV[1] = key prefix  (e.g. "ts_proxy:channel:<id>:buffer:chunk:")
+    # ARGV[2] = low index   (client_index + 1, first chunk the client needs)
+    # ARGV[3] = high index  (buffer head, most recent chunk)
+    #
+    # Returns: the index of the oldest existing chunk, or -1 if none exist.
+    _FIND_OLDEST_CHUNK_LUA = """
+    local prefix = ARGV[1]
+    local low    = tonumber(ARGV[2])
+    local high   = tonumber(ARGV[3])
+
+    if redis.call('EXISTS', prefix .. high) == 0 then
+        return -1
+    end
+
+    local result = high
+    while low <= high do
+        local mid = math.floor((low + high) / 2)
+        if redis.call('EXISTS', prefix .. mid) == 1 then
+            result = mid
+            high = mid - 1
+        else
+            low = mid + 1
+        end
+    end
+    return result
+    """
+
+    def find_oldest_available_chunk(self, client_index):
+        """Find the oldest (lowest-index) chunk that still exists in Redis.
+
+        Executes an atomic Lua binary search on the Redis server — one
+        round-trip, ~log2(N) EXISTS calls, no TOCTOU between steps.
+
+        The actual read attempt (get_optimized_client_data) is what
+        authoritatively detects expiration; this method is best-effort
+        positioning that self-corrects on the next iteration if the found
+        chunk also expires before the client can read it.
+
+        Args:
+            client_index: The client's current local_index (last consumed chunk).
+
+        Returns:
+            int or None: The local_index value the client should jump to
+                         (one before the first available chunk), or None if no
+                         chunks are available at all.
+        """
+        if not self.redis_client:
+            return None
+
+        low = client_index + 1   # First chunk the client needs
+        high = self.index        # Latest chunk written
+
+        if low > high:
+            return None
+
+        try:
+            # Uses EVALSHA under the hood — sends only the SHA hash,
+            # not the full script text, on every call after the first.
+            result = self._find_oldest_chunk_sha(
+                args=[
+                    RedisKeys.buffer_chunk_prefix(self.channel_id),
+                    low,
+                    high,
+                ],
+            )
+
+            if result == -1:
+                return None
+
+            # Return result - 1 so local_index points to one before the
+            # first available chunk (matching the "last consumed" convention).
+            return int(result) - 1
+
+        except Exception as e:
+            logger.error(f"Error running find_oldest_chunk Lua script for channel {self.channel_id}: {e}")
+            return None
+
     # Add a new method to safely create timers
     def schedule_timer(self, delay, callback, *args, **kwargs):
         """Schedule a timer and track it for proper cleanup"""
diff --git a/apps/proxy/ts_proxy/stream_generator.py b/apps/proxy/ts_proxy/stream_generator.py
index 50404f1d..8f631d53 100644
--- a/apps/proxy/ts_proxy/stream_generator.py
+++ b/apps/proxy/ts_proxy/stream_generator.py
@@ -224,17 +224,35 @@ class StreamGenerator:
                 self.empty_reads += 1
                 self.consecutive_empty += 1
 
-                # Check if we're too far behind (chunks expired from Redis)
+                # We got no data despite being behind the buffer head.
+                # The read itself is the authoritative signal — no separate
+                # existence check needed, avoiding TOCTOU races with Redis TTL.
                 chunks_behind = self.buffer.index - self.local_index
-                if chunks_behind > 50:  # If more than 50 chunks behind, jump forward
-                    # Calculate new position: stay a few chunks behind current buffer
-                    initial_behind = ConfigHelper.initial_behind_chunks()
-                    new_index = max(self.local_index, self.buffer.index - initial_behind)
+                if chunks_behind > 0:
+                    # Next chunk has expired — find the oldest chunk still in Redis
+                    new_index = self.buffer.find_oldest_available_chunk(self.local_index)
 
-                    logger.warning(f"[{self.client_id}] Client too far behind ({chunks_behind} chunks), jumping from {self.local_index} to {new_index}")
-                    self.local_index = new_index
-                    self.consecutive_empty = 0  # Reset since we're repositioning
-                    continue  # Try again immediately with new position
+                    if new_index is not None:
+                        skipped = new_index - self.local_index
+                        logger.warning(
+                            f"[{self.client_id}] Next chunk expired (index {self.local_index + 1}), "
+                            f"jumping to oldest available: {new_index + 1} "
+                            f"(skipped {skipped} chunks, buffer head at {self.buffer.index})"
+                        )
+                        self.local_index = new_index
+                    else:
+                        # No chunks available at all — jump to near the buffer head
+                        initial_behind = ConfigHelper.initial_behind_chunks()
+                        new_index = max(self.local_index, self.buffer.index - initial_behind)
+                        logger.warning(
+                            f"[{self.client_id}] No chunks available in buffer, "
+                            f"jumping to near buffer head: {new_index} "
+                            f"(buffer head at {self.buffer.index})"
+                        )
+                        self.local_index = new_index
+
+                    self.consecutive_empty = 0
+                    continue  # Retry immediately with the new position
 
                 if self._should_send_keepalive(self.local_index):
                     keepalive_packet = create_ts_packet('keepalive')

From d487bb5671236877c28dfe1e148e7072c9a15fe2 Mon Sep 17 00:00:00 2001
From: SergeantPanda <61642231+SergeantPanda@users.noreply.github.com>
Date: Tue, 3 Mar 2026 15:33:55 -0600
Subject: [PATCH 30/58] Revert "fix(wsgi): ensure gevent monkey-patching before
 Django import chain"

---
 CHANGELOG.md                   |  1 -
 dispatcharr/tests/__init__.py  |  0
 dispatcharr/tests/test_wsgi.py | 39 ----------------------------------
 dispatcharr/wsgi.py            | 15 -------------
 4 files changed, 55 deletions(-)
 delete mode 100644 dispatcharr/tests/__init__.py
 delete mode 100644 dispatcharr/tests/test_wsgi.py

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 55fac12c..fadc44e8 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -13,7 +13,6 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
-- Celery Redis connections using unpatched sockets under gevent: `dispatcharr/__init__.py` imports `celery.py` as part of `django.setup()`, initialising the Redis connection pool before uWSGI's gevent plugin patches stdlib sockets. Added `gevent.monkey.patch_all()` at the top of `wsgi.py` to guarantee all broker connections are gevent-aware. - Thanks [@endoze](https://github.com/endoze)
 - TS proxy streams dying after 30–200 seconds in multi-worker uWSGI/Celery deployments, caused by three interrelated bugs. (Fixes #992, #980) - Thanks [@PFalko](https://github.com/PFalko)
   - **Double ProxyServer instantiation**: `ProxyConfig.ready()` called `TSProxyServer()` directly while `TSProxyConfig.ready()` also called `TSProxyServer.get_instance()`, creating two instances per worker — each with its own cleanup thread. The orphaned thread could not extend ownership because it had no entries in `stream_managers`. Fixed by using `TSProxyServer.get_instance()` in `ProxyConfig.ready()`.
   - **`flushdb()` on every Redis client init**: `RedisClient.get_client()` called `client.flushdb()` whenever `_client` was `None`. Celery autoscale (`--autoscale=6,1`) spawning new workers mid-stream triggered this path, nuking all Redis keys including active ownership keys, client records, and channel metadata. Removed the `flushdb()` call entirely.
diff --git a/dispatcharr/tests/__init__.py b/dispatcharr/tests/__init__.py
deleted file mode 100644
index e69de29b..00000000
diff --git a/dispatcharr/tests/test_wsgi.py b/dispatcharr/tests/test_wsgi.py
deleted file mode 100644
index cdf7a2e9..00000000
--- a/dispatcharr/tests/test_wsgi.py
+++ /dev/null
@@ -1,39 +0,0 @@
-"""Tests for dispatcharr/wsgi.py gevent monkey-patching."""
-
-import os
-import subprocess
-import sys
-import unittest
-
-
-class TestWSGIGeventPatching(unittest.TestCase):
-    """Verify that wsgi.py applies gevent monkey-patching before Django imports.
-
-    These tests run in subprocesses because monkey.patch_all() must execute
-    before any other imports.  By the time Django's test runner loads this
-    file, ssl/socket are already imported, so calling patch_all() in-process
-    would fail.  A subprocess mirrors how uWSGI actually loads wsgi.py.
-    """
-
-    def test_socket_is_patched_after_wsgi_import(self):
-        """Loading wsgi.py first (as uWSGI does) should patch sockets."""
-        result = subprocess.run(
-            [
-                sys.executable, "-c",
-                "import dispatcharr.wsgi; "
-                "from gevent import monkey; "
-                "assert monkey.is_module_patched('socket'), "
-                "'socket module was not patched by wsgi.py'; "
-                "print('PASS')",
-            ],
-            capture_output=True,
-            text=True,
-            env={**os.environ, "DJANGO_SETTINGS_MODULE": "dispatcharr.settings"},
-            timeout=60,
-            cwd="/app",
-        )
-        self.assertEqual(
-            result.returncode,
-            0,
-            f"monkey-patching check failed:\n{result.stderr}",
-        )
diff --git a/dispatcharr/wsgi.py b/dispatcharr/wsgi.py
index 6e8860af..29c2ba01 100644
--- a/dispatcharr/wsgi.py
+++ b/dispatcharr/wsgi.py
@@ -1,21 +1,6 @@
 """
 WSGI config for dispatcharr project.
 """
-# When running under uWSGI with gevent, ensure monkey-patching is fully
-# applied before any other imports.  Django's setup triggers the import of
-# dispatcharr/__init__.py → celery.py, which initialises the Celery broker
-# transport.  If those imports happen before sockets are patched, Celery's
-# Redis connection pool holds unpatched sockets that silently fail under
-# gevent's cooperative scheduling.  Calling patch_all() here—before the
-# Django import chain—guarantees every socket (including pooled broker
-# connections) is gevent-aware.  The call is idempotent, so it's harmless
-# if uWSGI's gevent plugin has already patched.
-try:
-    from gevent import monkey
-    monkey.patch_all()
-except ImportError:
-    pass
-
 import os
 from django.core.wsgi import get_wsgi_application
 

From 24c20a2f4accf486b89c7e664d92571d2cdc845b Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Tue, 3 Mar 2026 15:59:49 -0600
Subject: [PATCH 31/58] changelog: Update changelog for proxy changes.

---
 CHANGELOG.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index fadc44e8..49e7be68 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -13,6 +13,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- TS proxy client stream lag recovery now only bumps clients forward when their next required chunk has genuinely expired from Redis (TTL), rather than unconditionally jumping if they fell more than 50 chunks behind. Clients are repositioned to the oldest available chunk (minimum data loss) using an atomic server-side Lua binary search, falling back to near the buffer head if nothing is available.
 - TS proxy streams dying after 30–200 seconds in multi-worker uWSGI/Celery deployments, caused by three interrelated bugs. (Fixes #992, #980) - Thanks [@PFalko](https://github.com/PFalko)
   - **Double ProxyServer instantiation**: `ProxyConfig.ready()` called `TSProxyServer()` directly while `TSProxyConfig.ready()` also called `TSProxyServer.get_instance()`, creating two instances per worker — each with its own cleanup thread. The orphaned thread could not extend ownership because it had no entries in `stream_managers`. Fixed by using `TSProxyServer.get_instance()` in `ProxyConfig.ready()`.
   - **`flushdb()` on every Redis client init**: `RedisClient.get_client()` called `client.flushdb()` whenever `_client` was `None`. Celery autoscale (`--autoscale=6,1`) spawning new workers mid-stream triggered this path, nuking all Redis keys including active ownership keys, client records, and channel metadata. Removed the `flushdb()` call entirely.

From 086259ef9dfad69c83efe11379a12536b652d29a Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Tue, 3 Mar 2026 16:28:30 -0600
Subject: [PATCH 32/58] changelog: Update changelog for stream lock PR.

---
 CHANGELOG.md | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 49e7be68..be8b88b4 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -13,6 +13,16 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- TS proxy connection slot leaks and TOCTOU races in stream initialization (Fixes #947) - Thanks [@CodeBormen](https://github.com/CodeBormen)
+  - **TOCTOU race in slot reservation**: `get_stream()` previously used a `GET`→check→`INCR` sequence, allowing concurrent requests to both read the same count below the limit and both reserve a slot, silently exceeding `max_streams`. Replaced with an atomic `INCR`-first pattern: increment unconditionally, check the result, roll back with `DECR` if over capacity.
+  - **Leak on URL generation failure**: `generate_stream_url()` called `get_stream()` (which `INCR`s the counter) but had no cleanup path if subsequent DB lookups or URL construction failed. The post-`get_stream()` block is now wrapped in a `try/except` that calls `release_stream()` on any error.
+  - **Leak on retry-loop timeout**: the retry loop in `stream_ts()` called a bare `get_stream()` on the first failure to classify the error reason. If a slot was available, this `INCR`'d the counter and set Redis keys that were never released when the loop timed out. A `release_stream()` call is now issued before returning 503.
+  - **Leak on `initialize_channel()` failure**: when `initialize_channel()` returned `False`, the connection slot allocated by the preceding `get_stream()` was never released. A `connection_allocated` flag now tracks whether this request performed the `INCR` (fresh initialization vs. joining an existing channel), and `release_stream()` is called guarded by that flag to prevent incorrect decrements when attaching to an already-running channel.
+  - **Safety net for unexpected exceptions**: the outer `except` in `stream_ts()` now checks `connection_allocated` and calls `release_stream()` as a last-resort cleanup for any unhandled exception that escapes before the channel is handed off to the stream lifecycle.
+  - **`release_stream()` now returns `bool`** and adds a metadata-hash fallback: if the primary `channel_stream` / `stream_profile` Redis keys have already been cleaned up by the proxy, it recovers `stream_id` and `profile_id` from the channel's metadata hash and clears those fields atomically to prevent duplicate `DECR`s on repeated calls.
+  - **`update_stream_profile()` uses a Redis pipeline** for the old-profile DECR + key update + new-profile INCR sequence, preventing counter drift if the process crashes between operations.
+  - **`stream_generator._cleanup()`** now falls back to `Stream.objects.get()` when the channel UUID resolves to a preview flow rather than a normal channel, rather than silently skipping the slot release.
+  - **VOD `cleanup_persistent_connection()`** fallback DECR is now conditional: it only decrements the profile counter when the connection tracking key had already expired by TTL (i.e., `remove_connection()` would have skipped the DECR), preventing double-decrements when the key is still present.
 - TS proxy client stream lag recovery now only bumps clients forward when their next required chunk has genuinely expired from Redis (TTL), rather than unconditionally jumping if they fell more than 50 chunks behind. Clients are repositioned to the oldest available chunk (minimum data loss) using an atomic server-side Lua binary search, falling back to near the buffer head if nothing is available.
 - TS proxy streams dying after 30–200 seconds in multi-worker uWSGI/Celery deployments, caused by three interrelated bugs. (Fixes #992, #980) - Thanks [@PFalko](https://github.com/PFalko)
   - **Double ProxyServer instantiation**: `ProxyConfig.ready()` called `TSProxyServer()` directly while `TSProxyConfig.ready()` also called `TSProxyServer.get_instance()`, creating two instances per worker — each with its own cleanup thread. The orphaned thread could not extend ownership because it had no entries in `stream_managers`. Fixed by using `TSProxyServer.get_instance()` in `ProxyConfig.ready()`.

From fc75a68195b59d1f00d1df2994c8741f443a6d76 Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Tue, 3 Mar 2026 18:12:32 -0600
Subject: [PATCH 33/58] =?UTF-8?q?Enhancement:=20new=20clients=20joining=20?=
 =?UTF-8?q?an=20active=20channel=20are=20now=20positioned=20a=20configurab?=
 =?UTF-8?q?le=20number=20of=20seconds=20behind=20live=20rather=20than=20a?=
 =?UTF-8?q?=20fixed=20chunk=20count.=20The=20start=20position=20is=20deter?=
 =?UTF-8?q?mined=20by=20wall-clock=20chunk=20receive=20time=20(stored=20as?=
 =?UTF-8?q?=20a=20Redis=20sorted=20set=20alongside=20the=20buffer),=20so?=
 =?UTF-8?q?=20the=20buffer=20depth=20is=20consistent=20in=20seconds=20rega?=
 =?UTF-8?q?rdless=20of=20stream=20bitrate.=20Setting=20the=20value=20to=20?=
 =?UTF-8?q?`0`=20starts=20clients=20at=20live=20with=20no=20buffer.=20Defa?=
 =?UTF-8?q?ults=20to=202=20seconds.=20Existing=20chunk-count=20gating=20fo?=
 =?UTF-8?q?r=20the=20first=20client=20connecting=20to=20a=20channel=20is?=
 =?UTF-8?q?=20unchanged.=20The=20setting=20is=20exposed=20in=20Settings=20?=
 =?UTF-8?q?=E2=86=92=20Proxy=20as=20"New=20Client=20Buffer=20(seconds)".?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 CHANGELOG.md                                  |  1 +
 apps/proxy/config.py                          |  2 +
 apps/proxy/ts_proxy/config_helper.py          |  9 +++
 apps/proxy/ts_proxy/redis_keys.py             |  6 ++
 apps/proxy/ts_proxy/stream_buffer.py          | 79 +++++++++++++++++++
 apps/proxy/ts_proxy/stream_generator.py       | 43 +++++++++-
 core/api_views.py                             |  1 +
 core/models.py                                |  1 +
 core/serializers.py                           |  6 ++
 .../forms/settings/ProxySettingsForm.jsx      | 12 ++-
 frontend/src/constants.js                     |  5 ++
 .../forms/settings/ProxySettingsFormUtils.js  |  1 +
 12 files changed, 161 insertions(+), 5 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index be8b88b4..88f6797a 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -9,6 +9,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+- New Client Buffer proxy setting: new clients joining an active channel are now positioned a configurable number of seconds behind live rather than a fixed chunk count. The start position is determined by wall-clock chunk receive time (stored as a Redis sorted set alongside the buffer), so the buffer depth is consistent in seconds regardless of stream bitrate. Setting the value to `0` starts clients at live with no buffer. Defaults to 2 seconds. Existing chunk-count gating for the first client connecting to a channel is unchanged. The setting is exposed in Settings → Proxy as "New Client Buffer (seconds)".
 - Channel table filter for channels that have stale streams: A new "Has Stale Streams" filter option in the channel table header menu highlights and filters channels containing at least one stale stream. Channels with stale streams are visually distinguished with an orange tint. The filter is mutually exclusive with "Only Empty Channels". - Thanks [@JCBird1012](https://github.com/JCBird1012)
 
 ### Fixed
diff --git a/apps/proxy/config.py b/apps/proxy/config.py
index 3b1ce967..dbdd8278 100644
--- a/apps/proxy/config.py
+++ b/apps/proxy/config.py
@@ -43,6 +43,7 @@ class BaseConfig:
                 "redis_chunk_ttl": 60,
                 "channel_shutdown_delay": 0,
                 "channel_init_grace_period": 5,
+                "new_client_behind_seconds": 2,
             }
 
         finally:
@@ -81,6 +82,7 @@ class TSConfig(BaseConfig):
     # Buffer settings
     INITIAL_BEHIND_CHUNKS = 4  # How many chunks behind to start a client (4 chunks = ~1MB)
     CHUNK_BATCH_SIZE = 5       # How many chunks to fetch in one batch
+    NEW_CLIENT_BEHIND_SECONDS = 2  # Start new clients this many seconds behind live (0 = start at live)
     KEEPALIVE_INTERVAL = 0.5   # Seconds between keepalive packets when at buffer head
     # Chunk read timeout
     CHUNK_TIMEOUT = 5        # Seconds to wait for each chunk read
diff --git a/apps/proxy/ts_proxy/config_helper.py b/apps/proxy/ts_proxy/config_helper.py
index d7d33558..f0f7a5b7 100644
--- a/apps/proxy/ts_proxy/config_helper.py
+++ b/apps/proxy/ts_proxy/config_helper.py
@@ -41,6 +41,15 @@ class ConfigHelper:
         """Get number of chunks to start behind"""
         return ConfigHelper.get('INITIAL_BEHIND_CHUNKS', 4)
 
+    @staticmethod
+    def new_client_behind_seconds():
+        """Get number of seconds behind live to start new clients.
+        0 means start at live (buffer head).
+        Loaded from DB proxy_settings so users can change it at runtime."""
+        from apps.proxy.config import TSConfig
+        settings = TSConfig.get_proxy_settings()
+        return settings.get('new_client_behind_seconds', 2)
+
     @staticmethod
     def keepalive_interval():
         """Get keepalive interval in seconds"""
diff --git a/apps/proxy/ts_proxy/redis_keys.py b/apps/proxy/ts_proxy/redis_keys.py
index 22b02648..29846f98 100644
--- a/apps/proxy/ts_proxy/redis_keys.py
+++ b/apps/proxy/ts_proxy/redis_keys.py
@@ -79,6 +79,12 @@ class RedisKeys:
         """Key for worker heartbeat"""
         return f"ts_proxy:worker:{worker_id}:heartbeat"
 
+    @staticmethod
+    def chunk_timestamps(channel_id):
+        """Sorted set mapping chunk receive-timestamps (score) to chunk indices (member).
+        Used for time-based client positioning."""
+        return f"ts_proxy:channel:{channel_id}:buffer:chunk_timestamps"
+
     @staticmethod
     def transcode_active(channel_id):
         """Key indicating active transcode process"""
diff --git a/apps/proxy/ts_proxy/stream_buffer.py b/apps/proxy/ts_proxy/stream_buffer.py
index 2bf99527..111468bf 100644
--- a/apps/proxy/ts_proxy/stream_buffer.py
+++ b/apps/proxy/ts_proxy/stream_buffer.py
@@ -45,14 +45,21 @@ class StreamBuffer:
         self._write_buffer = bytearray()
         self.target_chunk_size = ConfigHelper.get('BUFFER_CHUNK_SIZE', TS_PACKET_SIZE * 5644)  # ~1MB default
 
+        # Sorted-set key for chunk receive-timestamps (time-based positioning)
+        self.chunk_timestamps_key = RedisKeys.chunk_timestamps(channel_id) if channel_id else ""
+
         # Register Lua scripts once — subsequent calls use EVALSHA (just the
         # SHA hash) instead of sending the full script text on every invocation.
         if self.redis_client:
             self._find_oldest_chunk_sha = self.redis_client.register_script(
                 self._FIND_OLDEST_CHUNK_LUA
             )
+            self._find_chunk_by_time_sha = self.redis_client.register_script(
+                self._FIND_CHUNK_BY_TIME_LUA
+            )
         else:
             self._find_oldest_chunk_sha = None
+            self._find_chunk_by_time_sha = None
 
         # Track timers for proper cleanup
         self.stopping = False
@@ -101,6 +108,14 @@ class StreamBuffer:
                         chunk_key = RedisKeys.buffer_chunk(self.channel_id, chunk_index)
                         self.redis_client.setex(chunk_key, self.chunk_ttl, bytes(chunk_data))
 
+                        # Record receive timestamp for time-based client positioning
+                        if self.chunk_timestamps_key:
+                            now = time.time()
+                            self.redis_client.zadd(self.chunk_timestamps_key, {str(chunk_index): now})
+                            # Prune entries whose chunks have expired from Redis
+                            self.redis_client.zremrangebyscore(self.chunk_timestamps_key, '-inf', now - self.chunk_ttl)
+                            self.redis_client.expire(self.chunk_timestamps_key, self.chunk_ttl)
+
                         # Update local tracking
                         self.index = chunk_index
                         writes_done += 1
@@ -284,6 +299,13 @@ class StreamBuffer:
                 if hasattr(self, '_partial_packet'):
                     self._partial_packet = bytearray()
 
+            # Clean up the chunk timestamps sorted set
+            if self.redis_client and self.chunk_timestamps_key:
+                try:
+                    self.redis_client.delete(self.chunk_timestamps_key)
+                except Exception as e:
+                    logger.error(f"Error deleting chunk timestamps key: {e}")
+
         except Exception as e:
             logger.error(f"Error during buffer stop: {e}")
 
@@ -420,6 +442,63 @@ class StreamBuffer:
             logger.error(f"Error running find_oldest_chunk Lua script for channel {self.channel_id}: {e}")
             return None
 
+    # ------------------------------------------------------------------
+    # Lua script: atomic reverse-scan of the chunk_timestamps sorted set.
+    # Finds the chunk whose receive-timestamp is closest to (but <=) a
+    # target wall-clock time.  Returns the chunk index or -1.
+    #
+    # KEYS[1] = chunk_timestamps sorted-set key
+    # ARGV[1] = target timestamp  (time.time() - desired_seconds_behind)
+    # ------------------------------------------------------------------
+    _FIND_CHUNK_BY_TIME_LUA = """
+    local ts_key  = KEYS[1]
+    local target  = tonumber(ARGV[1])
+
+    -- ZREVRANGEBYSCORE returns members with score <= target, highest first.
+    local result = redis.call('ZREVRANGEBYSCORE', ts_key, target, '-inf', 'LIMIT', 0, 1)
+    if #result == 0 then
+        return -1
+    end
+    return tonumber(result[1])
+    """
+
+    def find_chunk_index_by_time(self, seconds_behind):
+        """Find the chunk index that was received approximately *seconds_behind*
+        seconds ago.
+
+        Uses an atomic Lua script against the chunk_timestamps sorted set so
+        no data can expire between the lookup and the read.
+
+        Returns:
+            int or None: The chunk index to position the client at (this is
+                         the *last consumed* convention, so the next read
+                         starts at index+1).  None if no suitable chunk
+                         exists.
+        """
+        if not self.redis_client or not self.chunk_timestamps_key:
+            return None
+
+        target_time = time.time() - seconds_behind
+
+        try:
+            result = self._find_chunk_by_time_sha(
+                keys=[self.chunk_timestamps_key],
+                args=[target_time],
+            )
+            if result is None or int(result) == -1:
+                # No chunk old enough — fall back to the oldest available chunk
+                oldest = self.redis_client.zrange(self.chunk_timestamps_key, 0, 0)
+                if oldest:
+                    return max(0, int(oldest[0]) - 1)  # "last consumed" convention
+                return None
+
+            # Return index - 1 so next read starts at that chunk
+            return max(0, int(result) - 1)
+
+        except Exception as e:
+            logger.error(f"Error in find_chunk_index_by_time for channel {self.channel_id}: {e}")
+            return None
+
     # Add a new method to safely create timers
     def schedule_timer(self, delay, callback, *args, **kwargs):
         """Schedule a timer and track it for proper cleanup"""
diff --git a/apps/proxy/ts_proxy/stream_generator.py b/apps/proxy/ts_proxy/stream_generator.py
index a8ee799f..05892293 100644
--- a/apps/proxy/ts_proxy/stream_generator.py
+++ b/apps/proxy/ts_proxy/stream_generator.py
@@ -186,10 +186,47 @@ class StreamGenerator:
             logger.error(f"[{self.client_id}] No buffer found for channel {self.channel_id}")
             return False
 
-        # Client state tracking - use config for initial position
-        initial_behind = ConfigHelper.initial_behind_chunks()
+        # Client state tracking — determine start position
+        # When behind_seconds > 0, use time-based positioning to start
+        # the client that many seconds behind live.
+        # When behind_seconds == 0, start at live (buffer head).
+        behind_seconds = ConfigHelper.new_client_behind_seconds()
         current_buffer_index = buffer.index
-        self.local_index = max(0, current_buffer_index - initial_behind)
+
+        if behind_seconds > 0:
+            time_index = buffer.find_chunk_index_by_time(behind_seconds)
+            if time_index is not None:
+                self.local_index = max(0, time_index)
+                logger.info(
+                    f"[{self.client_id}] Time-based positioning: "
+                    f"{behind_seconds}s behind -> index {self.local_index} "
+                    f"(buffer head at {current_buffer_index})"
+                )
+            else:
+                # Not enough buffer for the requested time — start as far
+                # back as possible (oldest available chunk).
+                oldest = buffer.find_oldest_available_chunk(0)
+                if oldest is not None:
+                    self.local_index = max(0, oldest)
+                    logger.info(
+                        f"[{self.client_id}] Buffer shorter than {behind_seconds}s, "
+                        f"starting at oldest available chunk {self.local_index} "
+                        f"(buffer head at {current_buffer_index})"
+                    )
+                else:
+                    # No timestamp data at all — start at live
+                    self.local_index = current_buffer_index
+                    logger.info(
+                        f"[{self.client_id}] No timestamp data, starting at live: "
+                        f"index {self.local_index} (buffer head at {current_buffer_index})"
+                    )
+        else:
+            # 0 = start at live (buffer head)
+            self.local_index = current_buffer_index
+            logger.info(
+                f"[{self.client_id}] Starting at live (behind_seconds=0): "
+                f"index {self.local_index} (buffer head at {current_buffer_index})"
+            )
 
         # Store important objects as instance variables
         self.buffer = buffer
diff --git a/core/api_views.py b/core/api_views.py
index 3d131ab9..43e55d02 100644
--- a/core/api_views.py
+++ b/core/api_views.py
@@ -183,6 +183,7 @@ class ProxySettingsViewSet(viewsets.ViewSet):
                 "redis_chunk_ttl": 60,
                 "channel_shutdown_delay": 0,
                 "channel_init_grace_period": 5,
+                "new_client_behind_seconds": 2,
             }
             settings_obj, created = CoreSettings.objects.get_or_create(
                 key=PROXY_SETTINGS_KEY,
diff --git a/core/models.py b/core/models.py
index 10f9073f..7ce7713d 100644
--- a/core/models.py
+++ b/core/models.py
@@ -329,6 +329,7 @@ class CoreSettings(models.Model):
             "redis_chunk_ttl": 60,
             "channel_shutdown_delay": 0,
             "channel_init_grace_period": 5,
+            "new_client_behind_seconds": 2,
         })
 
     # System Settings
diff --git a/core/serializers.py b/core/serializers.py
index 03ba33e8..b0b730ee 100644
--- a/core/serializers.py
+++ b/core/serializers.py
@@ -78,6 +78,7 @@ class ProxySettingsSerializer(serializers.Serializer):
     redis_chunk_ttl = serializers.IntegerField(min_value=10, max_value=3600)
     channel_shutdown_delay = serializers.IntegerField(min_value=0, max_value=300)
     channel_init_grace_period = serializers.IntegerField(min_value=0, max_value=60)
+    new_client_behind_seconds = serializers.IntegerField(min_value=0, max_value=120, required=False, default=2)
 
     def validate_buffering_timeout(self, value):
         if value < 0 or value > 300:
@@ -104,6 +105,11 @@ class ProxySettingsSerializer(serializers.Serializer):
             raise serializers.ValidationError("Channel init grace period must be between 0 and 60 seconds")
         return value
 
+    def validate_new_client_behind_seconds(self, value):
+        if value < 0 or value > 120:
+            raise serializers.ValidationError("New client buffer must be between 0 and 120 seconds")
+        return value
+
 
 class SystemNotificationSerializer(serializers.ModelSerializer):
     """Serializer for system notifications."""
diff --git a/frontend/src/components/forms/settings/ProxySettingsForm.jsx b/frontend/src/components/forms/settings/ProxySettingsForm.jsx
index a5c8d858..52440769 100644
--- a/frontend/src/components/forms/settings/ProxySettingsForm.jsx
+++ b/frontend/src/components/forms/settings/ProxySettingsForm.jsx
@@ -24,6 +24,7 @@ const ProxySettingsOptions = React.memo(({ proxySettingsForm }) => {
       'redis_chunk_ttl',
       'channel_shutdown_delay',
       'channel_init_grace_period',
+      'new_client_behind_seconds',
     ].includes(key);
   };
   const isFloatField = (key) => {
@@ -36,7 +37,9 @@ const ProxySettingsOptions = React.memo(({ proxySettingsForm }) => {
         ? 3600
         : key === 'channel_shutdown_delay'
           ? 300
-          : 60;
+          : key === 'new_client_behind_seconds'
+            ? 120
+            : 60;
   };
   return (
     <>
@@ -97,7 +100,12 @@ const ProxySettingsForm = React.memo(({ active }) => {
   useEffect(() => {
     if (settings) {
       if (settings['proxy_settings']?.value) {
-        proxySettingsForm.setValues(settings['proxy_settings'].value);
+        // Merge defaults so any newly-added keys not yet in the stored
+        // settings object still show their default value rather than blank.
+        proxySettingsForm.setValues({
+          ...getProxySettingDefaults(),
+          ...settings['proxy_settings'].value,
+        });
       }
     }
   }, [settings]);
diff --git a/frontend/src/constants.js b/frontend/src/constants.js
index d2caa45b..0c19eea1 100644
--- a/frontend/src/constants.js
+++ b/frontend/src/constants.js
@@ -55,6 +55,11 @@ export const PROXY_SETTINGS_OPTIONS = {
     label: 'Channel Initialization Grace Period',
     description: 'Grace period in seconds during channel initialization',
   },
+  new_client_behind_seconds: {
+    label: 'New Client Buffer (seconds)',
+    description:
+      'Seconds of received buffer to start behind live when a new client connects (0 = start at live). Note: this is chunk receive time, not video duration.',
+  },
 };
 
 export const M3U_FILTER_TYPES = [
diff --git a/frontend/src/utils/forms/settings/ProxySettingsFormUtils.js b/frontend/src/utils/forms/settings/ProxySettingsFormUtils.js
index 9ca185e8..e6e0464b 100644
--- a/frontend/src/utils/forms/settings/ProxySettingsFormUtils.js
+++ b/frontend/src/utils/forms/settings/ProxySettingsFormUtils.js
@@ -14,5 +14,6 @@ export const getProxySettingDefaults = () => {
     redis_chunk_ttl: 60,
     channel_shutdown_delay: 0,
     channel_init_grace_period: 5,
+    new_client_behind_seconds: 2,
   };
 };

From b9a9d929cee8e642cf1fff62bfbda83cac242cdd Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Tue, 3 Mar 2026 18:21:52 -0600
Subject: [PATCH 34/58] tests: Fix proxy settings test.

---
 .../forms/settings/__tests__/ProxySettingsFormUtils.test.js     | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/frontend/src/utils/forms/settings/__tests__/ProxySettingsFormUtils.test.js b/frontend/src/utils/forms/settings/__tests__/ProxySettingsFormUtils.test.js
index 8feefccb..a48e5a37 100644
--- a/frontend/src/utils/forms/settings/__tests__/ProxySettingsFormUtils.test.js
+++ b/frontend/src/utils/forms/settings/__tests__/ProxySettingsFormUtils.test.js
@@ -61,6 +61,7 @@ describe('ProxySettingsFormUtils', () => {
         redis_chunk_ttl: 60,
         channel_shutdown_delay: 0,
         channel_init_grace_period: 5,
+        new_client_behind_seconds: 2,
       });
     });
 
@@ -80,6 +81,7 @@ describe('ProxySettingsFormUtils', () => {
       expect(typeof result.redis_chunk_ttl).toBe('number');
       expect(typeof result.channel_shutdown_delay).toBe('number');
       expect(typeof result.channel_init_grace_period).toBe('number');
+      expect(typeof result.new_client_behind_seconds).toBe('number');
     });
   });
 });

From bd1e060bd213822055f3ed1758954bc8ce332a01 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Thu, 26 Feb 2026 00:11:15 -0600
Subject: [PATCH 35/58] fix(dvr): fix duplicate recording execution and add
 manual stop

Fixes #940: Duplicate DVR task executions were caused by ClockedSchedule objects being shared across PeriodicTasks, triggering run_recording multiple
times for the same recording. Added an idempotency guard in run_recording that exits early if status is already 'recording', 'completed', or 'stopped'.

revoke_task() now deletes the PeriodicTask and orphaned ClockedSchedule on execution rather than relying on Celery revocation alone.

Fixes #454: Added POST /api/channels/recordings/{id}/stop/ action that marks the recording as stopped and terminates only the specific DVR proxy client (identified via User-Agent: Dispatcharr-DVR/recording-{id}) without disrupting
simultaneous recordings on the same or other channels. The API returns
immediately; DVR client teardown and task revocation happen in a background thread to avoid 504 timeouts on the recordings list endpoint.

Additional changes:
- destroy() now only calls _stop_dvr_clients() for in-progress recordings, preventing accidental stream termination when deleting completed recordings
- WebSocket events differentiated: recording_stopped, recording_cancelled (in-progress cancel), and recording_cancelled (delete) with was_in_progress flag so the frontend shows distinct "Recording deleted" vs "Recording cancelled" notifications

- Frontend: Stop and Cancel split into separate buttons with confirmation
  modals; stopped recordings move immediately to Previously Recorded via
  optimistic status bucketing; removed dead 'Stopped' badge state

- 27 new unit tests covering scheduling, idempotency, revocation, and DVR client isolation
---
 apps/channels/api_views.py                    | 176 ++++--
 apps/channels/signals.py                      |  98 +++-
 apps/channels/tasks.py                        | 241 ++++++--
 .../tests/test_recording_scheduling.py        | 547 ++++++++++++++++++
 frontend/src/WebSocket.jsx                    |  28 +
 frontend/src/api.js                           |  11 +
 .../src/components/cards/RecordingCard.jsx    | 125 +++-
 .../src/utils/cards/RecordingCardUtils.js     |   4 +
 frontend/src/utils/pages/DVRUtils.js          |   2 +-
 9 files changed, 1112 insertions(+), 120 deletions(-)
 create mode 100644 apps/channels/tests/test_recording_scheduling.py

diff --git a/apps/channels/api_views.py b/apps/channels/api_views.py
index 9147532b..bd94b552 100644
--- a/apps/channels/api_views.py
+++ b/apps/channels/api_views.py
@@ -2248,6 +2248,54 @@ class RecurringRecordingRuleViewSet(viewsets.ModelViewSet):
             logger.warning(f"Failed to purge recordings for rule {rule_id}: {err}")
 
 
+def _stop_dvr_clients(channel_uuid, recording_id=None):
+    """Stop DVR recording clients for a channel.
+
+    If recording_id is provided, only the client whose User-Agent contains that
+    recording ID is stopped (safe for simultaneous recordings on the same channel).
+    If recording_id is None, all Dispatcharr-DVR clients for the channel are stopped
+    (used by destroy() when deleting a recording whose task_id is unknown).
+
+    Returns the number of DVR clients stopped.
+    """
+    from core.utils import RedisClient
+    from apps.proxy.ts_proxy.redis_keys import RedisKeys
+    from apps.proxy.ts_proxy.services.channel_service import ChannelService
+
+    r = RedisClient.get_client()
+    if not r:
+        return 0
+    client_set_key = RedisKeys.clients(channel_uuid)
+    client_ids = r.smembers(client_set_key) or []
+    stopped = 0
+    for raw_id in client_ids:
+        try:
+            cid = raw_id.decode("utf-8") if isinstance(raw_id, (bytes, bytearray)) else str(raw_id)
+            meta_key = RedisKeys.client_metadata(channel_uuid, cid)
+            ua = r.hget(meta_key, "user_agent")
+            ua_s = ua.decode("utf-8") if isinstance(ua, (bytes, bytearray)) else (ua or "")
+            if not (ua_s and "Dispatcharr-DVR" in ua_s):
+                continue
+            # When a recording_id is specified, only stop the client for that recording.
+            # Each run_recording task connects with User-Agent "Dispatcharr-DVR/recording-{id}",
+            # so we can safely target just this recording without affecting others on the channel.
+            if recording_id is not None and f"recording-{recording_id}" not in ua_s:
+                continue
+            try:
+                ChannelService.stop_client(channel_uuid, cid)
+                stopped += 1
+            except Exception as inner_e:
+                logger.debug(f"Failed to stop DVR client {cid} for channel {channel_uuid}: {inner_e}")
+        except Exception as inner:
+            logger.debug(f"Error while checking client metadata: {inner}")
+    # Do not call ChannelService.stop_channel() here.
+    # Stopping the channel proxy would terminate the source connection which may
+    # be shared with other recordings on the same channel.  The TS proxy server
+    # already detects when client count reaches zero and tears down the channel
+    # cleanly on its own (with the configured shutdown delay).
+    return stopped
+
+
 class RecordingViewSet(viewsets.ModelViewSet):
     queryset = Recording.objects.all()
     serializer_class = RecordingSerializer
@@ -2342,69 +2390,105 @@ class RecordingViewSet(viewsets.ModelViewSet):
         response["Content-Disposition"] = f"inline; filename=\"{file_name}\""
         return response
 
+    @action(detail=True, methods=["post"], url_path="stop")
+    def stop(self, request, pk=None):
+        """Stop a recording early while retaining the partial content for playback."""
+        instance = self.get_object()
+
+        # Mark as stopped in the DB first so run_recording detects it.
+        # This is the only operation that MUST be synchronous — run_recording reads
+        # the status field to decide whether the stream disconnection was deliberate.
+        cp = instance.custom_properties or {}
+        cp["status"] = "stopped"
+        cp["stopped_at"] = str(timezone.now())
+        instance.custom_properties = cp
+        instance.save(update_fields=["custom_properties"])
+
+        # Everything else (DVR client teardown, task revocation, WebSocket notification)
+        # is moved to a daemon thread so the API returns immediately.  Callers were
+        # seeing 5-15 s delays because _stop_dvr_clients / revoke_task / async_to_sync
+        # have occasional slow paths (Redis timeouts, Celery control broadcasts, etc.).
+        channel_uuid = str(instance.channel.uuid)
+        recording_id = instance.id
+        task_id = instance.task_id
+        channel_name = instance.channel.name
+
+        def _background_stop():
+            try:
+                stopped = _stop_dvr_clients(channel_uuid, recording_id=recording_id)
+                if stopped:
+                    logger.info(
+                        f"Stopped {stopped} DVR client(s) for channel {channel_uuid} (recording stopped early)"
+                    )
+            except Exception as e:
+                logger.debug(f"Unable to stop DVR clients for stopped recording: {e}")
+
+            try:
+                from apps.channels.signals import revoke_task
+                revoke_task(task_id)
+            except Exception as e:
+                logger.debug(f"Unable to revoke task for stopped recording: {e}")
+
+            try:
+                from core.utils import send_websocket_update
+                send_websocket_update('updates', 'update', {
+                    "success": True,
+                    "type": "recording_stopped",
+                    "channel": channel_name,
+                })
+            except Exception:
+                pass
+
+            # Close the DB connection opened by this thread so it isn't leaked
+            try:
+                from django.db import connection as _conn
+                _conn.close()
+            except Exception:
+                pass
+
+        import threading
+        threading.Thread(target=_background_stop, daemon=True).start()
+
+        return Response({"success": True, "status": "stopped"})
+
     def destroy(self, request, *args, **kwargs):
         """Delete the Recording and ensure any active DVR client connection is closed.
 
         Also removes the associated file(s) from disk if present.
         """
         instance = self.get_object()
-
-        # Attempt to close the DVR client connection for this channel if active
-        try:
-            channel_uuid = str(instance.channel.uuid)
-            # Lazy imports to avoid module overhead if proxy isn't used
-            from core.utils import RedisClient
-            from apps.proxy.ts_proxy.redis_keys import RedisKeys
-            from apps.proxy.ts_proxy.services.channel_service import ChannelService
-
-            r = RedisClient.get_client()
-            if r:
-                client_set_key = RedisKeys.clients(channel_uuid)
-                client_ids = r.smembers(client_set_key) or []
-                stopped = 0
-                for raw_id in client_ids:
-                    try:
-                        cid = raw_id.decode("utf-8") if isinstance(raw_id, (bytes, bytearray)) else str(raw_id)
-                        meta_key = RedisKeys.client_metadata(channel_uuid, cid)
-                        ua = r.hget(meta_key, "user_agent")
-                        ua_s = ua.decode("utf-8") if isinstance(ua, (bytes, bytearray)) else (ua or "")
-                        # Identify DVR recording client by its user agent
-                        if ua_s and "Dispatcharr-DVR" in ua_s:
-                            try:
-                                ChannelService.stop_client(channel_uuid, cid)
-                                stopped += 1
-                            except Exception as inner_e:
-                                logger.debug(f"Failed to stop DVR client {cid} for channel {channel_uuid}: {inner_e}")
-                    except Exception as inner:
-                        logger.debug(f"Error while checking client metadata: {inner}")
-                if stopped:
-                    logger.info(f"Stopped {stopped} DVR client(s) for channel {channel_uuid} due to recording cancellation")
-                # If no clients remain after stopping DVR clients, proactively stop the channel
-                try:
-                    remaining = r.scard(client_set_key) or 0
-                except Exception:
-                    remaining = 0
-                if remaining == 0:
-                    try:
-                        ChannelService.stop_channel(channel_uuid)
-                        logger.info(f"Stopped channel {channel_uuid} (no clients remain)")
-                    except Exception as sc_e:
-                        logger.debug(f"Unable to stop channel {channel_uuid}: {sc_e}")
-        except Exception as e:
-            logger.debug(f"Unable to stop DVR clients for cancelled recording: {e}")
+        channel_name = instance.channel.name
 
         # Capture paths before deletion
         cp = instance.custom_properties or {}
+        rec_status = cp.get("status", "")
+
+        # Only stop the DVR client if this recording is actively streaming.
+        # Stopping for completed/upcoming recordings would kill an unrelated
+        # in-progress recording on the same channel.
+        if rec_status == "recording":
+            try:
+                channel_uuid = str(instance.channel.uuid)
+                stopped = _stop_dvr_clients(channel_uuid)
+                if stopped:
+                    logger.info(f"Stopped {stopped} DVR client(s) for channel {channel_uuid} due to recording cancellation")
+            except Exception as e:
+                logger.debug(f"Unable to stop DVR clients for cancelled recording: {e}")
         file_path = cp.get("file_path")
         temp_ts_path = cp.get("_temp_file_path")
 
         # Perform DB delete first, then try to remove files
         response = super().destroy(request, *args, **kwargs)
 
-        # Notify frontends to refresh recordings
+        # Notify frontends
         try:
             from core.utils import send_websocket_update
-            send_websocket_update('updates', 'update', {"success": True, "type": "recordings_refreshed"})
+            send_websocket_update('updates', 'update', {
+                "success": True,
+                "type": "recording_cancelled",
+                "channel": channel_name,
+                "was_in_progress": rec_status == "recording",
+            })
         except Exception:
             pass
 
diff --git a/apps/channels/signals.py b/apps/channels/signals.py
index 75ef16dd..00b8fb7c 100644
--- a/apps/channels/signals.py
+++ b/apps/channels/signals.py
@@ -2,14 +2,15 @@
 
 from django.db.models.signals import m2m_changed, pre_save, post_save, post_delete
 from django.dispatch import receiver
-from django.utils.timezone import now
+from django.utils.timezone import now, is_aware, make_aware
 from celery.result import AsyncResult
+from django_celery_beat.models import ClockedSchedule, PeriodicTask
 from .models import Channel, Stream, ChannelProfile, ChannelProfileMembership, Recording
 from apps.m3u.models import M3UAccount
 from apps.epg.tasks import parse_programs_for_tvg_id
-import logging, requests, time
+import json
+import logging
 from .tasks import run_recording, prefetch_recording_artwork
-from django.utils.timezone import now, is_aware, make_aware
 from datetime import timedelta
 
 logger = logging.getLogger(__name__)
@@ -85,29 +86,76 @@ def create_profile_memberships(sender, instance, created, **kwargs):
             for channel in channels
         ])
 
+def _dvr_task_name(recording_id):
+    """Predictable PeriodicTask name for a DVR recording."""
+    return f"dvr-recording-{recording_id}"
+
+
 def schedule_recording_task(instance, eta=None):
-    # Use the explicitly-passed (and timezone-aware) eta if provided;
-    # fall back to instance.start_time only as a last resort.
+    """Schedule a recording task via ClockedSchedule + one-off PeriodicTask.
+
+    The task is stored in the database and dispatched by Celery Beat at the
+    scheduled time with no countdown.  This avoids the Redis visibility_timeout
+    redelivery bug that caused duplicate recordings when using apply_async
+    with long countdowns.  See https://github.com/Dispatcharr/Dispatcharr/issues/940
+    """
     if eta is None:
         eta = instance.start_time
-    # Ensure eta is timezone-aware before comparing against now()
     if eta is not None and not is_aware(eta):
         eta = make_aware(eta)
-    # countdown=0 fires immediately (in-progress programs whose start_time was
-    # clamped to now by the serializer), countdown>0 delays until start_time
-    # (future programs).  Using an integer countdown avoids any timezone
-    # serialization ambiguity that can occur with an absolute eta datetime.
-    countdown = max(0, int((eta - now()).total_seconds()))
-    # Pass recording_id first so task can persist metadata to the correct row
-    task = run_recording.apply_async(
-        args=[instance.id, instance.channel_id, str(instance.start_time), str(instance.end_time)],
-        countdown=countdown,
+    # Clamp to now so Beat dispatches immediately for past/current start times
+    if eta <= now():
+        eta = now()
+
+    task_args = [
+        instance.id,
+        instance.channel_id,
+        str(instance.start_time),
+        str(instance.end_time),
+    ]
+
+    clocked, _ = ClockedSchedule.objects.get_or_create(clocked_time=eta)
+    task_name = _dvr_task_name(instance.id)
+    PeriodicTask.objects.update_or_create(
+        name=task_name,
+        defaults={
+            "task": "apps.channels.tasks.run_recording",
+            "clocked": clocked,
+            "args": json.dumps(task_args),
+            "one_off": True,
+            "enabled": True,
+            "interval": None,
+            "crontab": None,
+            "solar": None,
+        },
     )
-    return task.id
+    return task_name
+
 
 def revoke_task(task_id):
-    if task_id:
+    """Cancel a pending recording task.
+
+    task_id is normally a PeriodicTask name (e.g. "dvr-recording-42").
+    For backwards compatibility with task_ids stored before this fix
+    (Celery async-result UUIDs), falls back to AsyncResult.revoke().
+    """
+    if not task_id:
+        return
+    # Primary path: delete the PeriodicTask and clean up its ClockedSchedule
+    try:
+        pt = PeriodicTask.objects.get(name=task_id)
+        old_clocked = pt.clocked
+        pt.delete()
+        if old_clocked and not PeriodicTask.objects.filter(clocked=old_clocked).exists():
+            old_clocked.delete()
+        return
+    except PeriodicTask.DoesNotExist:
+        pass
+    # Fallback for legacy Celery task UUIDs
+    try:
         AsyncResult(task_id).revoke()
+    except Exception:
+        pass
 
 @receiver(pre_save, sender=Recording)
 def revoke_old_task_on_update(sender, instance, **kwargs):
@@ -152,11 +200,17 @@ def schedule_task_on_save(sender, instance, created, **kwargs):
                 instance.save(update_fields=['task_id'])
             else:
                 print("Start time is in the past. Not scheduling.")
-        # Kick off poster/artwork prefetch to enrich Upcoming cards
-        try:
-            prefetch_recording_artwork.apply_async(args=[instance.id], countdown=1)
-        except Exception as e:
-            print("Error scheduling artwork prefetch:", e)
+        # Kick off poster/artwork prefetch to enrich Upcoming cards.
+        # Skip when the recording is already active or finished — run_recording
+        # handles its own poster resolution, and scheduling artwork prefetch
+        # while the task is running causes a race that can overwrite status.
+        cp = instance.custom_properties or {}
+        rec_status = cp.get("status", "")
+        if rec_status not in ("recording", "completed", "stopped", "interrupted"):
+            try:
+                prefetch_recording_artwork.apply_async(args=[instance.id], countdown=1)
+            except Exception as e:
+                print("Error scheduling artwork prefetch:", e)
     except Exception as e:
         import traceback
         print("Error in post_save signal:", e)
diff --git a/apps/channels/tasks.py b/apps/channels/tasks.py
index 8d49287b..87abb90d 100755
--- a/apps/channels/tasks.py
+++ b/apps/channels/tasks.py
@@ -1497,6 +1497,44 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     - Attempts to capture stream stats from TS proxy (codec, resolution, fps, etc.)
     - Attempts to capture a poster (via program.custom_properties) and store a Logo reference
     """
+    from .models import Recording, Logo
+
+    # --- Idempotency guard (prevents duplicate recordings from task redelivery) ---
+    try:
+        rec_check = Recording.objects.filter(id=recording_id).only("custom_properties").first()
+        if not rec_check:
+            # Recording was deleted before this task got a chance to run (e.g. user
+            # cancelled from the Upcoming list, or destroy() ran concurrently).
+            logger.info(
+                f"run_recording called for recording {recording_id} but it no longer exists — "
+                f"skipping (recording was cancelled before task started)."
+            )
+            return
+        status = (rec_check.custom_properties or {}).get("status", "")
+        if status in ("recording", "completed", "stopped"):
+            logger.warning(
+                f"run_recording called for recording {recording_id} but status "
+                f"is already '{status}' - skipping duplicate execution."
+            )
+            return
+    except Exception as e:
+        logger.debug(f"Idempotency guard check failed (proceeding anyway): {e}")
+
+    # --- Clean up the one-off PeriodicTask that dispatched task ---
+    try:
+        from django_celery_beat.models import ClockedSchedule as _CS, PeriodicTask as _PT
+        task_name = f"dvr-recording-{recording_id}"
+        try:
+            pt = _PT.objects.get(name=task_name)
+            old_clocked = pt.clocked
+            pt.delete()
+            if old_clocked and not _PT.objects.filter(clocked=old_clocked).exists():
+                old_clocked.delete()
+        except _PT.DoesNotExist:
+            pass
+    except Exception as e:
+        logger.debug(f"PeriodicTask cleanup failed (non-fatal): {e}")
+
     channel = Channel.objects.get(id=channel_id)
 
     start_time = datetime.fromisoformat(start_time_str)
@@ -1536,7 +1574,6 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     # Try to resolve the Recording row up front
     recording_obj = None
     try:
-        from .models import Recording, Logo
         recording_obj = Recording.objects.get(id=recording_id)
         # Prime custom_properties with file info/status
         cp = recording_obj.custom_properties or {}
@@ -1822,7 +1859,7 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
             with requests.get(
                 test_url,
                 headers={
-                    'User-Agent': 'Dispatcharr-DVR',
+                    'User-Agent': f'Dispatcharr-DVR/recording-{recording_id}',
                 },
                 stream=True,
                 timeout=(10, 15),
@@ -1878,6 +1915,38 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
         except Exception as e:
             last_error = str(e)
             logger.warning(f"DVR: attempt failed for base {base}: {e}")
+            # If there was a working stream that was interrupted (e.g. recording
+            # cancelled via stop_client), do not retry on another base — that would
+            # re-establish the DVR connection and make the recording re-appear.
+            if bytes_written > 0:
+                # Check if this was a deliberate "stop" vs an unexpected interruption
+                try:
+                    rec_check = Recording.objects.filter(id=recording_id).only("custom_properties").first()
+                    if rec_check and (rec_check.custom_properties or {}).get("status") == "stopped":
+                        # Graceful stop — not an error, just an early end
+                        interrupted = False
+                        logger.info(f"Recording {recording_id} was stopped by user — ending stream.")
+                    else:
+                        interrupted = True
+                        interrupted_reason = f"stream_interrupted: {e}"
+                except Exception:
+                    interrupted = True
+                    interrupted_reason = f"stream_interrupted: {e}"
+                break
+            # Also bail if the Recording was deleted (user cancelled before data arrived)
+            rec_exists = Recording.objects.filter(id=recording_id).exists()
+            if not rec_exists:
+                interrupted = True
+                interrupted_reason = "recording_deleted_during_connect"
+                break
+            # Check for "stopped" status even before data arrived
+            try:
+                rec_check = Recording.objects.filter(id=recording_id).only("custom_properties").first()
+                if rec_check and (rec_check.custom_properties or {}).get("status") == "stopped":
+                    interrupted = False
+                    break
+            except Exception:
+                pass
 
     if chosen_base is None and bytes_written == 0:
         interrupted = True
@@ -1892,39 +1961,48 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
         # We cannot compute accurate elapsed here; fine to leave as is
         pass
 
-    # If no bytes were written at all, mark detail
+    # If no bytes were written at all, check whether this was a deliberate stop or a
+    # genuine failure.  The exception handler above already sets interrupted=False when
+    # it detects "stopped" status, but do not override that decision here.
     if bytes_written == 0 and not interrupted:
-        interrupted = True
-        interrupted_reason = f"no_stream_data: {last_error or 'unknown'}"
-
-        # Update DB status immediately so the UI reflects the change on the event below
+        _deliberately_stopped = False
         try:
-            if recording_obj is None:
-                from .models import Recording
-                recording_obj = Recording.objects.get(id=recording_id)
-            cp_now = recording_obj.custom_properties or {}
-            cp_now.update({
-                "status": "interrupted" if interrupted else "completed",
-                "ended_at": str(datetime.now()),
-                "file_name": filename or cp_now.get("file_name"),
-                "file_path": final_path or cp_now.get("file_path"),
-            })
-            if interrupted and interrupted_reason:
-                cp_now["interrupted_reason"] = interrupted_reason
-            recording_obj.custom_properties = cp_now
-            recording_obj.save(update_fields=["custom_properties"])
-        except Exception as e:
-            logger.debug(f"Failed to update immediate recording status: {e}")
+            _rc = Recording.objects.filter(id=recording_id).only("custom_properties").first()
+            if _rc and (_rc.custom_properties or {}).get("status") == "stopped":
+                _deliberately_stopped = True
+        except Exception:
+            pass
 
-        async_to_sync(channel_layer.group_send)(
-            "updates",
-            {
-                "type": "update",
-                "data": {"success": True, "type": "recording_ended", "channel": channel.name}
-            },
-        )
-        # After the loop, the file and response are closed automatically.
-        logger.info(f"Finished recording for channel {channel.name}")
+        if not _deliberately_stopped:
+            interrupted = True
+            interrupted_reason = f"no_stream_data: {last_error or 'unknown'}"
+
+            # Update DB status immediately so the UI reflects the change on the event below
+            try:
+                if recording_obj is None:
+                    recording_obj = Recording.objects.get(id=recording_id)
+                cp_now = recording_obj.custom_properties or {}
+                cp_now.update({
+                    "status": "interrupted",
+                    "ended_at": str(datetime.now()),
+                    "file_name": filename or cp_now.get("file_name"),
+                    "file_path": final_path or cp_now.get("file_path"),
+                    "interrupted_reason": interrupted_reason,
+                })
+                recording_obj.custom_properties = cp_now
+                recording_obj.save(update_fields=["custom_properties"])
+            except Exception as e:
+                logger.debug(f"Failed to update immediate recording status: {e}")
+
+            async_to_sync(channel_layer.group_send)(
+                "updates",
+                {
+                    "type": "update",
+                    "data": {"success": True, "type": "recording_ended", "channel": channel.name}
+                },
+            )
+            # After the loop, the file and response are closed automatically.
+            logger.info(f"Finished recording for channel {channel.name}")
 
         # Log system event for recording end
         try:
@@ -1940,6 +2018,20 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
         except Exception as e:
             logger.error(f"Could not log recording end event: {e}")
 
+    # If the Recording was deleted (cancelled by user), skip post-processing
+    recording_cancelled = not Recording.objects.filter(id=recording_id).exists()
+    if recording_cancelled:
+        logger.info(f"Recording {recording_id} was cancelled — skipping remux and metadata.")
+        # Clean up any remaining temp or stub output files
+        for _cleanup_path in [temp_ts_path, final_path]:
+            try:
+                if _cleanup_path and os.path.exists(_cleanup_path):
+                    os.remove(_cleanup_path)
+                    logger.info(f"Cleaned up cancelled recording artifact: {_cleanup_path}")
+            except Exception:
+                pass
+        return
+
     # Remux TS to MKV container
     remux_success = False
     try:
@@ -2039,19 +2131,25 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     # Persist final metadata to Recording (status, ended_at, and stream stats if available)
     try:
         if recording_obj is None:
-            from .models import Recording
             recording_obj = Recording.objects.get(id=recording_id)
 
+        # Re-read from DB to get the latest status (stop endpoint may have set it)
+        recording_obj.refresh_from_db()
         cp = recording_obj.custom_properties or {}
-        cp.update({
-            "ended_at": str(datetime.now()),
-        })
-        if interrupted:
+        cp["ended_at"] = str(datetime.now())
+
+        # Determine final status.
+        # A deliberate user stop ("stopped" in DB) always wins over any interrupted flag —
+        # this handles races where the stream connection broke milliseconds before or
+        # after the stop endpoint committed its DB save.
+        db_status_now = cp.get("status", "")
+        if db_status_now == "stopped" or not interrupted:
+            cp["status"] = "completed"
+            cp.pop("interrupted_reason", None)
+        else:
             cp["status"] = "interrupted"
             if interrupted_reason:
                 cp["interrupted_reason"] = interrupted_reason
-        else:
-            cp["status"] = "completed"
         cp["bytes_written"] = bytes_written
         cp["remux_success"] = remux_success
 
@@ -2112,8 +2210,29 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
 
         # Removed: local thumbnail generation. We rely on EPG/VOD/TMDB/OMDb/keyless providers only.
 
+        # Final cancellation guard: destroy() may have deleted the record while
+        # remuxing.  If it's gone now, skip saving "interrupted" status and
+        # skip the notification — destroy() already sent recording_cancelled.
+        if not Recording.objects.filter(id=recording_id).exists():
+            logger.info(
+                f"Recording {recording_id} was deleted during post-processing — skipping final save."
+            )
+            return
+
         recording_obj.custom_properties = cp
         recording_obj.save(update_fields=["custom_properties"])
+
+        # Notify frontends so the UI refreshes immediately (e.g. "Stopped" → "Completed")
+        try:
+            async_to_sync(channel_layer.group_send)(
+                "updates",
+                {
+                    "type": "update",
+                    "data": {"success": True, "type": "recording_ended", "channel": channel.name},
+                },
+            )
+        except Exception:
+            pass
     except Exception as e:
         logger.debug(f"Unable to finalize Recording metadata: {e}")
 
@@ -2154,6 +2273,20 @@ def recover_recordings_on_startup():
         for rec in active:
             try:
                 cp = rec.custom_properties or {}
+                current_status = cp.get("status", "")
+
+                # Skip recordings that are already in a terminal or active state.
+                # "completed" / "stopped" — user stopped or it finished normally; do NOT
+                # overwrite the status and re-schedule (that would cause the
+                # Interrupted → In-Progress → Previously-Recorded ghost cycle).
+                # "recording" — a worker is already streaming; leave it alone.
+                if current_status in ("completed", "stopped", "recording"):
+                    logger.info(
+                        f"recover_recordings_on_startup: skipping recording {rec.id} "
+                        f"(status={current_status!r}, already in terminal/active state)."
+                    )
+                    continue
+
                 # Mark interrupted due to restart; will flip to 'recording' when task starts
                 cp["status"] = "interrupted"
                 cp["interrupted_reason"] = "server_restarted"
@@ -2167,11 +2300,20 @@ def recover_recordings_on_startup():
             except Exception as e:
                 logger.warning(f"Failed to resume recording {rec.id}: {e}")
 
-        # Ensure future recordings are scheduled
+        # Ensure future recordings are scheduled.
+        # With ClockedSchedule, PeriodicTasks survive restarts in the DB.
+        # Only recreate if the PeriodicTask is missing (safety net).
         upcoming = Recording.objects.filter(start_time__gt=now, end_time__gt=now)
         for rec in upcoming:
             try:
-                # Schedule task at start_time
+                from django_celery_beat.models import PeriodicTask as _PT
+                task_name = f"dvr-recording-{rec.id}"
+                if _PT.objects.filter(name=task_name).exists():
+                    if rec.task_id != task_name:
+                        rec.task_id = task_name
+                        rec.save(update_fields=["task_id"])
+                    continue
+                # PeriodicTask missing - recreate it
                 task_id = schedule_recording_task(rec)
                 if task_id:
                     rec.task_id = task_id
@@ -2553,6 +2695,13 @@ def prefetch_recording_artwork(recording_id):
         from .models import Recording
         rec = Recording.objects.get(id=recording_id)
         cp = rec.custom_properties or {}
+
+        # Bail out if the recording is already active or finished — run_recording
+        # handles poster resolution itself, and saving here can race with status updates.
+        current_status = cp.get("status", "")
+        if current_status in ("recording", "completed", "stopped", "interrupted"):
+            return "skipped: status is " + current_status
+
         program = cp.get("program") or {}
         poster_logo_id, poster_url = _resolve_poster_for_program(rec.channel.name, program)
         updated = False
@@ -2593,7 +2742,15 @@ def prefetch_recording_artwork(recording_id):
             pass
 
         if updated:
-            rec.custom_properties = cp
+            # Re-read from DB to avoid overwriting status changes made by
+            # the stop endpoint or run_recording's final metadata save.
+            rec.refresh_from_db()
+            fresh_cp = rec.custom_properties or {}
+            for key in ("poster_logo_id", "poster_url", "rating", "rating_system",
+                        "season", "episode", "onscreen_episode"):
+                if key in cp:
+                    fresh_cp[key] = cp[key]
+            rec.custom_properties = fresh_cp
             rec.save(update_fields=["custom_properties"])
             try:
                 from core.utils import send_websocket_update
diff --git a/apps/channels/tests/test_recording_scheduling.py b/apps/channels/tests/test_recording_scheduling.py
new file mode 100644
index 00000000..56bdbe87
--- /dev/null
+++ b/apps/channels/tests/test_recording_scheduling.py
@@ -0,0 +1,547 @@
+"""Tests for DVR recording scheduling with ClockedSchedule.
+
+   Prefers Celery ClockedScheduled over Redis time due to 3600s limitation on scheduled tasks;
+   scheduled tasks beyond 3600s cause Redis time to hand process between workers and duplicate jobs
+"""
+from datetime import timedelta
+from unittest.mock import patch, MagicMock
+
+from django.test import TestCase
+from django.utils import timezone
+from django_celery_beat.models import ClockedSchedule, PeriodicTask
+
+from apps.channels.models import Channel, Recording
+from apps.channels.signals import (
+    schedule_recording_task,
+    revoke_task,
+    _dvr_task_name,
+)
+
+
+class ScheduleRecordingTaskTests(TestCase):
+    """Tests for schedule_recording_task()."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(channel_number=1, name="Test Channel")
+
+    def tearDown(self):
+        PeriodicTask.objects.filter(name__startswith="dvr-recording-").delete()
+        ClockedSchedule.objects.all().delete()
+
+    @patch("apps.channels.signals.run_recording")
+    def test_future_recording_creates_periodic_task(self, mock_run_recording):
+        """Recordings in the future create a ClockedSchedule + PeriodicTask."""
+        future_time = timezone.now() + timedelta(hours=2)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=future_time,
+            end_time=future_time + timedelta(hours=1),
+        )
+
+        task_id = schedule_recording_task(rec, eta=future_time)
+
+        expected_name = f"dvr-recording-{rec.id}"
+        self.assertEqual(task_id, expected_name)
+
+        pt = PeriodicTask.objects.get(name=expected_name)
+        self.assertTrue(pt.one_off)
+        self.assertTrue(pt.enabled)
+        self.assertEqual(pt.task, "apps.channels.tasks.run_recording")
+        self.assertIsNotNone(pt.clocked)
+
+        # apply_async should not have been called
+        mock_run_recording.apply_async.assert_not_called()
+
+    @patch("apps.channels.signals.run_recording")
+    def test_immediate_recording_creates_periodic_task(self, mock_run_recording):
+        """Recordings starting now also use ClockedSchedule for consistency."""
+        now = timezone.now()
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=now,
+            end_time=now + timedelta(hours=1),
+        )
+
+        task_id = schedule_recording_task(rec, eta=now)
+
+        expected_name = f"dvr-recording-{rec.id}"
+        self.assertEqual(task_id, expected_name)
+        self.assertTrue(PeriodicTask.objects.filter(name=expected_name).exists())
+
+    @patch("apps.channels.signals.run_recording")
+    def test_past_start_time_clamps_to_now(self, mock_run_recording):
+        """Recordings with past start_time get clamped to now."""
+        past_time = timezone.now() - timedelta(minutes=5)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=past_time,
+            end_time=timezone.now() + timedelta(hours=1),
+        )
+
+        task_id = schedule_recording_task(rec, eta=past_time)
+
+        expected_name = f"dvr-recording-{rec.id}"
+        self.assertEqual(task_id, expected_name)
+        pt = PeriodicTask.objects.get(name=expected_name)
+        # Clocked time should be >= now
+        self.assertGreaterEqual(pt.clocked.clocked_time, past_time)
+
+    @patch("apps.channels.signals.run_recording")
+    def test_reschedule_updates_existing_periodic_task(self, mock_run_recording):
+        """Calling schedule_recording_task twice updates the existing PeriodicTask."""
+        future_time = timezone.now() + timedelta(hours=2)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=future_time,
+            end_time=future_time + timedelta(hours=1),
+        )
+
+        schedule_recording_task(rec, eta=future_time)
+
+        # Reschedule with a different time
+        new_eta = future_time + timedelta(hours=1)
+        schedule_recording_task(rec, eta=new_eta)
+
+        # Should still be exactly one PeriodicTask
+        task_name = f"dvr-recording-{rec.id}"
+        self.assertEqual(PeriodicTask.objects.filter(name=task_name).count(), 1)
+
+    @patch("apps.channels.signals.run_recording")
+    def test_naive_eta_is_made_aware(self, mock_run_recording):
+        """A naive (timezone-unaware) eta is made timezone-aware."""
+        from datetime import datetime
+        naive_eta = datetime(2030, 6, 15, 14, 0, 0)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=timezone.now() + timedelta(hours=1),
+            end_time=timezone.now() + timedelta(hours=2),
+        )
+
+        task_id = schedule_recording_task(rec, eta=naive_eta)
+
+        expected_name = f"dvr-recording-{rec.id}"
+        self.assertEqual(task_id, expected_name)
+        pt = PeriodicTask.objects.get(name=expected_name)
+        self.assertTrue(timezone.is_aware(pt.clocked.clocked_time))
+
+
+class RevokeTaskTests(TestCase):
+    """Tests for revoke_task()."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(channel_number=1, name="Test Channel")
+
+    def tearDown(self):
+        PeriodicTask.objects.filter(name__startswith="dvr-recording-").delete()
+        ClockedSchedule.objects.all().delete()
+
+    def test_revoke_deletes_periodic_task_and_clocked_schedule(self):
+        """revoke_task deletes the PeriodicTask and orphaned ClockedSchedule."""
+        eta = timezone.now() + timedelta(hours=5)
+        clocked = ClockedSchedule.objects.create(clocked_time=eta)
+        PeriodicTask.objects.create(
+            name="dvr-recording-10",
+            task="apps.channels.tasks.run_recording",
+            clocked=clocked,
+            one_off=True,
+            enabled=True,
+        )
+
+        revoke_task("dvr-recording-10")
+
+        self.assertFalse(PeriodicTask.objects.filter(name="dvr-recording-10").exists())
+        self.assertFalse(ClockedSchedule.objects.filter(id=clocked.id).exists())
+
+    def test_revoke_keeps_shared_clocked_schedule(self):
+        """ClockedSchedule is kept if another PeriodicTask still references it."""
+        eta = timezone.now() + timedelta(hours=5)
+        clocked = ClockedSchedule.objects.create(clocked_time=eta)
+        PeriodicTask.objects.create(
+            name="dvr-recording-10",
+            task="apps.channels.tasks.run_recording",
+            clocked=clocked,
+            one_off=True,
+        )
+        PeriodicTask.objects.create(
+            name="dvr-recording-11",
+            task="apps.channels.tasks.run_recording",
+            clocked=clocked,
+            one_off=True,
+        )
+
+        revoke_task("dvr-recording-10")
+
+        self.assertFalse(PeriodicTask.objects.filter(name="dvr-recording-10").exists())
+        self.assertTrue(ClockedSchedule.objects.filter(id=clocked.id).exists())
+
+    @patch("apps.channels.signals.AsyncResult")
+    def test_revoke_falls_back_to_async_result_for_legacy_ids(self, mock_async_result):
+        """revoke_task falls back to AsyncResult.revoke() for old-style UUIDs."""
+        revoke_task("550e8400-e29b-41d4-a716-446655440000")
+
+        mock_async_result.assert_called_once_with("550e8400-e29b-41d4-a716-446655440000")
+        mock_async_result.return_value.revoke.assert_called_once()
+
+    def test_revoke_none_is_noop(self):
+        """revoke_task(None) does nothing."""
+        revoke_task(None)  # Should not raise
+
+    def test_revoke_empty_string_is_noop(self):
+        """revoke_task('') does nothing."""
+        revoke_task("")  # Should not raise
+
+
+class DvrTaskNameTests(TestCase):
+    """Tests for the naming convention helper."""
+
+    def test_task_name_format(self):
+        self.assertEqual(_dvr_task_name(42), "dvr-recording-42")
+
+    def test_task_name_fits_in_charfield(self):
+        name = _dvr_task_name(999999999)
+        self.assertLessEqual(len(name), 255)
+
+
+class SignalIntegrationTests(TestCase):
+    """Integration tests for the post_save / post_delete signal handlers."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(channel_number=1, name="Test Channel")
+
+    def tearDown(self):
+        PeriodicTask.objects.filter(name__startswith="dvr-recording-").delete()
+        ClockedSchedule.objects.all().delete()
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_post_save_creates_periodic_task_for_future_recording(self, mock_artwork):
+        """Saving a future Recording creates a PeriodicTask via post_save signal."""
+        mock_artwork.apply_async.return_value = MagicMock()
+
+        future = timezone.now() + timedelta(hours=2)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=future,
+            end_time=future + timedelta(hours=1),
+        )
+
+        rec.refresh_from_db()
+        task_name = f"dvr-recording-{rec.id}"
+        self.assertEqual(rec.task_id, task_name)
+        self.assertTrue(PeriodicTask.objects.filter(name=task_name).exists())
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_post_delete_removes_periodic_task(self, mock_artwork):
+        """Deleting a Recording removes its PeriodicTask."""
+        mock_artwork.apply_async.return_value = MagicMock()
+
+        future = timezone.now() + timedelta(hours=2)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=future,
+            end_time=future + timedelta(hours=1),
+        )
+
+        rec.refresh_from_db()
+        task_name = rec.task_id
+        self.assertTrue(PeriodicTask.objects.filter(name=task_name).exists())
+
+        rec.delete()
+        self.assertFalse(PeriodicTask.objects.filter(name=task_name).exists())
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_bulk_delete_cleans_up_all_periodic_tasks(self, mock_artwork):
+        """Bulk deleting recordings cleans up all their PeriodicTasks."""
+        mock_artwork.apply_async.return_value = MagicMock()
+
+        future = timezone.now() + timedelta(hours=2)
+        rec_ids = []
+        for i in range(5):
+            rec = Recording.objects.create(
+                channel=self.channel,
+                start_time=future + timedelta(hours=i),
+                end_time=future + timedelta(hours=i + 1),
+            )
+            rec_ids.append(rec.id)
+
+        for rid in rec_ids:
+            self.assertTrue(
+                PeriodicTask.objects.filter(name=f"dvr-recording-{rid}").exists()
+            )
+
+        Recording.objects.filter(channel=self.channel).delete()
+
+        self.assertEqual(
+            PeriodicTask.objects.filter(name__startswith="dvr-recording-").count(), 0
+        )
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_pre_save_revokes_on_time_change(self, mock_artwork):
+        """Changing a recording's start_time revokes the old task and creates a new one."""
+        mock_artwork.apply_async.return_value = MagicMock()
+
+        future = timezone.now() + timedelta(hours=2)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=future,
+            end_time=future + timedelta(hours=1),
+        )
+
+        rec.refresh_from_db()
+        old_task_name = rec.task_id
+        self.assertTrue(PeriodicTask.objects.filter(name=old_task_name).exists())
+
+        # Change the start time — pre_save clears task_id, post_save reschedules
+        new_future = future + timedelta(hours=3)
+        rec.start_time = new_future
+        rec.end_time = new_future + timedelta(hours=1)
+        rec.save()
+
+        rec.refresh_from_db()
+        # Old PeriodicTask should be deleted; new one should exist
+        self.assertIsNotNone(rec.task_id)
+        self.assertTrue(
+            PeriodicTask.objects.filter(name=f"dvr-recording-{rec.id}").exists()
+        )
+
+
+class IdempotencyGuardTests(TestCase):
+    """Tests for the idempotency guard in run_recording()."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(channel_number=1, name="Test Channel")
+
+    @patch("apps.channels.tasks.get_channel_layer")
+    def test_skips_if_already_recording(self, mock_layer):
+        """run_recording returns early if status is already 'recording'."""
+        now = timezone.now()
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=now,
+            end_time=now + timedelta(hours=1),
+            custom_properties={"status": "recording", "started_at": str(now)},
+        )
+
+        from apps.channels.tasks import run_recording as run_rec_task
+        result = run_rec_task(rec.id, self.channel.id, str(now), str(now + timedelta(hours=1)))
+
+        self.assertIsNone(result)
+        # get_channel_layer should not have been called (returned before)
+        mock_layer.assert_not_called()
+
+    @patch("apps.channels.tasks.get_channel_layer")
+    def test_skips_if_already_completed(self, mock_layer):
+        """run_recording returns early if status is already 'completed'."""
+        now = timezone.now()
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(hours=2),
+            end_time=now - timedelta(hours=1),
+            custom_properties={"status": "completed"},
+        )
+
+        from apps.channels.tasks import run_recording as run_rec_task
+        result = run_rec_task(rec.id, self.channel.id, str(rec.start_time), str(rec.end_time))
+
+        self.assertIsNone(result)
+        mock_layer.assert_not_called()
+
+    @patch("apps.channels.tasks.get_channel_layer")
+    def test_skips_if_already_stopped(self, mock_layer):
+        """run_recording returns early if status is already 'stopped' (user stopped it early)."""
+        now = timezone.now()
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(hours=1),
+            end_time=now + timedelta(hours=1),
+            custom_properties={"status": "stopped", "stopped_at": str(now)},
+        )
+
+        from apps.channels.tasks import run_recording as run_rec_task
+        result = run_rec_task(rec.id, self.channel.id, str(rec.start_time), str(rec.end_time))
+
+        self.assertIsNone(result)
+        mock_layer.assert_not_called()
+
+
+class ArtworkPrefetchSignalGuardTests(TestCase):
+    """Tests that the post_save signal does not schedule artwork prefetch when
+    the recording is in an active or terminal state."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(channel_number=1, name="Test Channel")
+
+    def tearDown(self):
+        PeriodicTask.objects.filter(name__startswith="dvr-recording-").delete()
+        ClockedSchedule.objects.all().delete()
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_artwork_prefetch_not_scheduled_when_status_recording(self, mock_artwork):
+        """post_save must NOT schedule artwork prefetch when status='recording'
+        to prevent a race that overwrites the running task's status updates."""
+        future = timezone.now() + timedelta(hours=2)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=future,
+            end_time=future + timedelta(hours=1),
+            custom_properties={"status": "recording"},
+        )
+
+        # Simulate a save that run_recording itself might do mid-recording
+        rec.custom_properties = {"status": "recording", "file_path": "/data/recordings/test.mkv"}
+        rec.save(update_fields=["custom_properties"])
+
+        # apply_async was not called for the "recording" save
+        mock_artwork.apply_async.assert_not_called()
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_artwork_prefetch_not_scheduled_when_status_completed(self, mock_artwork):
+        """post_save must NOT schedule artwork prefetch when status='completed'."""
+        future = timezone.now() + timedelta(hours=2)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=future,
+            end_time=future + timedelta(hours=1),
+            custom_properties={"status": "completed"},
+        )
+
+        rec.custom_properties = {"status": "completed"}
+        rec.save(update_fields=["custom_properties"])
+
+        mock_artwork.apply_async.assert_not_called()
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_artwork_prefetch_not_scheduled_when_status_stopped(self, mock_artwork):
+        """post_save must NOT schedule artwork prefetch when status='stopped'."""
+        future = timezone.now() + timedelta(hours=2)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=future,
+            end_time=future + timedelta(hours=1),
+            custom_properties={"status": "stopped"},
+        )
+
+        rec.custom_properties = {"status": "stopped"}
+        rec.save(update_fields=["custom_properties"])
+
+        mock_artwork.apply_async.assert_not_called()
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_artwork_prefetch_scheduled_for_new_upcoming_recording(self, mock_artwork):
+        """post_save SHOULD schedule artwork prefetch for a newly created upcoming recording."""
+        mock_artwork.apply_async.return_value = MagicMock()
+        future = timezone.now() + timedelta(hours=2)
+        Recording.objects.create(
+            channel=self.channel,
+            start_time=future,
+            end_time=future + timedelta(hours=1),
+            custom_properties={},  # no status yet — should trigger prefetch
+        )
+
+        self.assertTrue(mock_artwork.apply_async.called)
+
+
+class DestroyDvrClientIsolationTests(TestCase):
+    """Tests that deleting a recording only stops DVR clients when the
+    recording is actively streaming — never for completed/upcoming recordings
+    that could share a channel with an unrelated in-progress recording."""
+
+    def setUp(self):
+        from django.contrib.auth import get_user_model
+        from rest_framework.test import APIRequestFactory, force_authenticate
+        self.channel = Channel.objects.create(channel_number=1, name="Test Channel")
+        User = get_user_model()
+        self.user = User.objects.create_user(
+            username="dvr_test_admin", password="pass",
+            user_level=User.UserLevel.ADMIN,
+        )
+        self.factory = APIRequestFactory()
+        self.force_authenticate = force_authenticate
+
+    def _delete_recording(self, rec):
+        from apps.channels.api_views import RecordingViewSet
+        request = self.factory.delete(f"/api/channels/recordings/{rec.id}/")
+        self.force_authenticate(request, user=self.user)
+        view = RecordingViewSet.as_view({"delete": "destroy"})
+        return view(request, pk=rec.id)
+
+    @patch("apps.channels.api_views._stop_dvr_clients")
+    def test_destroy_completed_recording_does_not_stop_dvr_clients(self, mock_stop):
+        """Deleting a completed recording must NOT call _stop_dvr_clients."""
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=timezone.now() - timedelta(hours=2),
+            end_time=timezone.now() - timedelta(hours=1),
+            custom_properties={"status": "completed", "file_path": "/data/recordings/test.mkv"},
+        )
+        self._delete_recording(rec)
+        mock_stop.assert_not_called()
+
+    @patch("apps.channels.api_views._stop_dvr_clients")
+    def test_destroy_upcoming_recording_does_not_stop_dvr_clients(self, mock_stop):
+        """Deleting an upcoming (scheduled) recording must NOT call _stop_dvr_clients."""
+        future = timezone.now() + timedelta(hours=2)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=future,
+            end_time=future + timedelta(hours=1),
+            custom_properties={},
+        )
+        self._delete_recording(rec)
+        mock_stop.assert_not_called()
+
+    @patch("apps.channels.api_views._stop_dvr_clients")
+    def test_destroy_active_recording_does_stop_dvr_clients(self, mock_stop):
+        """Deleting an in-progress recording MUST call _stop_dvr_clients."""
+        mock_stop.return_value = 1
+        now = timezone.now()
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(minutes=5),
+            end_time=now + timedelta(hours=1),
+            custom_properties={"status": "recording"},
+        )
+        self._delete_recording(rec)
+        mock_stop.assert_called_once_with(str(self.channel.uuid))
+
+
+class PeriodicTaskCleanupOnExecutionTests(TestCase):
+    """Tests for PeriodicTask cleanup when run_recording starts."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(channel_number=1, name="Test Channel")
+
+    def tearDown(self):
+        PeriodicTask.objects.filter(name__startswith="dvr-recording-").delete()
+        ClockedSchedule.objects.all().delete()
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    @patch("apps.channels.tasks.get_channel_layer")
+    def test_periodic_task_cleaned_up_on_execution(self, mock_layer, mock_artwork):
+        """When run_recording executes, it deletes its own PeriodicTask."""
+        mock_layer.return_value = MagicMock()
+        mock_artwork.apply_async.return_value = MagicMock()
+
+        future = timezone.now() + timedelta(hours=2)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=future,
+            end_time=future + timedelta(hours=1),
+            custom_properties={},
+        )
+
+        # post_save signal should have created the PeriodicTask
+        task_name = f"dvr-recording-{rec.id}"
+        self.assertTrue(PeriodicTask.objects.filter(name=task_name).exists())
+        pt = PeriodicTask.objects.get(name=task_name)
+        clocked_id = pt.clocked_id
+
+        from apps.channels.tasks import run_recording as run_rec_task
+        # This will proceed past guards, clean up the PeriodicTask, then
+        # eventually fail on the actual stream connection (expected)
+        try:
+            run_rec_task(rec.id, self.channel.id, str(future), str(future + timedelta(hours=1)))
+        except Exception:
+            pass
+
+        self.assertFalse(PeriodicTask.objects.filter(name=task_name).exists())
+        self.assertFalse(ClockedSchedule.objects.filter(id=clocked_id).exists())
diff --git a/frontend/src/WebSocket.jsx b/frontend/src/WebSocket.jsx
index c18a7635..f595b747 100644
--- a/frontend/src/WebSocket.jsx
+++ b/frontend/src/WebSocket.jsx
@@ -547,6 +547,34 @@ export const WebsocketProvider = ({ children }) => {
               }
               break;
 
+            case 'recording_stopped':
+              notifications.show({
+                title: 'Recording stopped',
+                message: `Recording stopped early for ${parsedEvent.data.channel || 'channel'}. Partial content has been saved.`,
+                color: 'yellow',
+              });
+              try {
+                await useChannelsStore.getState().fetchRecordings();
+              } catch (e) {
+                console.warn('Failed to refresh recordings on stop:', e);
+              }
+              break;
+
+            case 'recording_cancelled':
+              notifications.show({
+                title: parsedEvent.data.was_in_progress ? 'Recording cancelled' : 'Recording deleted',
+                message: parsedEvent.data.was_in_progress
+                  ? 'Recording cancelled and content removed.'
+                  : 'Recording deleted.',
+                color: 'red',
+              });
+              try {
+                await useChannelsStore.getState().fetchRecordings();
+              } catch (e) {
+                console.warn('Failed to refresh recordings on cancel:', e);
+              }
+              break;
+
             case 'epg_fetch_error':
               notifications.show({
                 title: 'EPG Source Error',
diff --git a/frontend/src/api.js b/frontend/src/api.js
index 10a836ae..4fbf1018 100644
--- a/frontend/src/api.js
+++ b/frontend/src/api.js
@@ -2662,6 +2662,17 @@ export default class API {
     }
   }
 
+  static async stopRecording(id) {
+    try {
+      await request(`${host}/api/channels/recordings/${id}/stop/`, {
+        method: 'POST',
+      });
+    } catch (e) {
+      errorNotification(`Failed to stop recording ${id}`, e);
+      throw e;
+    }
+  }
+
   static async runComskip(recordingId) {
     try {
       const resp = await request(
diff --git a/frontend/src/components/cards/RecordingCard.jsx b/frontend/src/components/cards/RecordingCard.jsx
index e1dbf020..2f428768 100644
--- a/frontend/src/components/cards/RecordingCard.jsx
+++ b/frontend/src/components/cards/RecordingCard.jsx
@@ -22,7 +22,7 @@ import {
   Text,
   Tooltip,
 } from '@mantine/core';
-import { AlertTriangle, SquareX } from 'lucide-react';
+import { AlertTriangle, Square, SquareX } from 'lucide-react';
 import RecordingSynopsis from '../RecordingSynopsis';
 import {
   deleteRecordingById,
@@ -34,6 +34,7 @@ import {
   getShowVideoUrl,
   removeRecording,
   runComSkip,
+  stopRecordingById,
 } from './../../utils/cards/RecordingCardUtils.js';
 
 const RecordingCard = ({
@@ -73,7 +74,7 @@ const RecordingCard = ({
   const status = customProps.status;
   const isTimeActive = now.isAfter(start) && now.isBefore(end);
   const isInterrupted = status === 'interrupted';
-  const isInProgress = isTimeActive; // Show as recording by time, regardless of status glitches
+  const isInProgress = isTimeActive && !isInterrupted && status !== 'completed' && status !== 'stopped';
   const isUpcoming = now.isBefore(start);
   const isSeriesGroup = Boolean(
     recording._group_count && recording._group_count > 1
@@ -117,10 +118,18 @@ const RecordingCard = ({
     }
   };
 
-  // Cancel handling for series groups
+  // Stop / Cancel / Delete state and handlers
   const [cancelOpen, setCancelOpen] = React.useState(false);
+  const [stopConfirmOpen, setStopConfirmOpen] = React.useState(false);
+  const [deleteConfirmOpen, setDeleteConfirmOpen] = React.useState(false);
   const [busy, setBusy] = React.useState(false);
-  const handleCancelClick = (e) => {
+
+  const handleStopClick = (e) => {
+    e.stopPropagation();
+    setStopConfirmOpen(true);
+  };
+
+  const handleDeleteClick = (e) => {
     e.stopPropagation();
     if (isRecurringRule) {
       onOpenRecurring?.(recording, true);
@@ -129,7 +138,30 @@ const RecordingCard = ({
     if (isSeriesGroup) {
       setCancelOpen(true);
     } else {
+      setDeleteConfirmOpen(true);
+    }
+  };
+
+  const confirmStop = async () => {
+    try {
+      setBusy(true);
+      await stopRecordingById(recording.id);
+    } catch (error) {
+      console.error('Failed to stop recording', error);
+    } finally {
+      setBusy(false);
+      setStopConfirmOpen(false);
+      try { await fetchRecordings(); } catch {}
+    }
+  };
+
+  const confirmDelete = async () => {
+    try {
+      setBusy(true);
       removeRecording(recording.id);
+    } finally {
+      setBusy(false);
+      setDeleteConfirmOpen(false);
     }
   };
 
@@ -279,18 +311,30 @@ const RecordingCard = ({
           </Stack>
         </Group>
 
-        <Center>
-          <Tooltip label={isUpcoming || isInProgress ? 'Cancel' : 'Delete'}>
+        <Group gap={4}>
+          {isInProgress && (
+            <Tooltip label="Stop recording (keep partial content)">
+              <ActionIcon
+                variant="transparent"
+                color="yellow.6"
+                onMouseDown={(e) => e.stopPropagation()}
+                onClick={handleStopClick}
+              >
+                <Square size="20" fill="currentColor" />
+              </ActionIcon>
+            </Tooltip>
+          )}
+          <Tooltip label={isInProgress ? 'Cancel & delete' : isUpcoming ? 'Cancel' : 'Delete'}>
             <ActionIcon
               variant="transparent"
               color="red.9"
               onMouseDown={(e) => e.stopPropagation()}
-              onClick={handleCancelClick}
+              onClick={handleDeleteClick}
             >
               <SquareX size="20" />
             </ActionIcon>
           </Tooltip>
-        </Center>
+        </Group>
       </Flex>
 
       <Flex gap="sm" align="center">
@@ -378,11 +422,74 @@ const RecordingCard = ({
       )}
     </Card>
   );
-  if (!isSeriesGroup) return MainCard;
+
+  // Confirmation modals for stop and cancel/delete
+  const ConfirmModals = (
+    <>
+      <Modal
+        opened={stopConfirmOpen}
+        onClose={() => setStopConfirmOpen(false)}
+        title="Stop Recording"
+        centered
+        size="md"
+        zIndex={9999}
+      >
+        <Stack gap="sm">
+          <Text>
+            The recording will be stopped early. The portion already recorded
+            will be saved and available for playback.
+          </Text>
+          <Group justify="flex-end">
+            <Button variant="default" onClick={() => setStopConfirmOpen(false)} disabled={busy}>
+              Go Back
+            </Button>
+            <Button color="yellow" loading={busy} onClick={confirmStop}>
+              Stop Recording
+            </Button>
+          </Group>
+        </Stack>
+      </Modal>
+
+      <Modal
+        opened={deleteConfirmOpen}
+        onClose={() => setDeleteConfirmOpen(false)}
+        title={isInProgress ? 'Cancel Recording' : isUpcoming ? 'Cancel Recording' : 'Delete Recording'}
+        centered
+        size="md"
+        zIndex={9999}
+      >
+        <Stack gap="sm">
+          <Text>
+            {isInProgress
+              ? 'The recording will be cancelled and all recorded content will be permanently deleted.'
+              : isUpcoming
+                ? 'This scheduled recording will be cancelled.'
+                : 'This recording and all associated files will be permanently deleted.'}
+          </Text>
+          <Group justify="flex-end">
+            <Button variant="default" onClick={() => setDeleteConfirmOpen(false)} disabled={busy}>
+              Go Back
+            </Button>
+            <Button color="red" loading={busy} onClick={confirmDelete}>
+              {isInProgress ? 'Cancel & Delete' : isUpcoming ? 'Cancel' : 'Delete'}
+            </Button>
+          </Group>
+        </Stack>
+      </Modal>
+    </>
+  );
+
+  if (!isSeriesGroup) return (
+    <>
+      {ConfirmModals}
+      {MainCard}
+    </>
+  );
 
   // Stacked look for series groups: render two shadow layers behind the main card
   return (
     <Box style={{ position: 'relative' }}>
+      {ConfirmModals}
       <Modal
         opened={cancelOpen}
         onClose={() => setCancelOpen(false)}
diff --git a/frontend/src/utils/cards/RecordingCardUtils.js b/frontend/src/utils/cards/RecordingCardUtils.js
index 6232064b..60dd2cd5 100644
--- a/frontend/src/utils/cards/RecordingCardUtils.js
+++ b/frontend/src/utils/cards/RecordingCardUtils.js
@@ -51,6 +51,10 @@ export const deleteRecordingById = async (recordingId) => {
   await API.deleteRecording(recordingId);
 };
 
+export const stopRecordingById = async (recordingId) => {
+  await API.stopRecording(recordingId);
+};
+
 export const deleteSeriesAndRule = async (seriesInfo) => {
   const { tvg_id, title } = seriesInfo;
   if (tvg_id) {
diff --git a/frontend/src/utils/pages/DVRUtils.js b/frontend/src/utils/pages/DVRUtils.js
index bcb200b2..fb80dd15 100644
--- a/frontend/src/utils/pages/DVRUtils.js
+++ b/frontend/src/utils/pages/DVRUtils.js
@@ -32,7 +32,7 @@ const dedupeById = (list, toUserTime, completed, now, inProgress, upcoming) => {
     const e = toUserTime(rec.end_time);
     const status = rec.custom_properties?.status;
 
-    if (status === 'interrupted' || status === 'completed') {
+    if (status === 'interrupted' || status === 'completed' || status === 'stopped') {
       completed.push(rec);
     } else {
       if (now.isAfter(s) && now.isBefore(e)) inProgress.push(rec);

From f28530a47b23ab2f9a6babbf3949f999cca997c4 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Sun, 1 Mar 2026 17:01:16 -0600
Subject: [PATCH 36/58] fix(dvr): add recording extend, fix cross-recording
 interference, artwork race, and keepalive gaps
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

File changes:

Backend

apps/channels/api_views.py — Added extend() action that uses queryset .update() to bypass signals, letting the running task pick up the new end_time via polling. Moved WS recording_stopped notification to synchronous (before HTTP response). Refactored destroy() to delete DB row first, send WS event immediately, defer cleanup to background thread.

apps/channels/signals.py — Added pre_save guard to skip task revocation when status is "recording". Added post_save reentrancy guard to skip artwork prefetch for internal field-only saves (custom_properties, task_id, end_time).

apps/channels/tasks.py — Added 2-second polling loop in streaming that checks for stop, deletion, and extended end_time. Added refresh_from_db() + merge before metadata save to prevent overwriting concurrent artwork prefetch. Replaced inline PeriodicTask cleanup with revoke_task()/_dvr_task_name(). Consolidated redundant DB queries in error path. Fixed TOCTOU os.path.exists + os.remove. Preserved "stopped" status in final-status logic instead of overwriting with "completed".

core/utils.py — Made send_websocket_update gevent-safe by detecting monkey-patching and offloading async_to_sync to gevent's native threadpool, fixing the event loop deadlock that caused cross-recording interference.

apps/proxy/ts_proxy/client_manager.py — Moved _trigger_stats_update to a background thread so remove_client() returns immediately.

apps/proxy/ts_proxy/stream_generator.py — Added Redis-based health check for non-owner workers in _should_send_keepalive(), so keepalives fire correctly even when stream_manager is None.

Frontend

frontend/src/WebSocket.jsx — Added 400ms debounced scheduleRecordingFetch() replacing all inline fetchRecordings() calls. Added recording_extended and recording_updated event handlers. recording_cancelled now does surgical removeRecording() by ID when available.

frontend/src/api.js — Added extendRecording(id, extraMinutes).

frontend/src/components/cards/RecordingCard.jsx — Added Extend menu (+15m/+30m/+1h) for in-progress recordings. Widened Comskip button to stopped/interrupted statuses.

frontend/src/components/forms/ProgramRecordingModal.jsx — Removed manual fetchRecordings() calls, relying on WS-driven debounced refresh.

frontend/src/components/forms/RecordingDetailsModal.jsx — Added stopped to playable statuses. Widened Comskip button. Removed manual fetchRecordings().

frontend/src/components/forms/RecurringRuleModal.jsx — Removed all manual fetchRecordings() calls, relying on WS events.

frontend/src/pages/Guide.jsx — Removed isLoading subscription that caused loading flash on debounced refresh. Removed manual fetchRecordings() after saving a series rule.

frontend/src/store/channels.jsx — Stripped isLoading/error state from fetchRecordings to prevent full-page loading indicators. Added no-op guards in removeRecording.

frontend/src/utils/cards/RecordingCardUtils.js — Added extendRecordingById wrapper.

Tests

apps/channels/tests/test_recording_extend.py (new) — 14 tests: extend endpoint validation, stacked extensions, signal bypass, pre_save guard behavior.

apps/channels/tests/test_recording_stop_cancel.py (new) — 18 tests: stop endpoint, DVR client teardown, cancel was_in_progress flag, run_recording race guards, signal reentrancy guards.

apps/channels/tests/test_ts_proxy_keepalive.py (new) — 21 tests: keepalive owner/non-owner logic, stats update error handling, client removal non-blocking, timing invariants.

apps/channels/tests/test_recording_scheduling.py — Updated mock assertion to match new _stop_dvr_clients call signature.

Summary:

Adds an Extend action for in-progress recordings that bypasses Django signals and lets the running task dynamically pick up the new end_time via its 2-second DB polling loop. Fixes cross-recording interference caused by send_websocket_update deadlocking the gevent event loop. Fixes an artwork race condition where run_recording overwrote concurrent prefetch data. Adds Redis-based keepalive health checks for non-owner workers to prevent DVR read timeouts during source transitions. Replaces manual fetchRecordings() calls throughout the frontend with a debounced WS-driven refresh. 53 new tests.
---
 apps/channels/api_views.py                    | 148 ++++++--
 apps/channels/signals.py                      |  19 +-
 apps/channels/tasks.py                        | 162 ++++++---
 apps/channels/tests/test_recording_extend.py  | 235 ++++++++++++
 .../tests/test_recording_scheduling.py        |   7 +-
 .../tests/test_recording_stop_cancel.py       | 341 ++++++++++++++++++
 .../channels/tests/test_ts_proxy_keepalive.py | 324 +++++++++++++++++
 apps/proxy/ts_proxy/client_manager.py         |  14 +-
 apps/proxy/ts_proxy/stream_generator.py       |  25 +-
 core/utils.py                                 |  46 +--
 frontend/src/WebSocket.jsx                    |  71 ++--
 frontend/src/api.js                           |  14 +
 .../src/components/cards/RecordingCard.jsx    |  55 ++-
 .../forms/ProgramRecordingModal.jsx           |  13 +-
 .../forms/RecordingDetailsModal.jsx           |  11 +-
 .../components/forms/RecurringRuleModal.jsx   |   9 +-
 frontend/src/pages/Guide.jsx                  |  12 +-
 frontend/src/store/channels.jsx               |  12 +-
 .../src/utils/cards/RecordingCardUtils.js     |   4 +
 19 files changed, 1328 insertions(+), 194 deletions(-)
 create mode 100644 apps/channels/tests/test_recording_extend.py
 create mode 100644 apps/channels/tests/test_recording_stop_cancel.py
 create mode 100644 apps/channels/tests/test_ts_proxy_keepalive.py

diff --git a/apps/channels/api_views.py b/apps/channels/api_views.py
index bd94b552..b2bc0e14 100644
--- a/apps/channels/api_views.py
+++ b/apps/channels/api_views.py
@@ -11,7 +11,8 @@ from django.shortcuts import get_object_or_404, get_list_or_404
 from django.db import transaction
 from django.db.models import Count, F
 from django.db.models import Q
-import os, json, requests, logging, mimetypes
+import os, json, requests, logging, mimetypes, threading
+from datetime import timedelta
 from django.utils.http import http_date
 from urllib.parse import unquote
 from apps.accounts.permissions import (
@@ -2404,15 +2405,27 @@ class RecordingViewSet(viewsets.ModelViewSet):
         instance.custom_properties = cp
         instance.save(update_fields=["custom_properties"])
 
-        # Everything else (DVR client teardown, task revocation, WebSocket notification)
-        # is moved to a daemon thread so the API returns immediately.  Callers were
-        # seeing 5-15 s delays because _stop_dvr_clients / revoke_task / async_to_sync
-        # have occasional slow paths (Redis timeouts, Celery control broadcasts, etc.).
+        # Send the WebSocket notification before returning the response.
+        # send_websocket_update is gevent-safe (offloads async_to_sync to a
+        # real OS thread when monkey-patching is active).
         channel_uuid = str(instance.channel.uuid)
         recording_id = instance.id
         task_id = instance.task_id
         channel_name = instance.channel.name
 
+        try:
+            from core.utils import send_websocket_update
+            send_websocket_update('updates', 'update', {
+                "success": True,
+                "type": "recording_stopped",
+                "channel": channel_name,
+            })
+        except Exception:
+            pass
+
+        # DVR client teardown and task revocation are deferred to a daemon thread
+        # because they have occasional slow paths (Redis timeouts, Celery control
+        # broadcasts) that would otherwise add 5-15 s to the HTTP response time.
         def _background_stop():
             try:
                 stopped = _stop_dvr_clients(channel_uuid, recording_id=recording_id)
@@ -2429,16 +2442,6 @@ class RecordingViewSet(viewsets.ModelViewSet):
             except Exception as e:
                 logger.debug(f"Unable to revoke task for stopped recording: {e}")
 
-            try:
-                from core.utils import send_websocket_update
-                send_websocket_update('updates', 'update', {
-                    "success": True,
-                    "type": "recording_stopped",
-                    "channel": channel_name,
-                })
-            except Exception:
-                pass
-
             # Close the DB connection opened by this thread so it isn't leaked
             try:
                 from django.db import connection as _conn
@@ -2446,52 +2449,106 @@ class RecordingViewSet(viewsets.ModelViewSet):
             except Exception:
                 pass
 
-        import threading
         threading.Thread(target=_background_stop, daemon=True).start()
 
         return Response({"success": True, "status": "stopped"})
 
+    @action(detail=True, methods=["post"], url_path="extend")
+    def extend(self, request, pk=None):
+        """Extend an in-progress recording's end_time without interrupting the stream.
+
+        The running task re-reads end_time every ~2 s and adjusts its deadline
+        dynamically.  The pre_save signal skips task revocation while the
+        recording status is 'recording'.
+        """
+        instance = self.get_object()
+        cp = instance.custom_properties or {}
+
+        if cp.get("status") in ("completed", "stopped", "interrupted"):
+            return Response(
+                {"success": False, "error": "Recording has already finished"},
+                status=status.HTTP_400_BAD_REQUEST,
+            )
+
+        try:
+            extra_minutes = int(request.data.get("extra_minutes", 0))
+        except (TypeError, ValueError):
+            extra_minutes = 0
+
+        if extra_minutes <= 0:
+            return Response(
+                {"success": False, "error": "extra_minutes must be a positive integer"},
+                status=status.HTTP_400_BAD_REQUEST,
+            )
+
+        new_end_time = instance.end_time + timedelta(minutes=extra_minutes)
+        # Use queryset .update() to bypass pre_save/post_save signals.
+        # This avoids the pre_save signal revoking the scheduled/running
+        # Celery task.  The running task's 2-second polling loop re-reads
+        # end_time from the DB and extends its deadline dynamically.
+        # If the task hasn't started yet (still in Beat's queue), it will
+        # read the updated end_time from the DB on its first poll cycle.
+        Recording.objects.filter(pk=instance.pk).update(end_time=new_end_time)
+
+        try:
+            from core.utils import send_websocket_update
+            send_websocket_update('updates', 'update', {
+                "success": True,
+                "type": "recording_extended",
+                "recording_id": instance.id,
+                "new_end_time": new_end_time.isoformat(),
+                "extra_minutes": extra_minutes,
+                "channel": instance.channel.name,
+            })
+        except Exception:
+            pass
+
+        return Response({"success": True, "new_end_time": new_end_time.isoformat()})
+
     def destroy(self, request, *args, **kwargs):
         """Delete the Recording and ensure any active DVR client connection is closed.
 
         Also removes the associated file(s) from disk if present.
+
+        Operation order matters for correctness:
+          1. Delete the DB record first — run_recording's cancellation guard
+             (Recording.objects.filter(id=...).exists()) will now return False,
+             preventing it from saving 'interrupted' status or sending
+             recording_ended after the stream is torn down.
+          2. Send recording_cancelled WebSocket immediately so the frontend
+             removes the card without waiting for the slow DVR client teardown.
+          3. Spawn a background thread to stop the DVR client and delete files.
+             This mirrors the stop() endpoint's approach and avoids the 5-15 s
+             delay that _stop_dvr_clients() can introduce.
         """
         instance = self.get_object()
+        recording_id = instance.pk
         channel_name = instance.channel.name
 
-        # Capture paths before deletion
+        # Capture state before the DB row is deleted
         cp = instance.custom_properties or {}
         rec_status = cp.get("status", "")
-
-        # Only stop the DVR client if this recording is actively streaming.
-        # Stopping for completed/upcoming recordings would kill an unrelated
-        # in-progress recording on the same channel.
-        if rec_status == "recording":
-            try:
-                channel_uuid = str(instance.channel.uuid)
-                stopped = _stop_dvr_clients(channel_uuid)
-                if stopped:
-                    logger.info(f"Stopped {stopped} DVR client(s) for channel {channel_uuid} due to recording cancellation")
-            except Exception as e:
-                logger.debug(f"Unable to stop DVR clients for cancelled recording: {e}")
         file_path = cp.get("file_path")
         temp_ts_path = cp.get("_temp_file_path")
+        channel_uuid = str(instance.channel.uuid)
 
-        # Perform DB delete first, then try to remove files
+        # 1. Delete the DB record (also fires post_delete → revoke_task_on_delete)
         response = super().destroy(request, *args, **kwargs)
 
-        # Notify frontends
+        # 2. Notify frontends immediately
         try:
             from core.utils import send_websocket_update
             send_websocket_update('updates', 'update', {
                 "success": True,
                 "type": "recording_cancelled",
+                "recording_id": recording_id,
                 "channel": channel_name,
                 "was_in_progress": rec_status == "recording",
             })
         except Exception:
             pass
 
+        # 3. Defer slow teardown to a background thread
         library_dir = '/data'
         allowed_roots = ['/data/', library_dir.rstrip('/') + '/']
 
@@ -2505,8 +2562,33 @@ class RecordingViewSet(viewsets.ModelViewSet):
             except Exception as ex:
                 logger.warning(f"Failed to delete recording artifact {path}: {ex}")
 
-        _safe_remove(file_path)
-        _safe_remove(temp_ts_path)
+        def _background_cancel():
+            # Only stop the DVR client if the recording was actively streaming.
+            # Stopping for completed/upcoming recordings would kill an unrelated
+            # in-progress recording on the same channel.
+            if rec_status == "recording":
+                try:
+                    stopped = _stop_dvr_clients(channel_uuid, recording_id=recording_id)
+                    if stopped:
+                        logger.info(
+                            f"Stopped {stopped} DVR client(s) for channel {channel_uuid} due to recording cancellation"
+                        )
+                except Exception as e:
+                    logger.debug(f"Unable to stop DVR clients for cancelled recording: {e}")
+
+            # Best-effort file cleanup in case run_recording already exited
+            # before the DB delete.
+            _safe_remove(file_path)
+            _safe_remove(temp_ts_path)
+
+            # Close the DB connection opened by this thread so it isn't leaked
+            try:
+                from django.db import connection as _conn
+                _conn.close()
+            except Exception:
+                pass
+
+        threading.Thread(target=_background_cancel, daemon=True).start()
 
         return response
 
diff --git a/apps/channels/signals.py b/apps/channels/signals.py
index 00b8fb7c..a42b8c04 100644
--- a/apps/channels/signals.py
+++ b/apps/channels/signals.py
@@ -97,7 +97,7 @@ def schedule_recording_task(instance, eta=None):
     The task is stored in the database and dispatched by Celery Beat at the
     scheduled time with no countdown.  This avoids the Redis visibility_timeout
     redelivery bug that caused duplicate recordings when using apply_async
-    with long countdowns.  See https://github.com/Dispatcharr/Dispatcharr/issues/940
+    with long countdowns.
     """
     if eta is None:
         eta = instance.start_time
@@ -136,8 +136,8 @@ def revoke_task(task_id):
     """Cancel a pending recording task.
 
     task_id is normally a PeriodicTask name (e.g. "dvr-recording-42").
-    For backwards compatibility with task_ids stored before this fix
-    (Celery async-result UUIDs), falls back to AsyncResult.revoke().
+    For backwards compatibility with legacy Celery async-result UUIDs,
+    falls back to AsyncResult.revoke().
     """
     if not task_id:
         return
@@ -168,6 +168,12 @@ def revoke_old_task_on_update(sender, instance, **kwargs):
             old.end_time != instance.end_time or
             old.channel_id != instance.channel_id
         ):
+            # Do NOT revoke while the recording is actively streaming.
+            # run_recording re-reads end_time from the DB every ~2 s and extends
+            # its internal deadline dynamically — revoking here would kill the task.
+            old_status = (old.custom_properties or {}).get("status", "")
+            if old_status == "recording":
+                return
             revoke_task(old.task_id)
             instance.task_id = None
     except Recording.DoesNotExist:
@@ -176,6 +182,13 @@ def revoke_old_task_on_update(sender, instance, **kwargs):
 @receiver(post_save, sender=Recording)
 def schedule_task_on_save(sender, instance, created, **kwargs):
     try:
+        # Skip processing for internal field-only saves (metadata updates,
+        # task_id assignment, end_time extensions) to prevent re-entrant
+        # artwork dispatch and redundant recording_updated WS events.
+        update_fields = kwargs.get('update_fields')
+        if not created and update_fields is not None and set(update_fields) <= {'custom_properties', 'task_id', 'end_time'}:
+            return
+
         if not instance.task_id:
             start_time = instance.start_time
 
diff --git a/apps/channels/tasks.py b/apps/channels/tasks.py
index 87abb90d..26df8c78 100755
--- a/apps/channels/tasks.py
+++ b/apps/channels/tasks.py
@@ -1503,11 +1503,8 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     try:
         rec_check = Recording.objects.filter(id=recording_id).only("custom_properties").first()
         if not rec_check:
-            # Recording was deleted before this task got a chance to run (e.g. user
-            # cancelled from the Upcoming list, or destroy() ran concurrently).
             logger.info(
-                f"run_recording called for recording {recording_id} but it no longer exists — "
-                f"skipping (recording was cancelled before task started)."
+                f"run_recording called for recording {recording_id} but it no longer exists — skipping."
             )
             return
         status = (rec_check.custom_properties or {}).get("status", "")
@@ -1520,18 +1517,10 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     except Exception as e:
         logger.debug(f"Idempotency guard check failed (proceeding anyway): {e}")
 
-    # --- Clean up the one-off PeriodicTask that dispatched task ---
+    # --- Clean up the one-off PeriodicTask that dispatched this task ---
     try:
-        from django_celery_beat.models import ClockedSchedule as _CS, PeriodicTask as _PT
-        task_name = f"dvr-recording-{recording_id}"
-        try:
-            pt = _PT.objects.get(name=task_name)
-            old_clocked = pt.clocked
-            pt.delete()
-            if old_clocked and not _PT.objects.filter(clocked=old_clocked).exists():
-                old_clocked.delete()
-        except _PT.DoesNotExist:
-            pass
+        from apps.channels.signals import revoke_task, _dvr_task_name
+        revoke_task(_dvr_task_name(recording_id))
     except Exception as e:
         logger.debug(f"PeriodicTask cleanup failed (non-fatal): {e}")
 
@@ -1541,8 +1530,7 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     end_time = datetime.fromisoformat(end_time_str)
 
     duration_seconds = int((end_time - start_time).total_seconds())
-    # Build output paths from templates
-    # We need program info; will refine after we load Recording cp below
+    # Build output paths from templates (refined after loading Recording cp below)
     filename = None
     final_path = None
     temp_ts_path = None
@@ -1575,6 +1563,16 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     recording_obj = None
     try:
         recording_obj = Recording.objects.get(id=recording_id)
+        # If the stop endpoint already wrote "stopped" before the task started,
+        # honor it instead of overwriting with "recording".
+        _pre_cp = recording_obj.custom_properties or {}
+        if _pre_cp.get("status") == "stopped":
+            logger.info(
+                f"run_recording {recording_id}: 'stopped' found in DB before stream started "
+                f"— task exits without connecting."
+            )
+            return
+
         # Prime custom_properties with file info/status
         cp = recording_obj.custom_properties or {}
         cp.update({
@@ -1817,8 +1815,38 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
         except Exception:
             pass
 
-        recording_obj.custom_properties = cp
+        # Re-read from DB to preserve concurrent changes (e.g., artwork
+        # prefetch may have saved poster/rating info while resolving).
+        recording_obj.refresh_from_db()
+        fresh_cp = recording_obj.custom_properties or {}
+
+        # If the stop endpoint set "stopped" while resolving, honor it.
+        if fresh_cp.get("status") == "stopped":
+            logger.info(
+                f"run_recording {recording_id}: 'stopped' found after metadata "
+                f"prep — task exits without streaming."
+            )
+            return
+
+        # Merge only the keys explicitly set into the fresh copy
+        for key in ("status", "started_at", "file_url", "output_file_url",
+                     "file_name", "file_path", "_temp_file_path",
+                     "poster_logo_id", "poster_url"):
+            if key in cp:
+                fresh_cp[key] = cp[key]
+        recording_obj.custom_properties = fresh_cp
         recording_obj.save(update_fields=["custom_properties"])
+
+        # Notify frontends so the tile picks up poster/metadata immediately
+        try:
+            from core.utils import send_websocket_update
+            send_websocket_update('updates', 'update', {
+                "success": True,
+                "type": "recording_updated",
+                "recording_id": recording_id,
+            })
+        except Exception:
+            pass
     except Exception as e:
         logger.debug(f"Unable to prime Recording metadata: {e}")
     interrupted = False
@@ -1888,7 +1916,12 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
                         elapsed = time.time() - started_at
                         if elapsed > duration_seconds:
                             break
-                        # Continue draining the stream
+                        # Continue draining the stream.
+                        # Periodically poll the DB for a user-initiated stop so
+                        # the task exits promptly if stop_client() was missed
+                        # (e.g., fired before the DVR client was registered in Redis).
+                        _stop_poll_interval = 2.0  # seconds between DB checks
+                        _last_stop_poll = time.time()
                         for chunk2 in response.iter_content(chunk_size=8192):
                             if not chunk2:
                                 continue
@@ -1897,6 +1930,50 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
                             elapsed = time.time() - started_at
                             if elapsed > duration_seconds:
                                 break
+                            # Periodic stopped-status + end_time extension check
+                            _now = time.time()
+                            if _now - _last_stop_poll >= _stop_poll_interval:
+                                _last_stop_poll = _now
+                                try:
+                                    _sc = Recording.objects.filter(
+                                        id=recording_id
+                                    ).only("custom_properties", "end_time").first()
+                                    if _sc is None:
+                                        # Recording was deleted — exit gracefully.
+                                        logger.info(
+                                            f"DVR recording {recording_id}: "
+                                            f"deleted — exiting stream loop"
+                                        )
+                                        interrupted = False
+                                        break
+                                    if (_sc.custom_properties or {}).get("status") == "stopped":
+                                        logger.info(
+                                            f"DVR recording {recording_id}: stop requested "
+                                            f"— exiting stream loop early"
+                                        )
+                                        break
+                                    # Check for extended end_time
+                                    try:
+                                        new_end = _sc.end_time
+                                        if new_end is not None:
+                                            from django.utils import timezone as _tz
+                                            if _tz.is_naive(new_end):
+                                                new_end = _tz.make_aware(new_end)
+                                            _ref = start_time
+                                            if _tz.is_naive(_ref):
+                                                _ref = _tz.make_aware(_ref)
+                                            new_duration = int((new_end - _ref).total_seconds())
+                                            if new_duration > duration_seconds:
+                                                logger.info(
+                                                    f"run_recording {recording_id}: "
+                                                    f"end_time extended to {new_end}, "
+                                                    f"new duration {new_duration}s"
+                                                )
+                                                duration_seconds = new_duration
+                                    except Exception:
+                                        pass
+                                except Exception:
+                                    pass
                         break  # exit outer for-loop once we switched to full drain
 
                 # If we wrote any bytes, treat as success and stop trying candidates
@@ -1933,16 +2010,14 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
                     interrupted = True
                     interrupted_reason = f"stream_interrupted: {e}"
                 break
-            # Also bail if the Recording was deleted (user cancelled before data arrived)
-            rec_exists = Recording.objects.filter(id=recording_id).exists()
-            if not rec_exists:
-                interrupted = True
-                interrupted_reason = "recording_deleted_during_connect"
-                break
-            # Check for "stopped" status even before data arrived
+            # Check if recording was deleted or stopped before data arrived
             try:
                 rec_check = Recording.objects.filter(id=recording_id).only("custom_properties").first()
-                if rec_check and (rec_check.custom_properties or {}).get("status") == "stopped":
+                if rec_check is None:
+                    interrupted = True
+                    interrupted_reason = "recording_deleted_during_connect"
+                    break
+                if (rec_check.custom_properties or {}).get("status") == "stopped":
                     interrupted = False
                     break
             except Exception:
@@ -1951,15 +2026,6 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     if chosen_base is None and bytes_written == 0:
         interrupted = True
         interrupted_reason = f"no_stream_data: {last_error or 'all_bases_failed'}"
-    else:
-        # If we ended before reaching planned duration, record reason
-        actual_elapsed = 0
-        try:
-            actual_elapsed = os.path.getsize(temp_ts_path) and (duration_seconds)  # Best effort; we streamed until duration or disconnect above
-        except Exception:
-            pass
-        # We cannot compute accurate elapsed here; fine to leave as is
-        pass
 
     # If no bytes were written at all, check whether this was a deliberate stop or a
     # genuine failure.  The exception handler above already sets interrupted=False when
@@ -2024,10 +2090,13 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
         logger.info(f"Recording {recording_id} was cancelled — skipping remux and metadata.")
         # Clean up any remaining temp or stub output files
         for _cleanup_path in [temp_ts_path, final_path]:
+            if not _cleanup_path:
+                continue
             try:
-                if _cleanup_path and os.path.exists(_cleanup_path):
-                    os.remove(_cleanup_path)
-                    logger.info(f"Cleaned up cancelled recording artifact: {_cleanup_path}")
+                os.remove(_cleanup_path)
+                logger.info(f"Cleaned up cancelled recording artifact: {_cleanup_path}")
+            except FileNotFoundError:
+                pass
             except Exception:
                 pass
         return
@@ -2138,12 +2207,14 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
         cp = recording_obj.custom_properties or {}
         cp["ended_at"] = str(datetime.now())
 
-        # Determine final status.
-        # A deliberate user stop ("stopped" in DB) always wins over any interrupted flag —
-        # this handles races where the stream connection broke milliseconds before or
-        # after the stop endpoint committed its DB save.
+        # Final status priority: stopped > completed > interrupted.
+        # "stopped" is set by the stop endpoint before stream teardown, so
+        # refresh_from_db() above guarantees it is visible here.
         db_status_now = cp.get("status", "")
-        if db_status_now == "stopped" or not interrupted:
+        if db_status_now == "stopped":
+            # Deliberate user stop — preserve; do not overwrite with "completed".
+            cp.pop("interrupted_reason", None)
+        elif not interrupted:
             cp["status"] = "completed"
             cp.pop("interrupted_reason", None)
         else:
@@ -2307,7 +2378,8 @@ def recover_recordings_on_startup():
         for rec in upcoming:
             try:
                 from django_celery_beat.models import PeriodicTask as _PT
-                task_name = f"dvr-recording-{rec.id}"
+                from apps.channels.signals import _dvr_task_name
+                task_name = _dvr_task_name(rec.id)
                 if _PT.objects.filter(name=task_name).exists():
                     if rec.task_id != task_name:
                         rec.task_id = task_name
diff --git a/apps/channels/tests/test_recording_extend.py b/apps/channels/tests/test_recording_extend.py
new file mode 100644
index 00000000..a083b6d3
--- /dev/null
+++ b/apps/channels/tests/test_recording_extend.py
@@ -0,0 +1,235 @@
+"""Tests for the Extend In-Progress Recording feature.
+
+Covers:
+  - extend() API endpoint (happy path and validation)
+  - pre_save signal guard: end_time change must NOT revoke a live recording
+  - pre_save signal guard: end_time change MUST still revoke an upcoming recording
+  - TOCTOU edge cases (extend on a completed/stopped/nonexistent recording)
+"""
+from datetime import timedelta
+from unittest.mock import MagicMock, patch
+
+from django.test import TestCase
+from django.utils import timezone
+from rest_framework.test import APIRequestFactory, force_authenticate
+
+from apps.channels.models import Channel, Recording
+from apps.channels.api_views import RecordingViewSet
+
+
+# ---------------------------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------------------------
+
+def _make_admin():
+    from django.contrib.auth import get_user_model
+    User = get_user_model()
+    u, _ = User.objects.get_or_create(
+        username="extend_test_admin",
+        defaults={"user_level": User.UserLevel.ADMIN},
+    )
+    u.set_password("pass")
+    u.save()
+    return u
+
+
+# ---------------------------------------------------------------------------
+# Extend endpoint tests
+# ---------------------------------------------------------------------------
+
+class ExtendEndpointTests(TestCase):
+    """Tests for POST /api/channels/recordings/{id}/extend/"""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(
+            channel_number=88, name="Extend Test Channel"
+        )
+        self.user = _make_admin()
+        self.factory = APIRequestFactory()
+
+    def _extend(self, rec, extra_minutes):
+        request = self.factory.post(
+            f"/api/channels/recordings/{rec.id}/extend/",
+            {"extra_minutes": extra_minutes},
+            format="json",
+        )
+        force_authenticate(request, user=self.user)
+        view = RecordingViewSet.as_view({"post": "extend"})
+        return view(request, pk=rec.id)
+
+    def _make_rec(self, status="recording"):
+        now = timezone.now()
+        return Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(hours=1),
+            end_time=now + timedelta(hours=1),
+            custom_properties={"status": status},
+        )
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_extend_updates_end_time_in_db(self, _ws):
+        rec = self._make_rec()
+        original_end = rec.end_time
+        response = self._extend(rec, 30)
+        self.assertEqual(response.status_code, 200)
+        self.assertTrue(response.data.get("success"))
+        rec.refresh_from_db()
+        expected = original_end + timedelta(minutes=30)
+        delta = abs((rec.end_time - expected).total_seconds())
+        self.assertLess(delta, 1, "end_time was not extended by the correct amount")
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_extend_stacks_multiple_extensions(self, _ws):
+        """Calling extend() twice adds both increments."""
+        rec = self._make_rec()
+        original_end = rec.end_time
+        self._extend(rec, 15)
+        self._extend(rec, 30)
+        rec.refresh_from_db()
+        expected = original_end + timedelta(minutes=45)
+        delta = abs((rec.end_time - expected).total_seconds())
+        self.assertLess(delta, 1)
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_extend_does_not_clear_task_id(self, _ws):
+        """The running Celery task must survive the DB save."""
+        rec = self._make_rec()
+        rec.task_id = "dvr-recording-999"
+        rec.save(update_fields=["task_id"])
+        self._extend(rec, 30)
+        rec.refresh_from_db()
+        self.assertEqual(rec.task_id, "dvr-recording-999")
+
+    def test_extend_returns_400_if_finished(self):
+        """Cannot extend a completed, stopped, or interrupted recording."""
+        for bad_status in ("completed", "stopped", "interrupted"):
+            with self.subTest(status=bad_status):
+                rec = self._make_rec(status=bad_status)
+                response = self._extend(rec, 30)
+                self.assertEqual(response.status_code, 400)
+                self.assertFalse(response.data.get("success"))
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_extend_succeeds_before_task_sets_status(self, _ws):
+        """Extend must work when status is empty (task hasn't started yet)."""
+        rec = self._make_rec(status="")
+        response = self._extend(rec, 15)
+        self.assertEqual(response.status_code, 200)
+        rec.refresh_from_db()
+        expected = rec.end_time  # already extended
+        self.assertTrue(response.data.get("success"))
+
+    @patch("apps.channels.signals.revoke_task")
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_extend_bypasses_signals_no_revoke(self, _ws, mock_revoke):
+        """Extend uses .update() to bypass pre_save — revoke_task must never fire."""
+        rec = self._make_rec(status="")
+        rec.task_id = "dvr-recording-500"
+        rec.save(update_fields=["task_id"])
+        self._extend(rec, 15)
+        self._extend(rec, 30)
+        mock_revoke.assert_not_called()
+        rec.refresh_from_db()
+        self.assertEqual(rec.task_id, "dvr-recording-500")
+
+    def test_extend_returns_400_for_zero_minutes(self):
+        response = self._extend(self._make_rec(), 0)
+        self.assertEqual(response.status_code, 400)
+
+    def test_extend_returns_400_for_negative_minutes(self):
+        response = self._extend(self._make_rec(), -15)
+        self.assertEqual(response.status_code, 400)
+
+    def test_extend_returns_400_for_non_numeric_minutes(self):
+        rec = self._make_rec()
+        request = self.factory.post(
+            f"/api/channels/recordings/{rec.id}/extend/",
+            {"extra_minutes": "lots"},
+            format="json",
+        )
+        force_authenticate(request, user=self.user)
+        view = RecordingViewSet.as_view({"post": "extend"})
+        response = view(request, pk=rec.id)
+        self.assertEqual(response.status_code, 400)
+
+    def test_extend_returns_404_for_nonexistent_recording(self):
+        request = self.factory.post(
+            "/api/channels/recordings/999999/extend/",
+            {"extra_minutes": 30},
+            format="json",
+        )
+        force_authenticate(request, user=self.user)
+        view = RecordingViewSet.as_view({"post": "extend"})
+        response = view(request, pk=999999)
+        self.assertEqual(response.status_code, 404)
+
+
+# ---------------------------------------------------------------------------
+# pre_save signal guard tests
+# ---------------------------------------------------------------------------
+
+class PreSaveExtendGuardTests(TestCase):
+    """The pre_save signal must NOT revoke a live recording when end_time changes,
+    but MUST still revoke a scheduled (upcoming) recording as before."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(
+            channel_number=77, name="Signal Guard Channel"
+        )
+
+    def _make_rec(self, status="", task_id="dvr-recording-42"):
+        now = timezone.now()
+        return Recording.objects.create(
+            channel=self.channel,
+            start_time=now + timedelta(hours=1),
+            end_time=now + timedelta(hours=2),
+            task_id=task_id,
+            custom_properties={"status": status} if status else {},
+        )
+
+    @patch("apps.channels.signals.revoke_task")
+    def test_end_time_change_does_not_revoke_live_recording(self, mock_revoke):
+        """When status='recording', extending end_time must not call revoke_task."""
+        rec = self._make_rec(status="recording", task_id="dvr-recording-42")
+        rec.end_time = rec.end_time + timedelta(minutes=30)
+        rec.save(update_fields=["end_time"])
+        mock_revoke.assert_not_called()
+
+    @patch("apps.channels.signals.revoke_task")
+    def test_task_id_preserved_after_extend_on_live_recording(self, mock_revoke):
+        """task_id must not be cleared for a live recording's end_time change."""
+        rec = self._make_rec(status="recording", task_id="dvr-recording-42")
+        original_task_id = rec.task_id
+        rec.end_time = rec.end_time + timedelta(minutes=30)
+        rec.save(update_fields=["end_time"])
+        rec.refresh_from_db()
+        self.assertEqual(rec.task_id, original_task_id)
+
+    @patch("apps.channels.signals.revoke_task")
+    def test_end_time_change_still_revokes_upcoming_recording(self, mock_revoke):
+        """The guard must NOT apply to upcoming recordings — existing behavior preserved."""
+        rec = self._make_rec(status="", task_id="dvr-recording-77")
+        rec.end_time = rec.end_time + timedelta(minutes=30)
+        rec.save(update_fields=["end_time"])
+        mock_revoke.assert_called_once_with("dvr-recording-77")
+
+    @patch("apps.channels.signals.revoke_task")
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_pre_save_guard_reads_db_status_not_memory_status(self, _ws, mock_revoke):
+        """pre_save reads status from DB (old object), not from the instance being saved."""
+        now = timezone.now()
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(hours=1),
+            end_time=now + timedelta(hours=1),
+            task_id="dvr-recording-66",
+            custom_properties={"status": "recording"},
+        )
+        # Simulate: DB status changes to 'completed' behind the instance's back
+        Recording.objects.filter(pk=rec.pk).update(
+            custom_properties={"status": "completed"}
+        )
+        rec.end_time = rec.end_time + timedelta(minutes=30)
+        rec.save(update_fields=["end_time"])
+        # revoke_task should be called because DB status is "completed", not "recording"
+        mock_revoke.assert_called_once_with("dvr-recording-66")
diff --git a/apps/channels/tests/test_recording_scheduling.py b/apps/channels/tests/test_recording_scheduling.py
index 56bdbe87..e9daecd7 100644
--- a/apps/channels/tests/test_recording_scheduling.py
+++ b/apps/channels/tests/test_recording_scheduling.py
@@ -1,7 +1,8 @@
 """Tests for DVR recording scheduling with ClockedSchedule.
 
-   Prefers Celery ClockedScheduled over Redis time due to 3600s limitation on scheduled tasks;
-   scheduled tasks beyond 3600s cause Redis time to hand process between workers and duplicate jobs
+Uses ClockedSchedule instead of apply_async with countdown because Redis
+visibility_timeout (default 3600s) causes task redelivery for long countdowns,
+leading to duplicate recordings.
 """
 from datetime import timedelta
 from unittest.mock import patch, MagicMock
@@ -501,7 +502,7 @@ class DestroyDvrClientIsolationTests(TestCase):
             custom_properties={"status": "recording"},
         )
         self._delete_recording(rec)
-        mock_stop.assert_called_once_with(str(self.channel.uuid))
+        mock_stop.assert_called_once_with(str(self.channel.uuid), recording_id=rec.id)
 
 
 class PeriodicTaskCleanupOnExecutionTests(TestCase):
diff --git a/apps/channels/tests/test_recording_stop_cancel.py b/apps/channels/tests/test_recording_stop_cancel.py
new file mode 100644
index 00000000..962160ee
--- /dev/null
+++ b/apps/channels/tests/test_recording_stop_cancel.py
@@ -0,0 +1,341 @@
+"""Tests for the DVR Stop/Cancel feature set.
+
+Covers:
+  - stop() endpoint
+  - destroy() was_in_progress field in recording_cancelled WebSocket event
+  - signals.py update_fields re-entrancy guard
+  - run_recording race guard before status write
+  - _stop_dvr_clients() DVR client isolation
+"""
+import time
+from datetime import timedelta
+from unittest.mock import MagicMock, AsyncMock, patch
+
+from django.test import TestCase
+from django.utils import timezone
+from rest_framework.test import APIRequestFactory, force_authenticate
+
+from apps.channels.models import Channel, Recording
+from apps.channels.api_views import RecordingViewSet, _stop_dvr_clients
+
+
+def _make_admin():
+    from django.contrib.auth import get_user_model
+    User = get_user_model()
+    u, _ = User.objects.get_or_create(
+        username="stop_test_admin",
+        defaults={"user_level": User.UserLevel.ADMIN},
+    )
+    u.set_password("pass")
+    u.save()
+    return u
+
+
+def _async_channel_layer_mock():
+    layer = MagicMock()
+    layer.group_send = AsyncMock()
+    return layer
+
+
+class StopEndpointTests(TestCase):
+    """Tests for POST /api/channels/recordings/{id}/stop/"""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(channel_number=99, name="Stop Test Channel")
+        self.user = _make_admin()
+        self.factory = APIRequestFactory()
+
+    def _stop(self, rec):
+        request = self.factory.post(f"/api/channels/recordings/{rec.id}/stop/")
+        force_authenticate(request, user=self.user)
+        view = RecordingViewSet.as_view({"post": "stop"})
+        return view(request, pk=rec.id)
+
+    def _make_rec(self, status="recording"):
+        now = timezone.now()
+        return Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(hours=1),
+            end_time=now + timedelta(hours=1),
+            custom_properties={"status": status},
+        )
+
+    @patch("core.utils.send_websocket_update")
+    @patch("threading.Thread")
+    def test_stop_writes_status_to_db_before_returning(self, mock_thread, mock_ws):
+        """DB write is synchronous — run_recording polls for this."""
+        mock_thread.return_value.start = MagicMock()
+        rec = self._make_rec()
+        response = self._stop(rec)
+        self.assertEqual(response.status_code, 200)
+        self.assertTrue(response.data.get("success"))
+        rec.refresh_from_db()
+        self.assertEqual(rec.custom_properties.get("status"), "stopped")
+
+    @patch("core.utils.send_websocket_update")
+    @patch("threading.Thread")
+    def test_stop_writes_stopped_at_timestamp(self, mock_thread, mock_ws):
+        mock_thread.return_value.start = MagicMock()
+        rec = self._make_rec()
+        self._stop(rec)
+        rec.refresh_from_db()
+        self.assertIn("stopped_at", rec.custom_properties)
+
+    def test_stop_calls_stop_dvr_clients_in_background(self):
+        rec = self._make_rec()
+        with patch("apps.channels.api_views._stop_dvr_clients", return_value=1) as mock_stop, \
+             patch("core.utils.send_websocket_update"):
+            self._stop(rec)
+            time.sleep(0.5)
+        self.assertTrue(mock_stop.called)
+        args, kwargs = mock_stop.call_args
+        actual_rec_id = kwargs.get("recording_id") or (args[1] if len(args) > 1 else None)
+        self.assertEqual(actual_rec_id, rec.id)
+
+    def test_stop_returns_404_for_nonexistent(self):
+        request = self.factory.post("/api/channels/recordings/99999/stop/")
+        force_authenticate(request, user=self.user)
+        view = RecordingViewSet.as_view({"post": "stop"})
+        self.assertEqual(view(request, pk=99999).status_code, 404)
+
+    @patch("core.utils.send_websocket_update")
+    @patch("threading.Thread")
+    def test_stop_idempotent_on_already_stopped(self, mock_thread, mock_ws):
+        mock_thread.return_value.start = MagicMock()
+        rec = self._make_rec(status="stopped")
+        self.assertEqual(self._stop(rec).status_code, 200)
+
+
+class CancelDestroyWasInProgressTests(TestCase):
+    """was_in_progress field in the recording_cancelled WebSocket event."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(channel_number=98, name="Cancel Test Channel")
+        self.user = _make_admin()
+        self.factory = APIRequestFactory()
+
+    def _delete(self, rec):
+        request = self.factory.delete(f"/api/channels/recordings/{rec.id}/")
+        force_authenticate(request, user=self.user)
+        return RecordingViewSet.as_view({"delete": "destroy"})(request, pk=rec.id)
+
+    @patch("apps.channels.api_views._stop_dvr_clients", return_value=1)
+    @patch("core.utils.send_websocket_update")
+    def test_in_progress_sends_was_in_progress_true(self, mock_ws, _):
+        now = timezone.now()
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(minutes=10),
+            end_time=now + timedelta(hours=1),
+            custom_properties={"status": "recording"},
+        )
+        self._delete(rec)
+        payload = mock_ws.call_args[0][2]
+        self.assertEqual(payload["type"], "recording_cancelled")
+        self.assertTrue(payload["was_in_progress"])
+
+    @patch("core.utils.send_websocket_update")
+    def test_completed_sends_was_in_progress_false(self, mock_ws):
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=timezone.now() - timedelta(hours=2),
+            end_time=timezone.now() - timedelta(hours=1),
+            custom_properties={"status": "completed"},
+        )
+        self._delete(rec)
+        self.assertFalse(mock_ws.call_args[0][2]["was_in_progress"])
+
+
+class SignalUpdateFieldsReentrancyGuardTests(TestCase):
+    """update_fields guard in schedule_task_on_save prevents redundant WS events."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(channel_number=97, name="Signal Guard Channel")
+
+    def _create_upcoming(self):
+        future = timezone.now() + timedelta(hours=2)
+        return Recording.objects.create(
+            channel=self.channel, start_time=future,
+            end_time=future + timedelta(hours=1), custom_properties={},
+        )
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_custom_properties_save_skips_artwork(self, mock_artwork):
+        rec = self._create_upcoming()
+        mock_artwork.reset_mock()
+        rec.custom_properties = {"poster_url": "https://example.com/p.jpg"}
+        rec.save(update_fields=["custom_properties"])
+        mock_artwork.apply_async.assert_not_called()
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_task_id_save_skips_artwork(self, mock_artwork):
+        rec = self._create_upcoming()
+        mock_artwork.reset_mock()
+        rec.task_id = "dvr-recording-999"
+        rec.save(update_fields=["task_id"])
+        mock_artwork.apply_async.assert_not_called()
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_combined_metadata_save_skips_artwork(self, mock_artwork):
+        rec = self._create_upcoming()
+        mock_artwork.reset_mock()
+        rec.task_id = "dvr-recording-1000"
+        rec.custom_properties = {"poster_url": "x"}
+        rec.save(update_fields=["custom_properties", "task_id"])
+        mock_artwork.apply_async.assert_not_called()
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_creation_dispatches_artwork(self, mock_artwork):
+        mock_artwork.apply_async.return_value = MagicMock()
+        self._create_upcoming()
+        self.assertTrue(mock_artwork.apply_async.called)
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_scheduling_field_update_dispatches_artwork(self, mock_artwork):
+        """save(update_fields=['start_time']) is not a metadata save — dispatch runs."""
+        mock_artwork.apply_async.return_value = MagicMock()
+        rec = self._create_upcoming()
+        mock_artwork.reset_mock()
+        future = timezone.now() + timedelta(hours=3)
+        rec.start_time = future
+        rec.end_time = future + timedelta(hours=1)
+        rec.save(update_fields=["start_time", "end_time"])
+        mock_artwork.apply_async.assert_called()
+
+
+class RunRecordingRaceGuardTests(TestCase):
+    """Race guard: stop() fires between idempotency check and status write."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(channel_number=96, name="Race Guard Channel")
+
+    def test_race_guard_exits_when_stopped_at_db_read(self):
+        """If Recording.objects.get() shows 'stopped', the task must exit
+        without writing 'recording' to the DB."""
+        from apps.channels.tasks import run_recording as run_rec
+        now = timezone.now()
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(minutes=1),
+            end_time=now + timedelta(hours=1),
+            custom_properties={},
+        )
+        mock_layer = _async_channel_layer_mock()
+        original_get = Recording.objects.get
+
+        def patched_get(*args, **kwargs):
+            obj = original_get(*args, **kwargs)
+            if kwargs.get("id") == rec.id or (args and args[0] == rec.id):
+                obj.custom_properties = {"status": "stopped"}
+            return obj
+
+        with patch("apps.channels.tasks.get_channel_layer", return_value=mock_layer), \
+             patch("core.utils.log_system_event", side_effect=Exception("skip")), \
+             patch.object(Recording.objects, "get", side_effect=patched_get):
+            result = run_rec(
+                rec.id, self.channel.id, str(rec.start_time), str(rec.end_time),
+            )
+
+        self.assertIsNone(result)
+        rec.refresh_from_db()
+        self.assertNotEqual(
+            rec.custom_properties.get("status"), "recording",
+            "Race guard failed: task overwrote 'stopped' with 'recording'",
+        )
+
+    def test_idempotency_guard_catches_stopped_before_channel_layer(self):
+        """When status='stopped' at the idempotency check, get_channel_layer is never called."""
+        from apps.channels.tasks import run_recording as run_rec
+        now = timezone.now()
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(minutes=5),
+            end_time=now + timedelta(hours=1),
+            custom_properties={"status": "stopped"},
+        )
+        with patch("apps.channels.tasks.get_channel_layer") as mock_get_layer:
+            result = run_rec(
+                rec.id, self.channel.id, str(rec.start_time), str(rec.end_time),
+            )
+        self.assertIsNone(result)
+        mock_get_layer.assert_not_called()
+
+
+class StopDvrClientsTests(TestCase):
+    """_stop_dvr_clients() DVR client isolation."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(channel_number=95, name="DVR Clients Channel")
+        self._redis = "core.utils.RedisClient"
+        self._sc = "apps.proxy.ts_proxy.services.channel_service.ChannelService.stop_client"
+        self._sch = "apps.proxy.ts_proxy.services.channel_service.ChannelService.stop_channel"
+
+    def _mock_redis(self, client_ids, ua_map):
+        r = MagicMock()
+        r.smembers.return_value = {c.encode() for c in client_ids}
+        def hget_side(key, field):
+            ks = key if isinstance(key, str) else key.decode("utf-8", errors="replace")
+            for cid, ua in ua_map.items():
+                if cid in ks:
+                    return ua.encode() if isinstance(ua, str) else ua
+            return b""
+        r.hget.side_effect = hget_side
+        return r
+
+    def test_returns_zero_when_redis_none(self):
+        with patch(self._redis) as rc:
+            rc.get_client.return_value = None
+            self.assertEqual(_stop_dvr_clients(str(self.channel.uuid)), 0)
+
+    def test_stops_only_matching_client_when_recording_id_given(self):
+        r = self._mock_redis(
+            ["client-a", "client-b"],
+            {"client-a": "Dispatcharr-DVR/recording-42",
+             "client-b": "Dispatcharr-DVR/recording-99"},
+        )
+        with patch(self._redis) as rc, patch(self._sc) as sc:
+            rc.get_client.return_value = r
+            result = _stop_dvr_clients(str(self.channel.uuid), recording_id=42)
+        self.assertEqual(result, 1)
+        stopped = [c[0][1] for c in sc.call_args_list]
+        self.assertIn("client-a", stopped)
+        self.assertNotIn("client-b", stopped)
+
+    def test_stops_all_dvr_clients_without_recording_id(self):
+        r = self._mock_redis(
+            ["client-a", "client-b"],
+            {"client-a": "Dispatcharr-DVR/recording-42",
+             "client-b": "Dispatcharr-DVR/recording-99"},
+        )
+        with patch(self._redis) as rc, patch(self._sc) as sc:
+            rc.get_client.return_value = r
+            result = _stop_dvr_clients(str(self.channel.uuid))
+        self.assertEqual(result, 2)
+
+    def test_skips_non_dvr_clients(self):
+        r = self._mock_redis(
+            ["viewer", "dvr-client"],
+            {"viewer": "Mozilla/5.0", "dvr-client": "Dispatcharr-DVR/recording-1"},
+        )
+        with patch(self._redis) as rc, patch(self._sc) as sc:
+            rc.get_client.return_value = r
+            result = _stop_dvr_clients(str(self.channel.uuid))
+        self.assertEqual(result, 1)
+        stopped = [c[0][1] for c in sc.call_args_list]
+        self.assertNotIn("viewer", stopped)
+
+    def test_returns_zero_for_empty_channel(self):
+        r = MagicMock()
+        r.smembers.return_value = set()
+        with patch(self._redis) as rc, patch(self._sc) as sc:
+            rc.get_client.return_value = r
+            self.assertEqual(_stop_dvr_clients(str(self.channel.uuid)), 0)
+        sc.assert_not_called()
+
+    def test_never_calls_stop_channel(self):
+        """Must not stop the whole channel proxy — only individual clients."""
+        r = self._mock_redis(["dvr-1"], {"dvr-1": "Dispatcharr-DVR/recording-1"})
+        with patch(self._redis) as rc, patch(self._sc), patch(self._sch) as sch:
+            rc.get_client.return_value = r
+            _stop_dvr_clients(str(self.channel.uuid))
+        sch.assert_not_called()
diff --git a/apps/channels/tests/test_ts_proxy_keepalive.py b/apps/channels/tests/test_ts_proxy_keepalive.py
new file mode 100644
index 00000000..db388b3d
--- /dev/null
+++ b/apps/channels/tests/test_ts_proxy_keepalive.py
@@ -0,0 +1,324 @@
+"""Tests for ts_proxy keepalive and stats-update behavior.
+
+Covers:
+  - stream_generator._should_send_keepalive() owner vs non-owner worker paths
+  - stream_generator._should_send_keepalive() Redis last_data health check
+  - client_manager._do_stats_update() error handling and WebSocket dispatch
+  - client_manager.remove_client() non-blocking stats update
+  - Keepalive/DVR-timeout timing invariants
+"""
+import threading
+import time
+from unittest.mock import MagicMock, patch
+
+from django.test import TestCase
+
+
+# ---------------------------------------------------------------------------
+# _should_send_keepalive: owner worker path
+# ---------------------------------------------------------------------------
+
+class OwnerWorkerKeepaliveTests(TestCase):
+    """Owner worker has a stream_manager; keepalive logic uses it directly."""
+
+    def _make_generator(self, healthy, at_buffer_head, consecutive_empty):
+        from apps.proxy.ts_proxy.stream_generator import StreamGenerator
+        gen = StreamGenerator.__new__(StreamGenerator)
+        gen.channel_id = "00000000-0000-0000-0000-000000000001"
+        gen.client_id = "test-client"
+
+        buffer = MagicMock()
+        buffer.index = 10 if at_buffer_head else 100
+        gen.local_index = 10
+        gen.buffer = buffer
+
+        stream_manager = MagicMock()
+        stream_manager.healthy = healthy
+        gen.stream_manager = stream_manager
+
+        gen.consecutive_empty = consecutive_empty
+        return gen
+
+    def test_owner_healthy_returns_false(self):
+        """Owner worker, healthy stream -> no keepalive."""
+        gen = self._make_generator(healthy=True, at_buffer_head=True, consecutive_empty=10)
+        self.assertFalse(gen._should_send_keepalive(gen.local_index))
+
+    def test_owner_unhealthy_at_head_returns_true(self):
+        """Owner worker, unhealthy stream, at buffer head -> send keepalive."""
+        gen = self._make_generator(healthy=False, at_buffer_head=True, consecutive_empty=10)
+        self.assertTrue(gen._should_send_keepalive(gen.local_index))
+
+    def test_owner_unhealthy_not_at_head_returns_false(self):
+        """Owner worker, unhealthy stream, but NOT at buffer head -> no keepalive."""
+        gen = self._make_generator(healthy=False, at_buffer_head=False, consecutive_empty=10)
+        self.assertFalse(gen._should_send_keepalive(gen.local_index))
+
+    def test_owner_insufficient_consecutive_empty_returns_false(self):
+        """Owner worker, unhealthy, at head but consecutive_empty < 5 -> no keepalive."""
+        gen = self._make_generator(healthy=False, at_buffer_head=True, consecutive_empty=3)
+        self.assertFalse(gen._should_send_keepalive(gen.local_index))
+
+    def test_owner_exactly_5_consecutive_empty_returns_true(self):
+        """consecutive_empty == 5 is the minimum threshold."""
+        gen = self._make_generator(healthy=False, at_buffer_head=True, consecutive_empty=5)
+        self.assertTrue(gen._should_send_keepalive(gen.local_index))
+
+
+# ---------------------------------------------------------------------------
+# _should_send_keepalive: non-owner worker path
+# ---------------------------------------------------------------------------
+
+class NonOwnerWorkerKeepaliveTests(TestCase):
+    """Non-owner worker has stream_manager=None; health determined from Redis."""
+
+    def _make_generator(self, consecutive_empty=10):
+        from apps.proxy.ts_proxy.stream_generator import StreamGenerator
+        gen = StreamGenerator.__new__(StreamGenerator)
+        gen.channel_id = "00000000-0000-0000-0000-000000000002"
+        gen.client_id = "test-client-nonowner"
+
+        buffer = MagicMock()
+        buffer.index = 10
+        gen.local_index = 10
+        gen.buffer = buffer
+
+        gen.stream_manager = None  # non-owner worker
+        gen.consecutive_empty = consecutive_empty
+        return gen
+
+    def _mock_proxy_server(self, last_data_value):
+        """Return a mock ProxyServer with a redis_client pre-configured."""
+        server = MagicMock()
+        redis_client = MagicMock()
+        server.redis_client = redis_client
+        redis_client.get.return_value = last_data_value
+        return server
+
+    def test_non_owner_fresh_data_returns_false(self):
+        """Non-owner, last_data < 10s ago -> stream healthy -> no keepalive."""
+        gen = self._make_generator()
+        fresh_ts = str(time.time() - 2.0).encode()
+        server = self._mock_proxy_server(fresh_ts)
+
+        with patch("apps.proxy.ts_proxy.stream_generator.ProxyServer") as MockPS:
+            MockPS.get_instance.return_value = server
+            result = gen._should_send_keepalive(gen.local_index)
+
+        self.assertFalse(result, "Fresh data should NOT trigger keepalive")
+
+    def test_non_owner_stale_data_returns_true(self):
+        """Non-owner, last_data >= 10s ago -> stream unhealthy -> send keepalive."""
+        gen = self._make_generator()
+        stale_ts = str(time.time() - 12.0).encode()
+        server = self._mock_proxy_server(stale_ts)
+
+        with patch("apps.proxy.ts_proxy.stream_generator.ProxyServer") as MockPS:
+            MockPS.get_instance.return_value = server
+            result = gen._should_send_keepalive(gen.local_index)
+
+        self.assertTrue(result, "Stale data (12s) should trigger keepalive")
+
+    def test_non_owner_exactly_at_timeout_returns_true(self):
+        """Data age exactly equal to CONNECTION_TIMEOUT (10s) -> send keepalive."""
+        gen = self._make_generator()
+        ts = str(time.time() - 10.0).encode()
+        server = self._mock_proxy_server(ts)
+
+        with patch("apps.proxy.ts_proxy.stream_generator.ProxyServer") as MockPS:
+            MockPS.get_instance.return_value = server
+            result = gen._should_send_keepalive(gen.local_index)
+
+        self.assertTrue(result, "Data at exactly timeout threshold should trigger keepalive")
+
+    def test_non_owner_no_redis_key_returns_true(self):
+        """Non-owner, last_data key missing from Redis -> assume unhealthy."""
+        gen = self._make_generator()
+        server = self._mock_proxy_server(None)
+
+        with patch("apps.proxy.ts_proxy.stream_generator.ProxyServer") as MockPS:
+            MockPS.get_instance.return_value = server
+            result = gen._should_send_keepalive(gen.local_index)
+
+        self.assertTrue(result, "Missing last_data key should trigger keepalive")
+
+    def test_non_owner_redis_client_none_returns_false(self):
+        """Non-owner, redis_client is None (disconnected) -> conservative, no keepalive."""
+        gen = self._make_generator()
+        server = MagicMock()
+        server.redis_client = None
+
+        with patch("apps.proxy.ts_proxy.stream_generator.ProxyServer") as MockPS:
+            MockPS.get_instance.return_value = server
+            result = gen._should_send_keepalive(gen.local_index)
+
+        self.assertFalse(result, "No redis_client -> conservative, no keepalive")
+
+    def test_non_owner_redis_exception_returns_false(self):
+        """Non-owner, Redis raises an exception -> conservative, no keepalive."""
+        gen = self._make_generator()
+
+        with patch("apps.proxy.ts_proxy.stream_generator.ProxyServer") as MockPS:
+            MockPS.get_instance.side_effect = Exception("Redis error")
+            result = gen._should_send_keepalive(gen.local_index)
+
+        self.assertFalse(result, "Redis error -> conservative, no keepalive")
+
+    def test_non_owner_not_at_buffer_head_returns_false(self):
+        """Non-owner, NOT at buffer head -> no keepalive regardless of Redis."""
+        gen = self._make_generator()
+        gen.buffer.index = 100  # far ahead of local_index=10
+        server = self._mock_proxy_server(None)
+
+        with patch("apps.proxy.ts_proxy.stream_generator.ProxyServer") as MockPS:
+            MockPS.get_instance.return_value = server
+            result = gen._should_send_keepalive(gen.local_index)
+
+        self.assertFalse(result)
+
+    def test_non_owner_insufficient_consecutive_empty_returns_false(self):
+        """Non-owner, at head, but consecutive_empty < 5 -> no keepalive."""
+        gen = self._make_generator(consecutive_empty=2)
+        stale_ts = str(time.time() - 30.0).encode()
+        server = self._mock_proxy_server(stale_ts)
+
+        with patch("apps.proxy.ts_proxy.stream_generator.ProxyServer") as MockPS:
+            MockPS.get_instance.return_value = server
+            result = gen._should_send_keepalive(gen.local_index)
+
+        self.assertFalse(result)
+
+
+# ---------------------------------------------------------------------------
+# _do_stats_update: error handling and WebSocket dispatch
+# ---------------------------------------------------------------------------
+
+class DoStatsUpdateTests(TestCase):
+    """_do_stats_update runs the actual Redis scan + WebSocket call."""
+
+    def _make_client_manager(self):
+        from apps.proxy.ts_proxy.client_manager import ClientManager
+        cm = ClientManager.__new__(ClientManager)
+        cm.channel_id = "00000000-0000-0000-0000-000000000004"
+        cm._heartbeat_running = False
+        return cm
+
+    def test_do_stats_update_calls_send_websocket_update(self):
+        """_do_stats_update must call send_websocket_update with channel_stats."""
+        cm = self._make_client_manager()
+
+        mock_redis = MagicMock()
+        mock_redis.scan.return_value = (0, [])
+
+        with patch("apps.proxy.ts_proxy.client_manager.send_websocket_update") as mock_ws, \
+             patch("redis.Redis.from_url", return_value=mock_redis):
+            cm._do_stats_update()
+
+        mock_ws.assert_called_once()
+        event_type = mock_ws.call_args[0][1]
+        self.assertEqual(event_type, "update")
+        payload = mock_ws.call_args[0][2]
+        self.assertEqual(payload["type"], "channel_stats")
+
+    def test_do_stats_update_does_not_raise_on_redis_error(self):
+        """Redis failure must be swallowed (logged), not propagated."""
+        cm = self._make_client_manager()
+
+        with patch("redis.Redis.from_url", side_effect=Exception("Redis down")):
+            try:
+                cm._do_stats_update()
+            except Exception as e:
+                self.fail(f"_do_stats_update raised an exception: {e}")
+
+    def test_do_stats_update_scans_channel_client_keys(self):
+        """Must scan for ts_proxy:channel:*:clients pattern."""
+        cm = self._make_client_manager()
+
+        mock_redis = MagicMock()
+        mock_redis.scan.return_value = (0, [])
+
+        with patch("apps.proxy.ts_proxy.client_manager.send_websocket_update"), \
+             patch("redis.Redis.from_url", return_value=mock_redis):
+            cm._do_stats_update()
+
+        scan_call = mock_redis.scan.call_args
+        self.assertIn("ts_proxy:channel:*:clients", str(scan_call))
+
+
+# ---------------------------------------------------------------------------
+# Integration: remove_client must not block on WebSocket
+# ---------------------------------------------------------------------------
+
+class ClientRemoveIntegrationTests(TestCase):
+    """When remove_client() fires, _trigger_stats_update must not block."""
+
+    def test_remove_client_does_not_block_on_websocket(self):
+        """remove_client() must return quickly even if WebSocket is slow."""
+        from apps.proxy.ts_proxy.client_manager import ClientManager
+
+        cm = ClientManager.__new__(ClientManager)
+        cm.channel_id = "00000000-0000-0000-0000-000000000005"
+        cm._heartbeat_running = False
+        cm.clients = {"test-client-1"}
+        cm.last_heartbeat_time = {"test-client-1": time.time()}
+        cm.last_active_time = time.time()
+        cm.client_set_key = f"ts_proxy:channel:{cm.channel_id}:clients"
+        cm.client_ttl = 60
+        cm.worker_id = "worker-1"
+        cm.proxy_server = MagicMock()
+        cm.proxy_server.am_i_owner.return_value = False
+        cm.lock = threading.Lock()
+
+        mock_redis = MagicMock()
+        mock_redis.hgetall.return_value = {b"ip_address": b"127.0.0.1"}
+        mock_redis.scard.return_value = 1
+        cm.redis_client = mock_redis
+
+        slow_ws_called = threading.Event()
+
+        def slow_websocket(*args, **kwargs):
+            time.sleep(2.0)
+            slow_ws_called.set()
+
+        start = time.time()
+        with patch("apps.proxy.ts_proxy.client_manager.send_websocket_update", side_effect=slow_websocket):
+            cm.remove_client("test-client-1")
+        elapsed = time.time() - start
+
+        self.assertLess(elapsed, 1.0,
+                        f"remove_client() blocked for {elapsed:.2f}s waiting for WebSocket "
+                        f"(should dispatch to background thread and return immediately)")
+
+
+# ---------------------------------------------------------------------------
+# DVR timeout threshold vs keepalive timing
+# ---------------------------------------------------------------------------
+
+class KeepaliveTimingTests(TestCase):
+    """Verify that keepalive threshold gives sufficient margin before DVR timeout."""
+
+    def test_keepalive_threshold_less_than_dvr_timeout(self):
+        """CONNECTION_TIMEOUT (keepalive trigger) must be < DVR read timeout (15s)."""
+        from apps.proxy.config import TSConfig as Config
+        connection_timeout = getattr(Config, "CONNECTION_TIMEOUT", 10)
+        dvr_read_timeout = 15  # hard-coded in run_recording: timeout=(10, 15)
+        self.assertLess(
+            connection_timeout,
+            dvr_read_timeout,
+            f"CONNECTION_TIMEOUT ({connection_timeout}s) must be < DVR timeout ({dvr_read_timeout}s) "
+            f"so keepalives fire before DVR times out",
+        )
+
+    def test_keepalive_interval_is_short(self):
+        """KEEPALIVE_INTERVAL must be short enough to send multiple keepalives in the gap."""
+        from apps.proxy.config import TSConfig as Config
+        interval = getattr(Config, "KEEPALIVE_INTERVAL", 0.5)
+        connection_timeout = getattr(Config, "CONNECTION_TIMEOUT", 10)
+        remaining_window = 15 - connection_timeout
+        self.assertGreater(
+            remaining_window / interval,
+            3,
+            f"KEEPALIVE_INTERVAL ({interval}s) is too long: only "
+            f"{remaining_window/interval:.1f} keepalives would fit in the "
+            f"{remaining_window}s window before DVR timeout",
+        )
diff --git a/apps/proxy/ts_proxy/client_manager.py b/apps/proxy/ts_proxy/client_manager.py
index a361bfa1..3351cf56 100644
--- a/apps/proxy/ts_proxy/client_manager.py
+++ b/apps/proxy/ts_proxy/client_manager.py
@@ -43,14 +43,20 @@ class ClientManager:
         self._registered_clients = set()  # Track already registered client IDs
 
     def _trigger_stats_update(self):
-        """Trigger a channel stats update via WebSocket"""
+        """Trigger a channel stats update via WebSocket in a background thread.
+
+        Offloaded so the caller is not blocked. send_websocket_update is
+        gevent-safe (offloads async_to_sync to a native OS thread).
+        """
+        threading.Thread(target=self._do_stats_update, daemon=True).start()
+
+    def _do_stats_update(self):
+        """Perform the stats update in the background."""
         try:
-            # Import here to avoid potential import issues
             from apps.proxy.ts_proxy.channel_status import ChannelStatus
             import redis
             from django.conf import settings
 
-            # Get all channels from Redis using settings
             redis_url = getattr(settings, 'REDIS_URL', 'redis://localhost:6379/0')
             redis_client = redis.Redis.from_url(redis_url, decode_responses=True)
             all_channels = []
@@ -59,7 +65,6 @@ class ClientManager:
             while True:
                 cursor, keys = redis_client.scan(cursor, match="ts_proxy:channel:*:clients", count=100)
                 for key in keys:
-                    # Extract channel ID from key
                     parts = key.split(':')
                     if len(parts) >= 4:
                         ch_id = parts[2]
@@ -70,7 +75,6 @@ class ClientManager:
                 if cursor == 0:
                     break
 
-            # Send WebSocket update using existing infrastructure
             send_websocket_update(
                 "updates",
                 "update",
diff --git a/apps/proxy/ts_proxy/stream_generator.py b/apps/proxy/ts_proxy/stream_generator.py
index 50404f1d..8ba38fc0 100644
--- a/apps/proxy/ts_proxy/stream_generator.py
+++ b/apps/proxy/ts_proxy/stream_generator.py
@@ -381,10 +381,29 @@ class StreamGenerator:
         """Determine if a keepalive packet should be sent."""
         # Check if we're caught up to buffer head
         at_buffer_head = local_index >= self.buffer.index
+        if not at_buffer_head or self.consecutive_empty < 5:
+            return False
 
-        # If we're at buffer head and no data is coming, send keepalive
-        stream_healthy = self.stream_manager.healthy if self.stream_manager else True
-        return at_buffer_head and not stream_healthy and self.consecutive_empty >= 5
+        if self.stream_manager is not None:
+            # Owner worker: use the in-memory health flag directly.
+            return not self.stream_manager.healthy
+        else:
+            # Non-owner worker: stream_manager only exists in the owner process.
+            # Approximate health from the Redis last_data timestamp; if stale
+            # beyond CONNECTION_TIMEOUT, send keepalives to prevent DVR timeout.
+            try:
+                proxy_server = ProxyServer.get_instance()
+                if proxy_server.redis_client:
+                    raw = proxy_server.redis_client.get(RedisKeys.last_data(self.channel_id))
+                    if raw:
+                        age = time.time() - float(raw)
+                        timeout_threshold = getattr(Config, 'CONNECTION_TIMEOUT', 10)
+                        return age >= timeout_threshold
+                    # No timestamp in Redis → key missing or expired → unhealthy
+                    return True
+            except Exception:
+                pass
+            return False
 
     def _is_ghost_client(self, local_index):
         """Check if this appears to be a ghost client (stuck but buffer advancing)."""
diff --git a/core/utils.py b/core/utils.py
index f0e02bb4..ef5aa702 100644
--- a/core/utils.py
+++ b/core/utils.py
@@ -294,30 +294,34 @@ def send_websocket_update(group_name, event_type, data, collect_garbage=False):
     """
     Standardized function to send WebSocket updates with proper memory management.
 
-    Args:
-        group_name: The WebSocket group to send to (e.g. 'updates')
-        event_type: The type of message (e.g. 'update')
-        data: The data to send
-        collect_garbage: Whether to force garbage collection after sending
+    In uWSGI + gevent deployments, async_to_sync creates an asyncio event loop
+    whose native EpollSelector blocks the entire OS thread, freezing all gevent
+    greenlets in the worker.  To avoid this, the actual send is offloaded to a
+    real OS thread from gevent's native threadpool when monkey-patching is active.
     """
     channel_layer = get_channel_layer()
-    try:
-        async_to_sync(channel_layer.group_send)(
-            group_name,
-            {
-                'type': event_type,
-                'data': data
-            }
-        )
-    except Exception as e:
-        logger.warning(f"Failed to send WebSocket update: {e}")
-    finally:
-        # Explicitly release references to help garbage collection
-        channel_layer = None
+    message = {'type': event_type, 'data': data}
 
-        # Force garbage collection if requested
-        if collect_garbage:
-            gc.collect()
+    def _do_send():
+        try:
+            async_to_sync(channel_layer.group_send)(group_name, message)
+        except Exception as e:
+            logger.warning(f"Failed to send WebSocket update: {e}")
+
+    try:
+        import gevent.monkey
+        if gevent.monkey.is_module_patched('threading'):
+            from gevent import get_hub
+            get_hub().threadpool.spawn(_do_send)
+            return
+    except Exception:
+        pass
+
+    # Not in a gevent-patched environment — call directly
+    _do_send()
+
+    if collect_garbage:
+        gc.collect()
 
 def send_websocket_event(event, success, data):
     """Acquire a lock to prevent concurrent task execution."""
diff --git a/frontend/src/WebSocket.jsx b/frontend/src/WebSocket.jsx
index f595b747..e687198c 100644
--- a/frontend/src/WebSocket.jsx
+++ b/frontend/src/WebSocket.jsx
@@ -19,6 +19,21 @@ import useAuthStore from './store/auth';
 
 export const WebsocketContext = createContext([false, () => {}, null]);
 
+// Debounce: coalesces rapid recording WS events into a single fetchRecordings()
+// call (400 ms window) to prevent redundant re-renders in the TV Guide.
+let _recordingFetchTimer = null;
+function scheduleRecordingFetch() {
+  if (_recordingFetchTimer) clearTimeout(_recordingFetchTimer);
+  _recordingFetchTimer = setTimeout(async () => {
+    _recordingFetchTimer = null;
+    try {
+      await useChannelsStore.getState().fetchRecordings();
+    } catch (e) {
+      console.warn('Failed to refresh recordings:', e);
+    }
+  }, 400);
+}
+
 export const WebsocketProvider = ({ children }) => {
   const [isReady, setIsReady] = useState(false);
   const [val, setVal] = useState(null);
@@ -193,9 +208,7 @@ export const WebsocketProvider = ({ children }) => {
                   loading: false,
                   autoClose: 4000,
                 });
-                try {
-                  await useChannelsStore.getState().fetchRecordings();
-                } catch {}
+                scheduleRecordingFetch();
               } else if (status === 'skipped') {
                 notifications.update({
                   id,
@@ -205,9 +218,7 @@ export const WebsocketProvider = ({ children }) => {
                   loading: false,
                   autoClose: 3000,
                 });
-                try {
-                  await useChannelsStore.getState().fetchRecordings();
-                } catch {}
+                scheduleRecordingFetch();
               } else if (status === 'error') {
                 notifications.update({
                   id,
@@ -217,9 +228,7 @@ export const WebsocketProvider = ({ children }) => {
                   loading: false,
                   autoClose: 6000,
                 });
-                try {
-                  await useChannelsStore.getState().fetchRecordings();
-                } catch {}
+                scheduleRecordingFetch();
               }
               break;
             }
@@ -508,19 +517,11 @@ export const WebsocketProvider = ({ children }) => {
               break;
 
             case 'recording_updated':
-              try {
-                await useChannelsStore.getState().fetchRecordings();
-              } catch (e) {
-                console.warn('Failed to refresh recordings on update:', e);
-              }
+              scheduleRecordingFetch();
               break;
 
             case 'recordings_refreshed':
-              try {
-                await useChannelsStore.getState().fetchRecordings();
-              } catch (e) {
-                console.warn('Failed to refresh recordings on refreshed:', e);
-              }
+              scheduleRecordingFetch();
               break;
 
             case 'recording_started':
@@ -528,11 +529,7 @@ export const WebsocketProvider = ({ children }) => {
                 title: 'Recording started!',
                 message: `Started recording channel ${parsedEvent.data.channel}`,
               });
-              try {
-                await useChannelsStore.getState().fetchRecordings();
-              } catch (e) {
-                console.warn('Failed to refresh recordings on start:', e);
-              }
+              scheduleRecordingFetch();
               break;
 
             case 'recording_ended':
@@ -540,11 +537,7 @@ export const WebsocketProvider = ({ children }) => {
                 title: 'Recording finished!',
                 message: `Stopped recording channel ${parsedEvent.data.channel}`,
               });
-              try {
-                await useChannelsStore.getState().fetchRecordings();
-              } catch (e) {
-                console.warn('Failed to refresh recordings on end:', e);
-              }
+              scheduleRecordingFetch();
               break;
 
             case 'recording_stopped':
@@ -553,11 +546,11 @@ export const WebsocketProvider = ({ children }) => {
                 message: `Recording stopped early for ${parsedEvent.data.channel || 'channel'}. Partial content has been saved.`,
                 color: 'yellow',
               });
-              try {
-                await useChannelsStore.getState().fetchRecordings();
-              } catch (e) {
-                console.warn('Failed to refresh recordings on stop:', e);
-              }
+              scheduleRecordingFetch();
+              break;
+
+            case 'recording_extended':
+              scheduleRecordingFetch();
               break;
 
             case 'recording_cancelled':
@@ -568,10 +561,12 @@ export const WebsocketProvider = ({ children }) => {
                   : 'Recording deleted.',
                 color: 'red',
               });
-              try {
-                await useChannelsStore.getState().fetchRecordings();
-              } catch (e) {
-                console.warn('Failed to refresh recordings on cancel:', e);
+              // Surgical removal by ID avoids a full fetchRecordings() re-render.
+              // Fall back to a full refresh if the ID is missing (e.g. older server).
+              if (parsedEvent.data.recording_id != null) {
+                useChannelsStore.getState().removeRecording(parsedEvent.data.recording_id);
+              } else {
+                scheduleRecordingFetch();
               }
               break;
 
diff --git a/frontend/src/api.js b/frontend/src/api.js
index 4fbf1018..43aca702 100644
--- a/frontend/src/api.js
+++ b/frontend/src/api.js
@@ -2673,6 +2673,20 @@ export default class API {
     }
   }
 
+  static async extendRecording(id, extraMinutes) {
+    try {
+      const resp = await request(`${host}/api/channels/recordings/${id}/extend/`, {
+        method: 'POST',
+        body: JSON.stringify({ extra_minutes: extraMinutes }),
+        headers: { 'Content-Type': 'application/json' },
+      });
+      return resp;
+    } catch (e) {
+      errorNotification(`Failed to extend recording ${id}`, e);
+      throw e;
+    }
+  }
+
   static async runComskip(recordingId) {
     try {
       const resp = await request(
diff --git a/frontend/src/components/cards/RecordingCard.jsx b/frontend/src/components/cards/RecordingCard.jsx
index 2f428768..2386748e 100644
--- a/frontend/src/components/cards/RecordingCard.jsx
+++ b/frontend/src/components/cards/RecordingCard.jsx
@@ -13,20 +13,21 @@ import {
   Box,
   Button,
   Card,
-  Center,
   Flex,
   Group,
   Image,
+  Menu,
   Modal,
   Stack,
   Text,
   Tooltip,
 } from '@mantine/core';
-import { AlertTriangle, Square, SquareX } from 'lucide-react';
+import { AlertTriangle, Plus, Square, SquareX } from 'lucide-react';
 import RecordingSynopsis from '../RecordingSynopsis';
 import {
   deleteRecordingById,
   deleteSeriesAndRule,
+  extendRecordingById,
   getPosterUrl,
   getRecordingUrl,
   getSeasonLabel,
@@ -43,7 +44,6 @@ const RecordingCard = ({
   onOpenRecurring,
   channel: channelProp = null,
 }) => {
-  const channels = useChannelsStore((s) => s.channels);
   const env_mode = useSettingsStore((s) => s.environment.env_mode);
   const showVideo = useVideoStore((s) => s.showVideo);
   const fetchRecordings = useChannelsStore((s) => s.fetchRecordings);
@@ -118,6 +118,27 @@ const RecordingCard = ({
     }
   };
 
+  const handleExtend = async (minutes, e) => {
+    e?.stopPropagation?.();
+    try {
+      await extendRecordingById(recording.id, minutes);
+      notifications.show({
+        title: 'Recording extended',
+        message: `Added ${minutes} minutes to this recording`,
+        color: 'teal',
+        autoClose: 2000,
+      });
+    } catch (error) {
+      console.error('Failed to extend recording', error);
+      notifications.show({
+        title: 'Extension failed',
+        message: 'Could not extend the recording',
+        color: 'red',
+        autoClose: 3000,
+      });
+    }
+  };
+
   // Stop / Cancel / Delete state and handlers
   const [cancelOpen, setCancelOpen] = React.useState(false);
   const [stopConfirmOpen, setStopConfirmOpen] = React.useState(false);
@@ -312,6 +333,30 @@ const RecordingCard = ({
         </Group>
 
         <Group gap={4}>
+          {isInProgress && (
+            <Tooltip label="Extend recording">
+              <Box display="inline-flex">
+                <Menu withinPortal position="bottom-end" shadow="md">
+                  <Menu.Target>
+                    <ActionIcon
+                      variant="transparent"
+                      color="teal.5"
+                      onMouseDown={(e) => e.stopPropagation()}
+                      onClick={(e) => e.stopPropagation()}
+                    >
+                      <Plus size={20} />
+                    </ActionIcon>
+                  </Menu.Target>
+                  <Menu.Dropdown onClick={(e) => e.stopPropagation()}>
+                    <Menu.Label>Extend recording by</Menu.Label>
+                    <Menu.Item onClick={(e) => handleExtend(15, e)}>+15 minutes</Menu.Item>
+                    <Menu.Item onClick={(e) => handleExtend(30, e)}>+30 minutes</Menu.Item>
+                    <Menu.Item onClick={(e) => handleExtend(60, e)}>+1 hour</Menu.Item>
+                  </Menu.Dropdown>
+                </Menu>
+              </Box>
+            </Tooltip>
+          )}
           {isInProgress && (
             <Tooltip label="Stop recording (keep partial content)">
               <ActionIcon
@@ -395,7 +440,7 @@ const RecordingCard = ({
 
             {!isUpcoming && <WatchRecording />}
             {!isUpcoming &&
-              customProps?.status === 'completed' &&
+              (customProps?.status === 'completed' || customProps?.status === 'stopped' || customProps?.status === 'interrupted') &&
               (!customProps?.comskip ||
                 customProps?.comskip?.status !== 'completed') && (
                 <Button
@@ -453,7 +498,7 @@ const RecordingCard = ({
       <Modal
         opened={deleteConfirmOpen}
         onClose={() => setDeleteConfirmOpen(false)}
-        title={isInProgress ? 'Cancel Recording' : isUpcoming ? 'Cancel Recording' : 'Delete Recording'}
+        title={isInProgress || isUpcoming ? 'Cancel Recording' : 'Delete Recording'}
         centered
         size="md"
         zIndex={9999}
diff --git a/frontend/src/components/forms/ProgramRecordingModal.jsx b/frontend/src/components/forms/ProgramRecordingModal.jsx
index 9503f26b..294cb2aa 100644
--- a/frontend/src/components/forms/ProgramRecordingModal.jsx
+++ b/frontend/src/components/forms/ProgramRecordingModal.jsx
@@ -1,6 +1,5 @@
 import React from 'react';
 import { Modal, Flex, Button } from '@mantine/core';
-import useChannelsStore from '../../store/channels.jsx';
 import { deleteRecordingById } from '../../utils/cards/RecordingCardUtils.js';
 import { deleteSeriesAndRule } from '../../utils/cards/RecordingCardUtils.js';
 import { deleteSeriesRuleByTvgId } from '../../pages/guideUtils.js';
@@ -22,11 +21,7 @@ export default function ProgramRecordingModal({
     } catch (error) {
       console.warn('Failed to delete recording', error);
     }
-    try {
-      await useChannelsStore.getState().fetchRecordings();
-    } catch (error) {
-      console.warn('Failed to refresh recordings after delete', error);
-    }
+    // recording_cancelled WS event triggers the debounced fetchRecordings()
     onClose();
   };
 
@@ -35,11 +30,7 @@ export default function ProgramRecordingModal({
       tvg_id: program.tvg_id,
       title: program.title,
     });
-    try {
-      await useChannelsStore.getState().fetchRecordings();
-    } catch (error) {
-      console.warn('Failed to refresh recordings after series delete', error);
-    }
+    // recordings_refreshed WS event triggers the debounced fetchRecordings()
     onClose();
   };
 
diff --git a/frontend/src/components/forms/RecordingDetailsModal.jsx b/frontend/src/components/forms/RecordingDetailsModal.jsx
index 5684adc3..bfd7abda 100644
--- a/frontend/src/components/forms/RecordingDetailsModal.jsx
+++ b/frontend/src/components/forms/RecordingDetailsModal.jsx
@@ -70,6 +70,7 @@ const RecordingDetailsModal = ({
   const fileUrl = customProps.file_url || customProps.output_file_url;
   const canWatchRecording =
     (customProps.status === 'completed' ||
+      customProps.status === 'stopped' ||
       customProps.status === 'interrupted') &&
     Boolean(fileUrl);
 
@@ -200,11 +201,7 @@ const RecordingDetailsModal = ({
       } catch (error) {
         console.error('Failed to delete upcoming recording', error);
       }
-      try {
-        await useChannelsStore.getState().fetchRecordings();
-      } catch (error) {
-        console.error('Failed to refresh recordings after delete', error);
-      }
+      // recording_cancelled WS event triggers the debounced fetchRecordings()
     };
 
     const handleOnMainCardClick = () => {
@@ -360,7 +357,9 @@ const RecordingDetailsModal = ({
               {onWatchLive && <WatchLive />}
               {onWatchRecording && <WatchRecording />}
               {onEdit && start.isAfter(userNow()) && <Edit />}
-              {customProps.status === 'completed' &&
+              {(customProps.status === 'completed' ||
+                customProps.status === 'stopped' ||
+                customProps.status === 'interrupted') &&
                 (!customProps?.comskip ||
                   customProps?.comskip?.status !== 'completed') && (
                   <Button
diff --git a/frontend/src/components/forms/RecurringRuleModal.jsx b/frontend/src/components/forms/RecurringRuleModal.jsx
index 9d8f5684..630dbee6 100644
--- a/frontend/src/components/forms/RecurringRuleModal.jsx
+++ b/frontend/src/components/forms/RecurringRuleModal.jsx
@@ -37,7 +37,6 @@ const RecurringRuleModal = ({ opened, onClose, ruleId, onEditOccurrence }) => {
   const channels = useChannelsStore((s) => s.channels);
   const recurringRules = useChannelsStore((s) => s.recurringRules);
   const fetchRecurringRules = useChannelsStore((s) => s.fetchRecurringRules);
-  const fetchRecordings = useChannelsStore((s) => s.fetchRecordings);
   const recordings = useChannelsStore((s) => s.recordings);
   const { toUserTime, userNow } = useTimeHelpers();
   const { timeFormat: timeformat, dateFormat: dateformat } =
@@ -125,7 +124,7 @@ const RecurringRuleModal = ({ opened, onClose, ruleId, onEditOccurrence }) => {
     setSaving(true);
     try {
       await updateRecurringRule(ruleId, values);
-      await Promise.all([fetchRecurringRules(), fetchRecordings()]);
+      await fetchRecurringRules(); // recordings_refreshed WS event handles recording list update
       notifications.show({
         title: 'Recurring rule updated',
         message: 'Schedule adjustments saved',
@@ -145,7 +144,7 @@ const RecurringRuleModal = ({ opened, onClose, ruleId, onEditOccurrence }) => {
     setDeleting(true);
     try {
       await deleteRecurringRuleById(ruleId);
-      await Promise.all([fetchRecurringRules(), fetchRecordings()]);
+      await fetchRecurringRules(); // recordings_refreshed WS event handles recording list update
       notifications.show({
         title: 'Recurring rule removed',
         message: 'All future occurrences were cancelled',
@@ -165,7 +164,7 @@ const RecurringRuleModal = ({ opened, onClose, ruleId, onEditOccurrence }) => {
     setSaving(true);
     try {
       await updateRecurringRuleEnabled(ruleId, checked);
-      await Promise.all([fetchRecurringRules(), fetchRecordings()]);
+      await fetchRecurringRules(); // recordings_refreshed WS event handles recording list update
       notifications.show({
         title: checked ? 'Recurring rule enabled' : 'Recurring rule paused',
         message: checked
@@ -186,7 +185,7 @@ const RecurringRuleModal = ({ opened, onClose, ruleId, onEditOccurrence }) => {
     setBusyOccurrence(occurrence.id);
     try {
       await deleteRecordingById(occurrence.id);
-      await fetchRecordings();
+      // recording_cancelled WS event handles recording list update
       notifications.show({
         title: 'Occurrence cancelled',
         message: 'The selected airing was removed',
diff --git a/frontend/src/pages/Guide.jsx b/frontend/src/pages/Guide.jsx
index 310ab78f..980f145e 100644
--- a/frontend/src/pages/Guide.jsx
+++ b/frontend/src/pages/Guide.jsx
@@ -93,7 +93,6 @@ export default function TVChannelGuide({ startDate, endDate }) {
   const recordings = useChannelsStore((s) => s.recordings);
   const channelGroups = useChannelsStore((s) => s.channelGroups);
   const profiles = useChannelsStore((s) => s.profiles);
-  const isLoading = useChannelsStore((s) => s.isLoading);
   const [isProgramsLoading, setIsProgramsLoading] = useState(true);
   const logos = useLogosStore((s) => s.logos);
 
@@ -699,14 +698,7 @@ export default function TVChannelGuide({ startDate, endDate }) {
   const saveSeriesRule = useCallback(async (program, mode) => {
     await createSeriesRule(program, mode);
     await evaluateSeriesRule(program);
-    try {
-      await useChannelsStore.getState().fetchRecordings();
-    } catch (error) {
-      console.warn(
-        'Failed to refresh recordings after saving series rule',
-        error
-      );
-    }
+    // recordings_refreshed WS event triggers the debounced fetchRecordings()
     showNotification({
       title: mode === 'new' ? 'Record new episodes' : 'Record all episodes',
     });
@@ -1351,7 +1343,7 @@ export default function TVChannelGuide({ startDate, endDate }) {
           pos="relative"
         >
           <LoadingOverlay
-            visible={isLoading || isProgramsLoading || isChannelsLoading}
+            visible={isProgramsLoading || isChannelsLoading}
           />
           {nowPosition >= 0 && (
             <Box
diff --git a/frontend/src/store/channels.jsx b/frontend/src/store/channels.jsx
index 15aa5776..24ad2d9d 100644
--- a/frontend/src/store/channels.jsx
+++ b/frontend/src/store/channels.jsx
@@ -459,15 +459,10 @@ const useChannelsStore = create((set, get) => ({
   },
 
   fetchRecordings: async () => {
-    set({ isLoading: true, error: null });
     try {
-      set({
-        recordings: await api.getRecordings(),
-        isLoading: false,
-      });
+      set({ recordings: await api.getRecordings() });
     } catch (error) {
       console.error('Failed to fetch recordings:', error);
-      set({ error: 'Failed to load recordings.', isLoading: false });
     }
   },
 
@@ -494,11 +489,16 @@ const useChannelsStore = create((set, get) => ({
       const target = String(id);
       const current = state.recordings;
       if (Array.isArray(current)) {
+        // Early return if item doesn't exist — avoids a new array reference
+        // (and thus a needless re-render) when called redundantly, e.g. both
+        // the optimistic API delete and the WS recording_cancelled handler.
+        if (!current.some((r) => String(r?.id) === target)) return {};
         return {
           recordings: current.filter((r) => String(r?.id) !== target),
         };
       }
       if (current && typeof current === 'object') {
+        if (!Object.values(current).some((r) => String(r?.id) === target)) return {};
         const next = { ...current };
         for (const k of Object.keys(next)) {
           try {
diff --git a/frontend/src/utils/cards/RecordingCardUtils.js b/frontend/src/utils/cards/RecordingCardUtils.js
index 60dd2cd5..3ca5703a 100644
--- a/frontend/src/utils/cards/RecordingCardUtils.js
+++ b/frontend/src/utils/cards/RecordingCardUtils.js
@@ -55,6 +55,10 @@ export const stopRecordingById = async (recordingId) => {
   await API.stopRecording(recordingId);
 };
 
+export const extendRecordingById = async (recordingId, extraMinutes) => {
+  await API.extendRecording(recordingId, extraMinutes);
+};
+
 export const deleteSeriesAndRule = async (seriesInfo) => {
   const { tvg_id, title } = seriesInfo;
   if (tvg_id) {

From e5206545ddd002e607276586b42b28a888ba60c3 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Sun, 1 Mar 2026 18:25:04 -0600
Subject: [PATCH 37/58] fix(dvr): add stream reconnection resilience, fix
 comskip exit code handling

- Add reconnection loop to DVR streaming: on transient connectivity loss (ReadTimeout, ConnectionError, ChunkedEncodingError), the recording task retries the same TS proxy base up to 5 times with a 2-second delay, appending to the existing file. Counter resets on successful data receipt. TS container format tolerates brief discontinuities; MKV remux normalises timestamps.
- Extract _check_recording_cancelled() helper to consolidate repeated stop/delete DB checks across the streaming and reconnection paths.
- Fix comskip treating exit code 1 (no commercials detected) as a fatal error.
  Use check=False and handle return codes explicitly: 0 = commercials found,
  1 = no commercials (skipped), negative = signal, other = real error.
- Add human-readable reason messages for comskip 'skipped' notifications in
  WebSocket handler.
- Remove redundant conditional Channel import in stream_generator._cleanup() that shadowed the module-level import, causing UnboundLocalError when the conditional branch did not execute.
---
 apps/channels/tasks.py                  | 312 +++++++++++++++---------
 apps/proxy/ts_proxy/stream_generator.py |   1 -
 frontend/src/WebSocket.jsx              |   6 +-
 3 files changed, 202 insertions(+), 117 deletions(-)

diff --git a/apps/channels/tasks.py b/apps/channels/tasks.py
index 26df8c78..14ddaef0 100755
--- a/apps/channels/tasks.py
+++ b/apps/channels/tasks.py
@@ -1878,59 +1878,86 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     interrupted = False
     interrupted_reason = None
 
-    # We'll attempt each base until we receive some data
-    for base in candidates:
+    def _check_recording_cancelled(rid):
+        """Check if a recording was stopped by user or deleted.
+
+        Returns (should_exit, is_interrupted, reason) where should_exit
+        indicates the stream loop must terminate.
+        """
         try:
-            test_url = f"{base.rstrip('/')}/proxy/ts/stream/{channel.uuid}"
-            logger.info(f"DVR: trying TS base {base} -> {test_url}")
+            rec = Recording.objects.filter(id=rid).only("custom_properties").first()
+            if rec is None:
+                return True, True, "recording_deleted"
+            if (rec.custom_properties or {}).get("status") == "stopped":
+                return True, False, "stopped_by_user"
+        except Exception:
+            pass
+        return False, False, None
 
-            with requests.get(
-                test_url,
-                headers={
-                    'User-Agent': f'Dispatcharr-DVR/recording-{recording_id}',
-                },
-                stream=True,
-                timeout=(10, 15),
-            ) as response:
-                response.raise_for_status()
+    # --- Stream with reconnection ---
+    # Iterate candidate base URLs until one delivers data.  Once streaming,
+    # transient connectivity losses (upstream provider drops, proxy
+    # reconnects) trigger automatic re-connection to the same base rather
+    # than aborting the recording.  TS container format tolerates the brief
+    # discontinuity; the MKV remux normalises timestamps afterwards.
+    _dvr_max_reconnects = 5
+    _dvr_reconnect_delay = 2.0  # seconds
 
-                # Open the file and start copying; if we get any data within a short window, accept this base
-                got_any_data = False
-                test_window = 3.0  # seconds to detect first bytes
-                window_start = time.time()
+    for base in candidates:
+        test_url = f"{base.rstrip('/')}/proxy/ts/stream/{channel.uuid}"
+        logger.info(f"DVR recording {recording_id}: trying TS base {base}")
 
-                with open(temp_ts_path, 'wb') as file:
-                    started_at = time.time()
-                    for chunk in response.iter_content(chunk_size=8192):
-                        if not chunk:
-                            # keep-alives may be empty; continue
-                            if not got_any_data and (time.time() - window_start) > test_window:
-                                break
-                            continue
-                        # We have data
-                        got_any_data = True
-                        chosen_base = base
-                        # Fall through to full recording loop using this same response/connection
-                        file.write(chunk)
-                        bytes_written += len(chunk)
-                        elapsed = time.time() - started_at
-                        if elapsed > duration_seconds:
-                            break
-                        # Continue draining the stream.
-                        # Periodically poll the DB for a user-initiated stop so
-                        # the task exits promptly if stop_client() was missed
-                        # (e.g., fired before the DVR client was registered in Redis).
-                        _stop_poll_interval = 2.0  # seconds between DB checks
-                        _last_stop_poll = time.time()
-                        for chunk2 in response.iter_content(chunk_size=8192):
-                            if not chunk2:
+        _reconnects = 0
+        _file_mode = 'wb'
+        _stream_started_at = None
+        _done = False
+
+        while True:  # Reconnection loop for this base
+            try:
+                with requests.get(
+                    test_url,
+                    headers={
+                        'User-Agent': f'Dispatcharr-DVR/recording-{recording_id}',
+                    },
+                    stream=True,
+                    timeout=(10, 15),
+                ) as response:
+                    response.raise_for_status()
+
+                    _test_window = 3.0
+                    _window_start = time.time()
+                    _stop_poll_interval = 2.0
+                    _last_stop_poll = time.time()
+
+                    with open(temp_ts_path, _file_mode) as file:
+                        if _stream_started_at is None:
+                            _stream_started_at = time.time()
+
+                        for chunk in response.iter_content(chunk_size=8192):
+                            if not chunk:
+                                if not chosen_base and (time.time() - _window_start) > _test_window:
+                                    break
                                 continue
-                            file.write(chunk2)
-                            bytes_written += len(chunk2)
-                            elapsed = time.time() - started_at
+
+                            if not chosen_base:
+                                chosen_base = base
+
+                            # Data received after reconnect — connection restored
+                            if _reconnects > 0:
+                                logger.info(
+                                    f"DVR recording {recording_id}: "
+                                    f"stream resumed after reconnect"
+                                )
+                                _reconnects = 0
+
+                            file.write(chunk)
+                            bytes_written += len(chunk)
+
+                            elapsed = time.time() - _stream_started_at
                             if elapsed > duration_seconds:
                                 break
-                            # Periodic stopped-status + end_time extension check
+
+                            # Periodic DB poll: stop, delete, end_time extension
                             _now = time.time()
                             if _now - _last_stop_poll >= _stop_poll_interval:
                                 _last_stop_poll = _now
@@ -1939,7 +1966,6 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
                                         id=recording_id
                                     ).only("custom_properties", "end_time").first()
                                     if _sc is None:
-                                        # Recording was deleted — exit gracefully.
                                         logger.info(
                                             f"DVR recording {recording_id}: "
                                             f"deleted — exiting stream loop"
@@ -1948,11 +1974,10 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
                                         break
                                     if (_sc.custom_properties or {}).get("status") == "stopped":
                                         logger.info(
-                                            f"DVR recording {recording_id}: stop requested "
-                                            f"— exiting stream loop early"
+                                            f"DVR recording {recording_id}: "
+                                            f"stop requested — exiting stream loop"
                                         )
                                         break
-                                    # Check for extended end_time
                                     try:
                                         new_end = _sc.end_time
                                         if new_end is not None:
@@ -1962,10 +1987,12 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
                                             _ref = start_time
                                             if _tz.is_naive(_ref):
                                                 _ref = _tz.make_aware(_ref)
-                                            new_duration = int((new_end - _ref).total_seconds())
+                                            new_duration = int(
+                                                (new_end - _ref).total_seconds()
+                                            )
                                             if new_duration > duration_seconds:
                                                 logger.info(
-                                                    f"run_recording {recording_id}: "
+                                                    f"DVR recording {recording_id}: "
                                                     f"end_time extended to {new_end}, "
                                                     f"new duration {new_duration}s"
                                                 )
@@ -1974,54 +2001,100 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
                                         pass
                                 except Exception:
                                     pass
-                        break  # exit outer for-loop once we switched to full drain
 
-                # If we wrote any bytes, treat as success and stop trying candidates
+                    # iter_content exhausted or loop exited normally
+                    if bytes_written > 0:
+                        logger.info(
+                            f"DVR recording {recording_id}: "
+                            f"stream complete, {bytes_written} bytes written"
+                        )
+                        _done = True
+                    else:
+                        last_error = f"no_data_from_{base}"
+                        logger.warning(
+                            f"DVR recording {recording_id}: no data from "
+                            f"{base} within {_test_window}s, trying next base"
+                        )
+                        try:
+                            if os.path.exists(temp_ts_path) and os.path.getsize(temp_ts_path) == 0:
+                                os.remove(temp_ts_path)
+                        except FileNotFoundError:
+                            pass
+                    break  # Exit reconnection loop
+
+            except (ReadTimeout, ReqConnectionError, ChunkedEncodingError) as e:
                 if bytes_written > 0:
-                    logger.info(f"DVR: selected TS base {base}; wrote initial {bytes_written} bytes")
+                    # Active stream lost — check cancellation before reconnecting
+                    should_exit, is_int, reason = _check_recording_cancelled(recording_id)
+                    if should_exit:
+                        interrupted = is_int
+                        interrupted_reason = reason
+                        if reason == "stopped_by_user":
+                            logger.info(
+                                f"DVR recording {recording_id}: "
+                                f"stopped by user — ending stream"
+                            )
+                        _done = True
+                        break
+
+                    _reconnects += 1
+                    if _reconnects <= _dvr_max_reconnects:
+                        logger.warning(
+                            f"DVR recording {recording_id}: connection lost "
+                            f"({type(e).__name__}), reconnecting "
+                            f"({_reconnects}/{_dvr_max_reconnects}) "
+                            f"in {_dvr_reconnect_delay}s..."
+                        )
+                        time.sleep(_dvr_reconnect_delay)
+                        _file_mode = 'ab'
+                        continue
+
+                    logger.error(
+                        f"DVR recording {recording_id}: max reconnects "
+                        f"({_dvr_max_reconnects}) exceeded — ending recording"
+                    )
+                    interrupted = True
+                    interrupted_reason = (
+                        f"stream_interrupted: max reconnects exceeded ({e})"
+                    )
+                    _done = True
                     break
-                else:
-                    last_error = f"no_data_from_{base}"
-                    logger.warning(f"DVR: no data received from {base} within {test_window}s, trying next base")
-                    # Clean up empty temp file
-                    try:
-                        if os.path.exists(temp_ts_path) and os.path.getsize(temp_ts_path) == 0:
-                            os.remove(temp_ts_path)
-                    except Exception:
-                        pass
-        except Exception as e:
-            last_error = str(e)
-            logger.warning(f"DVR: attempt failed for base {base}: {e}")
-            # If there was a working stream that was interrupted (e.g. recording
-            # cancelled via stop_client), do not retry on another base — that would
-            # re-establish the DVR connection and make the recording re-appear.
-            if bytes_written > 0:
-                # Check if this was a deliberate "stop" vs an unexpected interruption
-                try:
-                    rec_check = Recording.objects.filter(id=recording_id).only("custom_properties").first()
-                    if rec_check and (rec_check.custom_properties or {}).get("status") == "stopped":
-                        # Graceful stop — not an error, just an early end
+
+                # No data received yet — try next candidate base
+                last_error = str(e)
+                logger.warning(f"DVR recording {recording_id}: base {base} failed: {e}")
+                should_exit, is_int, reason = _check_recording_cancelled(recording_id)
+                if should_exit:
+                    interrupted = is_int
+                    interrupted_reason = reason
+                    _done = True
+                break
+
+            except Exception as e:
+                last_error = str(e)
+                logger.warning(f"DVR recording {recording_id}: base {base} failed: {e}")
+                if bytes_written > 0:
+                    should_exit, is_int, reason = _check_recording_cancelled(recording_id)
+                    if should_exit and reason == "stopped_by_user":
                         interrupted = False
-                        logger.info(f"Recording {recording_id} was stopped by user — ending stream.")
+                        logger.info(
+                            f"DVR recording {recording_id}: "
+                            f"stopped by user — ending stream"
+                        )
                     else:
                         interrupted = True
                         interrupted_reason = f"stream_interrupted: {e}"
-                except Exception:
-                    interrupted = True
-                    interrupted_reason = f"stream_interrupted: {e}"
+                    _done = True
+                    break
+                should_exit, is_int, reason = _check_recording_cancelled(recording_id)
+                if should_exit:
+                    interrupted = is_int
+                    interrupted_reason = reason
+                    _done = True
                 break
-            # Check if recording was deleted or stopped before data arrived
-            try:
-                rec_check = Recording.objects.filter(id=recording_id).only("custom_properties").first()
-                if rec_check is None:
-                    interrupted = True
-                    interrupted_reason = "recording_deleted_during_connect"
-                    break
-                if (rec_check.custom_properties or {}).get("status") == "stopped":
-                    interrupted = False
-                    break
-            except Exception:
-                pass
+
+        if _done:
+            break
 
     if chosen_base is None and bytes_written == 0:
         interrupted = True
@@ -2487,34 +2560,43 @@ def comskip_process_recording(recording_id: int):
                 cmd.extend([f"--ini={ini_path}"])
                 break
         cmd.append(file_path)
-        subprocess.run(
+        result = subprocess.run(
             cmd,
-            check=True,
             stdout=subprocess.PIPE,
             stderr=subprocess.PIPE,
             text=True,
         )
-    except subprocess.CalledProcessError as e:
-        stderr_tail = (e.stderr or "").strip().splitlines()
-        stderr_tail = stderr_tail[-5:] if stderr_tail else []
-        detail = {
-            "status": "error",
-            "reason": "comskip_failed",
-            "returncode": e.returncode,
-        }
-        if e.returncode and e.returncode < 0:
-            try:
-                detail["signal"] = signal.Signals(-e.returncode).name
-            except Exception:
-                detail["signal"] = f"signal_{-e.returncode}"
-        if stderr_tail:
-            detail["stderr"] = "\n".join(stderr_tail)
-        if selected_ini:
-            detail["ini_path"] = selected_ini
-        cp["comskip"] = detail
-        _persist_custom_properties()
-        _ws('error', {"reason": "comskip_failed", "returncode": e.returncode})
-        return "comskip_failed"
+        # comskip exit codes: 0 = commercials found, 1 = no commercials detected.
+        # Negative codes indicate killed by signal; anything else is a real error.
+        if result.returncode == 1:
+            # No commercials detected — not an error.
+            cp["comskip"] = {"status": "completed", "skipped": True}
+            if selected_ini:
+                cp["comskip"]["ini_path"] = selected_ini
+            _persist_custom_properties()
+            _ws('skipped', {"reason": "no_commercials_detected"})
+            return "no_commercials"
+        elif result.returncode != 0:
+            stderr_tail = (result.stderr or "").strip().splitlines()
+            stderr_tail = stderr_tail[-5:] if stderr_tail else []
+            detail = {
+                "status": "error",
+                "reason": "comskip_failed",
+                "returncode": result.returncode,
+            }
+            if result.returncode < 0:
+                try:
+                    detail["signal"] = signal.Signals(-result.returncode).name
+                except Exception:
+                    detail["signal"] = f"signal_{-result.returncode}"
+            if stderr_tail:
+                detail["stderr"] = "\n".join(stderr_tail)
+            if selected_ini:
+                detail["ini_path"] = selected_ini
+            cp["comskip"] = detail
+            _persist_custom_properties()
+            _ws('error', {"reason": "comskip_failed", "returncode": result.returncode})
+            return "comskip_failed"
     except Exception as e:
         cp["comskip"] = {"status": "error", "reason": f"comskip_failed: {e}"}
         _persist_custom_properties()
diff --git a/apps/proxy/ts_proxy/stream_generator.py b/apps/proxy/ts_proxy/stream_generator.py
index 8ba38fc0..f66572d6 100644
--- a/apps/proxy/ts_proxy/stream_generator.py
+++ b/apps/proxy/ts_proxy/stream_generator.py
@@ -456,7 +456,6 @@ class StreamGenerator:
                             client_count = proxy_server.client_managers[self.channel_id].get_total_client_count()
                             # Only the last client or owner should release the stream
                             if client_count <= 1 and proxy_server.am_i_owner(self.channel_id):
-                                from apps.channels.models import Channel
                                 try:
                                     # Get the channel by UUID
                                     channel = Channel.objects.get(uuid=self.channel_id)
diff --git a/frontend/src/WebSocket.jsx b/frontend/src/WebSocket.jsx
index e687198c..1df2e295 100644
--- a/frontend/src/WebSocket.jsx
+++ b/frontend/src/WebSocket.jsx
@@ -210,10 +210,14 @@ export const WebsocketProvider = ({ children }) => {
                 });
                 scheduleRecordingFetch();
               } else if (status === 'skipped') {
+                const reasonMap = {
+                  no_commercials_detected: 'No commercials were detected in this recording',
+                  no_commercials: 'No commercials were detected in this recording',
+                };
                 notifications.update({
                   id,
                   title: 'No commercials to remove',
-                  message: parsedEvent.data.reason || '',
+                  message: reasonMap[parsedEvent.data.reason] || parsedEvent.data.reason || '',
                   color: 'teal',
                   loading: false,
                   autoClose: 3000,

From cf3165ea88bc30d9033d2cbff7d1d5913d0399ab Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Sun, 1 Mar 2026 20:30:56 -0600
Subject: [PATCH 38/58] fix(dvr): add DB retry resilience, fix recording card
 logos, prevent incorrect artwork

Add exponential backoff retry (_db_retry helper) for transient DB errors during final metadata save, startup recovery, and initial TS proxy
connection. Add FFmpeg remux retry for transient I/O errors. Retry same proxy base before falling back to next candidate on initial connection failure.

Fix DVR card logos not displaying: channel summary API returns logo_id (integer) but frontend expected a nested logo object with cache_url.

Add getChannelLogoUrl() helper that handles both data shapes. Backend artwork prefetch and run_recording now fall back to the channel's own logo when no show-specific artwork is found, ensuring poster_logo_id is always populated.

Prevent incorrect artwork by skipping external API searches (TVMaze, iTunes) when no program title is available - channel names like
"USA A&E SD*" returned unrelated results from fuzzy matching.

Add 13 tests for retry logic (test_dvr_retry.py)
---
 apps/channels/tasks.py                        | 304 ++++++++++++------
 apps/channels/tests/test_dvr_retry.py         | 294 +++++++++++++++++
 .../src/components/cards/RecordingCard.jsx    |   3 +-
 .../forms/RecordingDetailsModal.jsx           |   5 +-
 frontend/src/pages/DVR.jsx                    |   5 +-
 .../src/utils/cards/RecordingCardUtils.js     |  10 +
 6 files changed, 512 insertions(+), 109 deletions(-)
 create mode 100644 apps/channels/tests/test_dvr_retry.py

diff --git a/apps/channels/tasks.py b/apps/channels/tasks.py
index 14ddaef0..4bbd082c 100755
--- a/apps/channels/tasks.py
+++ b/apps/channels/tasks.py
@@ -30,6 +30,29 @@ from urllib.parse import quote
 
 logger = logging.getLogger(__name__)
 
+
+def _db_retry(fn, max_retries=3, base_interval=1, label="DB operation"):
+    """Execute fn() with exponential backoff retry on transient DB errors.
+
+    Follows the same backoff pattern as RedisClient.get_client().
+    Resets stale connections between attempts so the ORM reconnects.
+    """
+    from django.db import OperationalError, close_old_connections
+    for attempt in range(max_retries):
+        try:
+            return fn()
+        except OperationalError:
+            if attempt + 1 >= max_retries:
+                raise
+            wait = base_interval * (2 ** attempt)
+            logger.warning(
+                f"{label}: failed, retrying in {wait}s "
+                f"({attempt + 1}/{max_retries})..."
+            )
+            close_old_connections()
+            time.sleep(wait)
+
+
 # PostgreSQL btree index has a limit of ~2704 bytes (1/3 of 8KB page size)
 # We use 2000 as a safe maximum to account for multibyte characters
 def validate_logo_url(logo_url, max_length=2000):
@@ -1743,10 +1766,12 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
             except Exception as e:
                 logger.debug(f"External poster lookup failed: {e}")
 
-        # Keyless fallback providers (no API keys required)
-        if not poster_url and not poster_logo_id:
+        # Keyless fallback providers (no API keys required) — only search when
+        # real program title is present; channel names return unrelated results.
+        _program_title = (program.get('title') or '').strip()
+        if not poster_url and not poster_logo_id and _program_title:
             try:
-                title = (program.get('title') or channel.name or '').strip()
+                title = _program_title
                 if title:
                     # 1) TVMaze (TV shows) - singlesearch by title
                     try:
@@ -1802,6 +1827,10 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
             except Exception as e:
                 logger.debug(f"Unable to persist poster to Logo: {e}")
 
+        # Fall back to the channel's own logo so the card always has an image
+        if not poster_logo_id and not poster_url and channel.logo_id:
+            poster_logo_id = channel.logo_id
+
         if poster_logo_id:
             cp["poster_logo_id"] = poster_logo_id
         if poster_url and "poster_url" not in cp:
@@ -1894,14 +1923,16 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
             pass
         return False, False, None
 
-    # --- Stream with reconnection ---
-    # Iterate candidate base URLs until one delivers data.  Once streaming,
-    # transient connectivity losses (upstream provider drops, proxy
-    # reconnects) trigger automatic re-connection to the same base rather
-    # than aborting the recording.  TS container format tolerates the brief
-    # discontinuity; the MKV remux normalises timestamps afterwards.
+    # --- Retry / reconnection constants ---
+    # Stream reconnection: retry the same TS proxy base on transient
+    # connectivity loss.  Counter resets when data resumes.
     _dvr_max_reconnects = 5
     _dvr_reconnect_delay = 2.0  # seconds
+    # DB save retry: exponential backoff (1s, 2s, 4s) for transient errors.
+    _dvr_db_max_retries = 3
+    _dvr_db_retry_interval = 1  # seconds (base for exponential backoff)
+    # FFmpeg remux retry: covers transient I/O errors.
+    _dvr_remux_max_retries = 2
 
     for base in candidates:
         test_url = f"{base.rstrip('/')}/proxy/ts/stream/{channel.uuid}"
@@ -2060,14 +2091,28 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
                     _done = True
                     break
 
-                # No data received yet — try next candidate base
-                last_error = str(e)
-                logger.warning(f"DVR recording {recording_id}: base {base} failed: {e}")
+                # No data received yet — retry same base before moving on
                 should_exit, is_int, reason = _check_recording_cancelled(recording_id)
                 if should_exit:
                     interrupted = is_int
                     interrupted_reason = reason
                     _done = True
+                    break
+                _reconnects += 1
+                if _reconnects <= _dvr_max_reconnects:
+                    logger.warning(
+                        f"DVR recording {recording_id}: initial connection "
+                        f"to {base} failed ({type(e).__name__}), retrying "
+                        f"({_reconnects}/{_dvr_max_reconnects}) "
+                        f"in {_dvr_reconnect_delay}s..."
+                    )
+                    time.sleep(_dvr_reconnect_delay)
+                    continue
+                last_error = str(e)
+                logger.warning(
+                    f"DVR recording {recording_id}: base {base} exhausted "
+                    f"retries ({_dvr_max_reconnects}): {e}"
+                )
                 break
 
             except Exception as e:
@@ -2174,101 +2219,115 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
                 pass
         return
 
-    # Remux TS to MKV container
+    # Remux TS to MKV container with retry for transient I/O errors
     remux_success = False
-    try:
-        if temp_ts_path and os.path.exists(temp_ts_path):
-            # First attempt: Direct TS to MKV remux
-            result = subprocess.run([
-                "ffmpeg", "-y",
-                "-fflags", "+genpts+igndts+discardcorrupt",  # Regenerate timestamps, ignore DTS
-                "-err_detect", "ignore_err",   # Ignore minor stream errors
-                "-i", temp_ts_path,
-                "-map", "0", # Map all streams
-                "-c", "copy",
-                final_path
-            ], stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True)
-
-            # Check if FFmpeg succeeded (return code 0) and output file is valid
-            if result.returncode == 0 and os.path.exists(final_path) and os.path.getsize(final_path) > 0:
-                remux_success = True
-                logger.info(f"Direct TS→MKV remux succeeded for {os.path.basename(final_path)}")
-            else:
-                # Direct remux failed - try fallback: TS → MP4 → MKV to fix timestamps
-                logger.warning(f"Direct TS→MKV remux failed (return code: {result.returncode}), trying fallback TS→MP4→MKV")
-
-                # Clean up partial/failed MKV
-                try:
-                    if os.path.exists(final_path):
-                        os.remove(final_path)
-                except Exception:
-                    pass
-
-                # Step 1: TS → MP4 (MP4 container handles broken timestamps better)
-                temp_mp4_path = os.path.splitext(temp_ts_path)[0] + ".mp4"
-                result_mp4 = subprocess.run([
+    for _remux_attempt in range(_dvr_remux_max_retries):
+        try:
+            if temp_ts_path and os.path.exists(temp_ts_path):
+                # First attempt: Direct TS to MKV remux
+                result = subprocess.run([
                     "ffmpeg", "-y",
-                    "-fflags", "+genpts+igndts+discardcorrupt",
-                    "-err_detect", "ignore_err",
+                    "-fflags", "+genpts+igndts+discardcorrupt",  # Regenerate timestamps, ignore DTS
+                    "-err_detect", "ignore_err",   # Ignore minor stream errors
                     "-i", temp_ts_path,
-                    "-map", "0",
+                    "-map", "0",  # Map all streams
                     "-c", "copy",
-                    temp_mp4_path
+                    final_path
                 ], stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True)
 
-                if result_mp4.returncode == 0 and os.path.exists(temp_mp4_path) and os.path.getsize(temp_mp4_path) > 0:
-                    logger.info(f"TS→MP4 conversion succeeded, now converting MP4→MKV")
+                # Check if FFmpeg succeeded (return code 0) and output file is valid
+                if result.returncode == 0 and os.path.exists(final_path) and os.path.getsize(final_path) > 0:
+                    remux_success = True
+                    logger.info(f"Direct TS→MKV remux succeeded for {os.path.basename(final_path)}")
+                else:
+                    # Direct remux failed - try fallback: TS → MP4 → MKV to fix timestamps
+                    logger.warning(f"Direct TS→MKV remux failed (return code: {result.returncode}), trying fallback TS→MP4→MKV")
 
-                    # Step 2: MP4 → MKV (clean timestamps from MP4)
-                    result_mkv = subprocess.run([
-                        "ffmpeg", "-y",
-                        "-i", temp_mp4_path,
-                        "-map", "0",
-                        "-c", "copy",
-                        final_path
-                    ], stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True)
-
-                    if result_mkv.returncode == 0 and os.path.exists(final_path) and os.path.getsize(final_path) > 0:
-                        remux_success = True
-                        logger.info(f"Fallback TS→MP4→MKV remux succeeded for {os.path.basename(final_path)}")
-                    else:
-                        logger.error(f"MP4→MKV conversion failed (return code: {result_mkv.returncode})")
-
-                    # Clean up temp MP4
+                    # Clean up partial/failed MKV
                     try:
-                        if os.path.exists(temp_mp4_path):
-                            os.remove(temp_mp4_path)
+                        if os.path.exists(final_path):
+                            os.remove(final_path)
                     except Exception:
                         pass
+
+                    # Step 1: TS → MP4 (MP4 container handles broken timestamps better)
+                    temp_mp4_path = os.path.splitext(temp_ts_path)[0] + ".mp4"
+                    result_mp4 = subprocess.run([
+                        "ffmpeg", "-y",
+                        "-fflags", "+genpts+igndts+discardcorrupt",
+                        "-err_detect", "ignore_err",
+                        "-i", temp_ts_path,
+                        "-map", "0",
+                        "-c", "copy",
+                        temp_mp4_path
+                    ], stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True)
+
+                    if result_mp4.returncode == 0 and os.path.exists(temp_mp4_path) and os.path.getsize(temp_mp4_path) > 0:
+                        logger.info(f"TS→MP4 conversion succeeded, now converting MP4→MKV")
+
+                        # Step 2: MP4 → MKV (clean timestamps from MP4)
+                        result_mkv = subprocess.run([
+                            "ffmpeg", "-y",
+                            "-i", temp_mp4_path,
+                            "-map", "0",
+                            "-c", "copy",
+                            final_path
+                        ], stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True)
+
+                        if result_mkv.returncode == 0 and os.path.exists(final_path) and os.path.getsize(final_path) > 0:
+                            remux_success = True
+                            logger.info(f"Fallback TS→MP4→MKV remux succeeded for {os.path.basename(final_path)}")
+                        else:
+                            logger.error(f"MP4→MKV conversion failed (return code: {result_mkv.returncode})")
+
+                        # Clean up temp MP4
+                        try:
+                            if os.path.exists(temp_mp4_path):
+                                os.remove(temp_mp4_path)
+                        except Exception:
+                            pass
+                    else:
+                        logger.error(f"TS→MP4 conversion failed (return code: {result_mp4.returncode})")
+
+                # Clean up temp TS file only on successful remux
+                if remux_success:
+                    try:
+                        os.remove(temp_ts_path)
+                        logger.debug(f"Cleaned up temp TS file: {temp_ts_path}")
+                    except Exception as e:
+                        logger.warning(f"Failed to remove temp TS file: {e}")
                 else:
-                    logger.error(f"TS→MP4 conversion failed (return code: {result_mp4.returncode})")
+                    # Keep TS file for debugging/manual recovery if remux failed
+                    logger.warning(f"Remux failed - keeping temp TS file for recovery: {temp_ts_path}")
+                    # Clean up any partial MKV
+                    try:
+                        if os.path.exists(final_path):
+                            os.remove(final_path)
+                            logger.debug(f"Cleaned up partial MKV file: {final_path}")
+                    except Exception:
+                        pass
+            break  # Completed (success or deterministic failure)
 
-            # Clean up temp TS file only on successful remux
-            if remux_success:
-                try:
-                    os.remove(temp_ts_path)
-                    logger.debug(f"Cleaned up temp TS file: {temp_ts_path}")
-                except Exception as e:
-                    logger.warning(f"Failed to remove temp TS file: {e}")
+        except Exception as e:
+            # Clean up partial output before potential retry
+            try:
+                if os.path.exists(final_path):
+                    os.remove(final_path)
+            except Exception:
+                pass
+            if _remux_attempt + 1 < _dvr_remux_max_retries:
+                _wait = _dvr_db_retry_interval * (2 ** _remux_attempt)
+                logger.warning(
+                    f"DVR recording {recording_id}: remux failed "
+                    f"({type(e).__name__}), retrying in {_wait}s "
+                    f"({_remux_attempt + 1}/{_dvr_remux_max_retries})..."
+                )
+                time.sleep(_wait)
             else:
-                # Keep TS file for debugging/manual recovery if remux failed
-                logger.warning(f"Remux failed - keeping temp TS file for recovery: {temp_ts_path}")
-                # Clean up any partial MKV
-                try:
-                    if os.path.exists(final_path):
-                        os.remove(final_path)
-                        logger.debug(f"Cleaned up partial MKV file: {final_path}")
-                except Exception:
-                    pass
-
-    except Exception as e:
-        logger.warning(f"MKV remux failed with exception: {e}")
-        # Clean up any partial files on exception
-        try:
-            if os.path.exists(final_path):
-                os.remove(final_path)
-        except Exception:
-            pass
+                logger.warning(
+                    f"DVR recording {recording_id}: remux failed "
+                    f"after {_dvr_remux_max_retries} attempts: {e}"
+                )
 
     # Persist final metadata to Recording (status, ended_at, and stream stats if available)
     try:
@@ -2364,7 +2423,17 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
             return
 
         recording_obj.custom_properties = cp
-        recording_obj.save(update_fields=["custom_properties"])
+
+        def _save_final_metadata():
+            recording_obj.custom_properties = cp
+            recording_obj.save(update_fields=["custom_properties"])
+
+        _db_retry(
+            _save_final_metadata,
+            max_retries=_dvr_db_max_retries,
+            base_interval=_dvr_db_retry_interval,
+            label=f"DVR recording {recording_id}: metadata save",
+        )
 
         # Notify frontends so the UI refreshes immediately (e.g. "Stopped" → "Completed")
         try:
@@ -2412,8 +2481,14 @@ def recover_recordings_on_startup():
 
         now = timezone.now()
 
-        # Resume in-window recordings
-        active = Recording.objects.filter(start_time__lte=now, end_time__gt=now)
+        # Resume in-window recordings.  DB queries and saves use _db_retry
+        # to tolerate transient connection errors common during startup.
+        active = _db_retry(
+            lambda: list(Recording.objects.filter(
+                start_time__lte=now, end_time__gt=now
+            )),
+            label="DVR recovery: fetching active recordings",
+        )
         for rec in active:
             try:
                 cp = rec.custom_properties or {}
@@ -2435,7 +2510,10 @@ def recover_recordings_on_startup():
                 cp["status"] = "interrupted"
                 cp["interrupted_reason"] = "server_restarted"
                 rec.custom_properties = cp
-                rec.save(update_fields=["custom_properties"])
+                _db_retry(
+                    lambda r=rec: r.save(update_fields=["custom_properties"]),
+                    label=f"DVR recovery: recording {rec.id} status update",
+                )
 
                 # Start recording for remaining window
                 run_recording.apply_async(
@@ -2447,7 +2525,12 @@ def recover_recordings_on_startup():
         # Ensure future recordings are scheduled.
         # With ClockedSchedule, PeriodicTasks survive restarts in the DB.
         # Only recreate if the PeriodicTask is missing (safety net).
-        upcoming = Recording.objects.filter(start_time__gt=now, end_time__gt=now)
+        upcoming = _db_retry(
+            lambda: list(Recording.objects.filter(
+                start_time__gt=now, end_time__gt=now
+            )),
+            label="DVR recovery: fetching upcoming recordings",
+        )
         for rec in upcoming:
             try:
                 from django_celery_beat.models import PeriodicTask as _PT
@@ -2456,13 +2539,19 @@ def recover_recordings_on_startup():
                 if _PT.objects.filter(name=task_name).exists():
                     if rec.task_id != task_name:
                         rec.task_id = task_name
-                        rec.save(update_fields=["task_id"])
+                        _db_retry(
+                            lambda r=rec: r.save(update_fields=["task_id"]),
+                            label=f"DVR recovery: recording {rec.id} task_id update",
+                        )
                     continue
                 # PeriodicTask missing - recreate it
                 task_id = schedule_recording_task(rec)
                 if task_id:
                     rec.task_id = task_id
-                    rec.save(update_fields=["task_id"])
+                    _db_retry(
+                        lambda r=rec: r.save(update_fields=["task_id"]),
+                        label=f"DVR recovery: recording {rec.id} task_id update",
+                    )
             except Exception as e:
                 logger.warning(f"Failed to schedule recording {rec.id}: {e}")
 
@@ -2782,10 +2871,13 @@ def _resolve_poster_for_program(channel_name, program):
         except Exception:
             pass
 
-    # Keyless providers (TVMaze & iTunes)
-    if not poster_url and not poster_logo_id:
+    # Keyless providers (TVMaze & iTunes) — only when real program title is 
+    # present. Searching by channel name returns unrelated art work from
+    # fuzzy API matching.
+    program_title = program.get('title') if isinstance(program, dict) else None
+    if not poster_url and not poster_logo_id and program_title:
         try:
-            title = (program.get('title') if isinstance(program, dict) else None) or channel_name
+            title = program_title
             if title:
                 # TVMaze
                 try:
@@ -2858,6 +2950,10 @@ def prefetch_recording_artwork(recording_id):
 
         program = cp.get("program") or {}
         poster_logo_id, poster_url = _resolve_poster_for_program(rec.channel.name, program)
+        # Fall back to the channel's own logo so the card always has an image
+        # rather than relying on the frontend's async channelsById lookup.
+        if not poster_logo_id and not poster_url and rec.channel.logo_id:
+            poster_logo_id = rec.channel.logo_id
         updated = False
         if poster_logo_id and cp.get("poster_logo_id") != poster_logo_id:
             cp["poster_logo_id"] = poster_logo_id
diff --git a/apps/channels/tests/test_dvr_retry.py b/apps/channels/tests/test_dvr_retry.py
new file mode 100644
index 00000000..7d6b3a97
--- /dev/null
+++ b/apps/channels/tests/test_dvr_retry.py
@@ -0,0 +1,294 @@
+"""Tests for DVR retry logic.
+
+Covers:
+  - _db_retry(): exponential backoff, max retries, connection reset
+  - Final metadata save retry in run_recording post-processing
+  - Initial TS proxy connection retry (per-base retry on retriable errors)
+  - recover_recordings_on_startup DB retry wrappers
+"""
+from datetime import timedelta
+from unittest.mock import MagicMock, patch, call
+
+from django.db import OperationalError
+from django.test import TestCase
+from django.utils import timezone
+
+from apps.channels.models import Channel, Recording
+from apps.channels.tasks import _db_retry
+
+
+# ---------------------------------------------------------------------------
+# _db_retry unit tests
+# ---------------------------------------------------------------------------
+
+class DbRetryTests(TestCase):
+    """Tests for the _db_retry() exponential backoff helper."""
+
+    @patch("apps.channels.tasks.time.sleep")
+    @patch("apps.channels.tasks.close_old_connections", create=True)
+    def test_succeeds_on_first_attempt(self, _close, _sleep):
+        """No retry needed when fn succeeds immediately."""
+        result = _db_retry(lambda: "ok", max_retries=3)
+        self.assertEqual(result, "ok")
+        _sleep.assert_not_called()
+
+    @patch("apps.channels.tasks.time.sleep")
+    @patch("django.db.close_old_connections")
+    def test_retries_on_operational_error_then_succeeds(self, mock_close, mock_sleep):
+        """Retry succeeds on second attempt after OperationalError."""
+        call_count = {"n": 0}
+
+        def flaky():
+            call_count["n"] += 1
+            if call_count["n"] == 1:
+                raise OperationalError("connection reset")
+            return "recovered"
+
+        result = _db_retry(flaky, max_retries=3, base_interval=1)
+        self.assertEqual(result, "recovered")
+        self.assertEqual(call_count["n"], 2)
+
+    @patch("apps.channels.tasks.time.sleep")
+    @patch("django.db.close_old_connections")
+    def test_raises_after_max_retries_exhausted(self, mock_close, mock_sleep):
+        """Raises OperationalError after all retries fail."""
+        def always_fail():
+            raise OperationalError("db gone")
+
+        with self.assertRaises(OperationalError):
+            _db_retry(always_fail, max_retries=3, base_interval=1)
+
+    @patch("apps.channels.tasks.time.sleep")
+    @patch("django.db.close_old_connections")
+    def test_exponential_backoff_timing(self, mock_close, mock_sleep):
+        """Sleep durations follow exponential backoff: 1s, 2s, 4s."""
+        call_count = {"n": 0}
+
+        def fail_twice():
+            call_count["n"] += 1
+            if call_count["n"] <= 2:
+                raise OperationalError("retry me")
+            return "done"
+
+        _db_retry(fail_twice, max_retries=3, base_interval=1)
+        mock_sleep.assert_has_calls([call(1), call(2)])
+
+    @patch("apps.channels.tasks.time.sleep")
+    @patch("django.db.close_old_connections")
+    def test_close_old_connections_called_between_retries(self, mock_close, mock_sleep):
+        """Stale DB connections are reset before each retry attempt."""
+        call_count = {"n": 0}
+
+        def fail_once():
+            call_count["n"] += 1
+            if call_count["n"] == 1:
+                raise OperationalError("stale conn")
+            return "ok"
+
+        _db_retry(fail_once, max_retries=3)
+        mock_close.assert_called_once()
+
+    @patch("apps.channels.tasks.time.sleep")
+    @patch("django.db.close_old_connections")
+    def test_non_operational_error_not_retried(self, mock_close, mock_sleep):
+        """Non-OperationalError exceptions propagate immediately."""
+        def raise_value_error():
+            raise ValueError("not a DB error")
+
+        with self.assertRaises(ValueError):
+            _db_retry(raise_value_error, max_retries=3)
+        mock_sleep.assert_not_called()
+
+    @patch("apps.channels.tasks.time.sleep")
+    @patch("django.db.close_old_connections")
+    def test_returns_fn_return_value(self, mock_close, mock_sleep):
+        """Return value of fn() is passed through."""
+        result = _db_retry(lambda: {"key": "value"}, max_retries=3)
+        self.assertEqual(result, {"key": "value"})
+
+    @patch("apps.channels.tasks.time.sleep")
+    @patch("django.db.close_old_connections")
+    def test_single_retry_allowed(self, mock_close, mock_sleep):
+        """max_retries=1 means no retry — fail immediately."""
+        with self.assertRaises(OperationalError):
+            _db_retry(
+                lambda: (_ for _ in ()).throw(OperationalError("fail")),
+                max_retries=1,
+            )
+        mock_sleep.assert_not_called()
+
+
+# ---------------------------------------------------------------------------
+# Final metadata save retry integration tests
+# ---------------------------------------------------------------------------
+
+class FinalMetadataSaveRetryTests(TestCase):
+    """The final recording metadata save must retry on transient DB errors."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(
+            channel_number=95, name="Retry Test Channel"
+        )
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_metadata_save_uses_db_retry(self, _ws):
+        """Verify recording metadata is saved via _db_retry (retries on OperationalError)."""
+        now = timezone.now()
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(hours=1),
+            end_time=now + timedelta(hours=1),
+            custom_properties={"status": "recording"},
+        )
+        # Directly call _db_retry to save metadata as run_recording does
+        cp = rec.custom_properties.copy()
+        cp["status"] = "completed"
+        cp["ended_at"] = str(now)
+        cp["bytes_written"] = 1024
+
+        def _save():
+            rec.custom_properties = cp
+            rec.save(update_fields=["custom_properties"])
+
+        _db_retry(_save, max_retries=3, base_interval=1, label="test save")
+        rec.refresh_from_db()
+        self.assertEqual(rec.custom_properties["status"], "completed")
+        self.assertEqual(rec.custom_properties["bytes_written"], 1024)
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_metadata_survives_transient_save_failure(self, _ws):
+        """Simulate OperationalError on first save, success on retry."""
+        now = timezone.now()
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(hours=1),
+            end_time=now + timedelta(hours=1),
+            custom_properties={"status": "recording"},
+        )
+        cp = {"status": "completed", "bytes_written": 2048}
+        call_count = {"n": 0}
+        _real_save = rec.save
+
+        def patched_save(**kwargs):
+            call_count["n"] += 1
+            if call_count["n"] == 1:
+                raise OperationalError("connection reset by peer")
+            return _real_save(**kwargs)
+
+        with patch.object(rec, "save", side_effect=patched_save):
+            with patch("apps.channels.tasks.time.sleep"):
+                with patch("django.db.close_old_connections"):
+                    def _save():
+                        rec.custom_properties = cp
+                        rec.save(update_fields=["custom_properties"])
+                    _db_retry(_save, max_retries=3, base_interval=1, label="test")
+
+        rec.refresh_from_db()
+        self.assertEqual(rec.custom_properties["status"], "completed")
+
+
+# ---------------------------------------------------------------------------
+# Initial connection retry tests
+# ---------------------------------------------------------------------------
+
+class InitialConnectionRetryTests(TestCase):
+    """When a candidate base URL fails with a retriable error before data flows,
+    the DVR task must retry the same base before moving to the next candidate."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(
+            channel_number=96, name="Connection Retry Channel"
+        )
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_retriable_error_increments_reconnect_counter(self, _ws):
+        """Verify the reconnection counter logic: same base retried before fallback."""
+        # This tests the logic structure directly rather than running the full task.
+        # The reconnection counter starts at 0, increments on each retriable error,
+        # and the loop 'continue's (retries same base) until max is exceeded.
+        max_reconnects = 5
+        reconnects = 0
+        attempts_on_base = 0
+
+        # Simulate the inner reconnection loop logic
+        while True:
+            attempts_on_base += 1
+            # Simulate retriable error with no data
+            reconnects += 1
+            if reconnects <= max_reconnects:
+                continue  # Would retry same base
+            break  # Would move to next base
+
+        self.assertEqual(attempts_on_base, max_reconnects + 1)
+        self.assertEqual(reconnects, max_reconnects + 1)
+
+
+# ---------------------------------------------------------------------------
+# recover_recordings_on_startup retry tests
+# ---------------------------------------------------------------------------
+
+class RecoveryRetryTests(TestCase):
+    """DB operations in recover_recordings_on_startup must use _db_retry."""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(
+            channel_number=97, name="Recovery Retry Channel"
+        )
+
+    @patch("apps.channels.tasks.run_recording.apply_async")
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_recovery_save_retries_on_operational_error(self, _ws, mock_async):
+        """Recovery status update uses _db_retry — survives one OperationalError."""
+        now = timezone.now()
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(minutes=30),
+            end_time=now + timedelta(minutes=30),
+            custom_properties={},
+        )
+        # Simulate what recovery does: mark interrupted, then save with retry
+        cp = rec.custom_properties or {}
+        cp["status"] = "interrupted"
+        cp["interrupted_reason"] = "server_restarted"
+        rec.custom_properties = cp
+
+        call_count = {"n": 0}
+        _real_save = Recording.save
+
+        def patched_save(self_rec, **kwargs):
+            call_count["n"] += 1
+            if call_count["n"] == 1:
+                raise OperationalError("db temporarily unavailable")
+            return _real_save(self_rec, **kwargs)
+
+        with patch.object(Recording, "save", patched_save):
+            with patch("apps.channels.tasks.time.sleep"):
+                with patch("django.db.close_old_connections"):
+                    _db_retry(
+                        lambda: rec.save(update_fields=["custom_properties"]),
+                        max_retries=3,
+                        label="test recovery",
+                    )
+
+        rec.refresh_from_db()
+        self.assertEqual(rec.custom_properties.get("status"), "interrupted")
+        self.assertEqual(rec.custom_properties.get("interrupted_reason"), "server_restarted")
+
+    def test_db_retry_fetches_recording_list(self):
+        """_db_retry correctly returns query results for recording list fetch."""
+        now = timezone.now()
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(minutes=30),
+            end_time=now + timedelta(minutes=30),
+            custom_properties={},
+        )
+        result = _db_retry(
+            lambda: list(Recording.objects.filter(
+                start_time__lte=now, end_time__gt=now
+            )),
+            label="test query",
+        )
+        self.assertGreaterEqual(len(result), 1)
+        ids = [r.id for r in result]
+        self.assertIn(rec.id, ids)
diff --git a/frontend/src/components/cards/RecordingCard.jsx b/frontend/src/components/cards/RecordingCard.jsx
index 2386748e..073ec542 100644
--- a/frontend/src/components/cards/RecordingCard.jsx
+++ b/frontend/src/components/cards/RecordingCard.jsx
@@ -28,6 +28,7 @@ import {
   deleteRecordingById,
   deleteSeriesAndRule,
   extendRecordingById,
+  getChannelLogoUrl,
   getPosterUrl,
   getRecordingUrl,
   getSeasonLabel,
@@ -64,7 +65,7 @@ const RecordingCard = ({
   const posterUrl = getPosterUrl(
     customProps.poster_logo_id,
     customProps,
-    channel?.logo?.cache_url,
+    getChannelLogoUrl(channel),
     env_mode
   );
 
diff --git a/frontend/src/components/forms/RecordingDetailsModal.jsx b/frontend/src/components/forms/RecordingDetailsModal.jsx
index bfd7abda..65599e24 100644
--- a/frontend/src/components/forms/RecordingDetailsModal.jsx
+++ b/frontend/src/components/forms/RecordingDetailsModal.jsx
@@ -20,6 +20,7 @@ import useVideoStore from '../../store/useVideoStore.jsx';
 import { notifications } from '@mantine/notifications';
 import {
   deleteRecordingById,
+  getChannelLogoUrl,
   getPosterUrl,
   getRecordingUrl,
   getSeasonLabel,
@@ -158,7 +159,7 @@ const RecordingDetailsModal = ({
         url: getPosterUrl(
           childRec.custom_properties?.poster_logo_id,
           undefined,
-          ch?.logo?.cache_url
+          getChannelLogoUrl(ch)
         ),
       },
     });
@@ -325,7 +326,7 @@ const RecordingDetailsModal = ({
             posterUrl={getPosterUrl(
               childRec.custom_properties?.poster_logo_id,
               childRec.custom_properties,
-              channelsById[childRec.channel]?.logo?.cache_url
+              getChannelLogoUrl(channelsById[childRec.channel])
             )}
             env_mode={env_mode}
             onWatchLive={handleOnWatchLive}
diff --git a/frontend/src/pages/DVR.jsx b/frontend/src/pages/DVR.jsx
index 60822dae..0c04a51c 100644
--- a/frontend/src/pages/DVR.jsx
+++ b/frontend/src/pages/DVR.jsx
@@ -24,6 +24,7 @@ import RecurringRuleModal from '../components/forms/RecurringRuleModal.jsx';
 import RecordingCard from '../components/cards/RecordingCard.jsx';
 import { categorizeRecordings } from '../utils/pages/DVRUtils.js';
 import {
+  getChannelLogoUrl,
   getPosterUrl,
   getRecordingUrl,
   getShowVideoUrl,
@@ -158,7 +159,7 @@ const DVRPage = () => {
         url: getPosterUrl(
           detailsRecording.custom_properties?.poster_logo_id,
           undefined,
-          channelsById[detailsRecording.channel]?.logo?.cache_url
+          getChannelLogoUrl(channelsById[detailsRecording.channel])
         ),
       },
     });
@@ -302,7 +303,7 @@ const DVRPage = () => {
               posterUrl={getPosterUrl(
                 detailsRecording.custom_properties?.poster_logo_id,
                 detailsRecording.custom_properties,
-                channelsById[detailsRecording.channel]?.logo?.cache_url
+                getChannelLogoUrl(channelsById[detailsRecording.channel])
               )}
               env_mode={useSettingsStore.getState().environment.env_mode}
               onWatchLive={handleOnWatchLive}
diff --git a/frontend/src/utils/cards/RecordingCardUtils.js b/frontend/src/utils/cards/RecordingCardUtils.js
index 3ca5703a..216494b8 100644
--- a/frontend/src/utils/cards/RecordingCardUtils.js
+++ b/frontend/src/utils/cards/RecordingCardUtils.js
@@ -19,6 +19,16 @@ export const removeRecording = (id) => {
   });
 };
 
+/**
+ * Resolve the channel logo cache URL from either a full channel object
+ * (has logo.cache_url) or a summary object (has logo_id integer).
+ */
+export const getChannelLogoUrl = (channel) => {
+  if (!channel) return null;
+  if (channel.logo_id) return `/api/channels/logos/${channel.logo_id}/cache/`;
+  return channel.logo?.cache_url || null;
+};
+
 export const getPosterUrl = (posterLogoId, customProperties, posterUrl) => {
   let purl = posterLogoId
     ? `/api/channels/logos/${posterLogoId}/cache/`

From 759688390bb1e0781ca24d424af21b262042b5de Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Mon, 2 Mar 2026 20:41:52 -0600
Subject: [PATCH 39/58] fix(dvr): fix series episode list, card layout, rule
 cleanup, and recording resilience
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Backend:
- Add terminal status guard to stop() endpoint -prevents overwriting completed/interrupted/failed with "stopped" (returns 409 Conflict)
- Include currently-airing programs in series rule evaluation by filtering on end_time instead of start_time
- Clean up future recordings when deleting a series rule while preserving previously recorded episodes
- Add logo proxy negative cache to prevent repeated requests to unreachable hosts from blocking Daphne workers
- Add URL validation with HEAD-then-GET fallback and per-worker cache
- Add EPG time-slot matching (80% overlap threshold) to enrich manual recordings with program metadata
- Add multi-source poster resolution pipeline (EPG, VOD, TMDB, OMDb, TVMaze, iTunes) with URL validation at each stage
- Add editable recording metadata endpoint with user_edited flag to prevent EPG auto-overwrite
- Add refresh-artwork endpoint to re-run poster resolution on demand
- Batch-fetch PeriodicTask names in schedule_upcoming to avoid N+1

Frontend:
- Fix series modal showing only one episode — preserve _group_count from categorized prop when merging with store version so the episode list renders correctly
- Fix recording card button positioning — use flex column layout with marginTop auto so buttons align at the bottom regardless of content
- Add delete fallback in RecurringRuleModal for orphaned recordings from deleted rules
- Use end_time filter in upcoming episodes list to include currently-airing episodes
- Add inline metadata editing (title/description) in details modal
- Add refresh artwork button in details modal
- Replace hardcoded /logo.png fallbacks with imported default logo
- Add React.memo with custom comparator to RecordingCard

Tests:
- Add 21 EPG time-slot matching tests (overlap threshold, edge cases)
- Add 16 recording metadata tests (validation, artwork, logo cache)
- Add 17 URL validation tests (HEAD/GET fallback, caching, eviction)
---
 apps/channels/api_views.py                    | 201 +++++-
 apps/channels/tasks.py                        | 608 +++++++++---------
 apps/channels/tests/test_epg_matching.py      | 180 ++++++
 .../channels/tests/test_recording_metadata.py | 379 +++++++++++
 apps/channels/tests/test_validate_url.py      | 169 +++++
 apps/proxy/ts_proxy/client_manager.py         |   2 +-
 frontend/package-lock.json                    |  18 -
 frontend/src/api.js                           |  24 +
 .../src/components/cards/RecordingCard.jsx    |  42 +-
 .../forms/RecordingDetailsModal.jsx           | 175 ++++-
 .../components/forms/RecurringRuleModal.jsx   |  42 +-
 frontend/src/pages/DVR.jsx                    |   5 +-
 frontend/src/pages/__tests__/DVR.test.jsx     |  14 +-
 .../src/utils/cards/RecordingCardUtils.js     |   7 +-
 .../__tests__/RecordingCardUtils.test.js      |   4 +-
 .../utils/forms/RecordingDetailsModalUtils.js |   5 +-
 16 files changed, 1495 insertions(+), 380 deletions(-)
 create mode 100644 apps/channels/tests/test_epg_matching.py
 create mode 100644 apps/channels/tests/test_recording_metadata.py
 create mode 100644 apps/channels/tests/test_validate_url.py

diff --git a/apps/channels/api_views.py b/apps/channels/api_views.py
index b2bc0e14..2d4e5077 100644
--- a/apps/channels/api_views.py
+++ b/apps/channels/api_views.py
@@ -11,7 +11,7 @@ from django.shortcuts import get_object_or_404, get_list_or_404
 from django.db import transaction
 from django.db.models import Count, F
 from django.db.models import Q
-import os, json, requests, logging, mimetypes, threading
+import os, json, requests, logging, mimetypes, threading, time
 from datetime import timedelta
 from django.utils.http import http_date
 from urllib.parse import unquote
@@ -73,6 +73,12 @@ from rest_framework.pagination import PageNumberPagination
 
 logger = logging.getLogger(__name__)
 
+# Negative cache for remote logo URLs that failed to fetch.
+# Prevents repeated blocking requests to unreachable hosts (e.g., dead CDNs)
+# from exhausting Daphne workers.  Keyed by URL, value is expiry timestamp.
+_logo_fetch_failures = {}
+_LOGO_FAIL_TTL = 300  # seconds
+
 
 class OrInFilter(django_filters.Filter):
     """
@@ -1957,6 +1963,12 @@ class LogoViewSet(viewsets.ModelViewSet):
             return response
 
         else:  # Remote image
+            # Skip URLs that recently failed to avoid blocking Daphne workers
+            # on unreachable hosts (e.g., dead CDNs referenced by old recordings).
+            fail_expiry = _logo_fetch_failures.get(logo_url)
+            if fail_expiry and time.monotonic() < fail_expiry:
+                raise Http404("Remote image temporarily unavailable")
+
             try:
                 # Get the default user agent
                 try:
@@ -1975,6 +1987,9 @@ class LogoViewSet(viewsets.ModelViewSet):
                     headers={'User-Agent': user_agent}
                 )
                 if remote_response.status_code == 200:
+                    # Success — clear any previous failure entry
+                    _logo_fetch_failures.pop(logo_url, None)
+
                     # Try to get content type from response headers first
                     content_type = remote_response.headers.get("Content-Type")
 
@@ -1998,14 +2013,19 @@ class LogoViewSet(viewsets.ModelViewSet):
                         os.path.basename(logo_url)
                     )
                     return response
+                # Non-200 response — cache the failure and evict stale entries
+                now = time.monotonic()
+                _logo_fetch_failures[logo_url] = now + _LOGO_FAIL_TTL
+                if len(_logo_fetch_failures) > 256:
+                    for k in [k for k, v in _logo_fetch_failures.items() if v <= now]:
+                        _logo_fetch_failures.pop(k, None)
                 raise Http404("Remote image not found")
-            except requests.exceptions.Timeout:
-                logger.warning(f"Timeout fetching logo from {logo_url}")
-                raise Http404("Logo request timed out")
-            except requests.exceptions.ConnectionError:
-                logger.warning(f"Connection error fetching logo from {logo_url}")
-                raise Http404("Unable to connect to logo server")
             except requests.RequestException as e:
+                now = time.monotonic()
+                _logo_fetch_failures[logo_url] = now + _LOGO_FAIL_TTL
+                if len(_logo_fetch_failures) > 256:
+                    for k in [k for k, v in _logo_fetch_failures.items() if v <= now]:
+                        _logo_fetch_failures.pop(k, None)
                 logger.warning(f"Error fetching logo from {logo_url}: {e}")
                 raise Http404("Error fetching remote image")
 
@@ -2396,10 +2416,22 @@ class RecordingViewSet(viewsets.ModelViewSet):
         """Stop a recording early while retaining the partial content for playback."""
         instance = self.get_object()
 
+        cp = instance.custom_properties or {}
+        current_status = cp.get("status", "")
+
+        # Reject stop on recordings that are already in a terminal state.
+        # Without this guard, stop() would overwrite "completed" or
+        # "interrupted" with "stopped", losing the original outcome.
+        terminal = {"completed", "interrupted", "failed"}
+        if current_status in terminal:
+            return Response(
+                {"success": False, "error": f"Recording is already {current_status}"},
+                status=status.HTTP_409_CONFLICT,
+            )
+
         # Mark as stopped in the DB first so run_recording detects it.
         # This is the only operation that MUST be synchronous — run_recording reads
         # the status field to decide whether the stream disconnection was deliberate.
-        cp = instance.custom_properties or {}
         cp["status"] = "stopped"
         cp["stopped_at"] = str(timezone.now())
         instance.custom_properties = cp
@@ -2442,7 +2474,6 @@ class RecordingViewSet(viewsets.ModelViewSet):
             except Exception as e:
                 logger.debug(f"Unable to revoke task for stopped recording: {e}")
 
-            # Close the DB connection opened by this thread so it isn't leaked
             try:
                 from django.db import connection as _conn
                 _conn.close()
@@ -2505,6 +2536,121 @@ class RecordingViewSet(viewsets.ModelViewSet):
 
         return Response({"success": True, "new_end_time": new_end_time.isoformat()})
 
+    @action(detail=True, methods=["post"], url_path="refresh-artwork")
+    def refresh_artwork(self, request, pk=None):
+        """Re-run the poster resolution pipeline for this recording.
+
+        Useful when a recording fell back to a channel logo or default logo
+        because external sources were temporarily unavailable.
+        """
+        instance = self.get_object()
+
+        def _background_refresh(rec_id):
+            try:
+                from .tasks import _resolve_poster_for_program
+                from .models import Recording
+                from core.utils import send_websocket_update
+                from django.db import close_old_connections
+
+                rec = Recording.objects.select_related("channel").get(id=rec_id)
+                cp = rec.custom_properties or {}
+                program = cp.get("program") or {}
+
+                poster_logo_id, poster_url = _resolve_poster_for_program(
+                    rec.channel.name, program, channel_logo_id=rec.channel.logo_id,
+                )
+
+                # Refresh and merge to avoid overwriting concurrent changes.
+                # Only upgrade — never replace a real poster with a channel logo fallback.
+                rec.refresh_from_db()
+                fresh_cp = rec.custom_properties or {}
+                updated = False
+                is_channel_logo_fallback = (
+                    poster_logo_id == rec.channel.logo_id
+                    and not poster_url
+                )
+                if program and program.get("id"):
+                    fresh_cp["program"] = program
+                    updated = True
+                if not is_channel_logo_fallback:
+                    if poster_logo_id and fresh_cp.get("poster_logo_id") != poster_logo_id:
+                        fresh_cp["poster_logo_id"] = poster_logo_id
+                        updated = True
+                    if poster_url and fresh_cp.get("poster_url") != poster_url:
+                        fresh_cp["poster_url"] = poster_url
+                        updated = True
+
+                if updated:
+                    rec.custom_properties = fresh_cp
+                    rec.save(update_fields=["custom_properties"])
+
+                send_websocket_update('updates', 'update', {
+                    "success": True,
+                    "type": "recording_updated",
+                    "recording_id": rec_id,
+                })
+            except Exception as e:
+                logger.debug(f"refresh-artwork background failed for {rec_id}: {e}")
+            finally:
+                close_old_connections()
+
+        t = threading.Thread(target=_background_refresh, args=(instance.id,), daemon=True)
+        t.start()
+
+        return Response({"success": True, "message": "Artwork refresh started"})
+
+    @action(detail=True, methods=["post"], url_path="update-metadata")
+    def update_metadata(self, request, pk=None):
+        """Update user-editable recording metadata (title, description).
+
+        Sets user_edited flag to prevent EPG auto-enrichment from overwriting
+        the user's changes on subsequent task runs.
+        """
+        instance = self.get_object()
+        title = request.data.get("title")
+        description = request.data.get("description")
+
+        if title is None and description is None:
+            return Response(
+                {"success": False, "error": "No fields to update"},
+                status=status.HTTP_400_BAD_REQUEST,
+            )
+
+        # Strip whitespace; treat blank strings as "no change"
+        clean_title = str(title).strip() if title is not None else None
+        clean_desc = str(description).strip() if description is not None else None
+
+        if not clean_title and not clean_desc:
+            return Response(
+                {"success": False, "error": "Title and description cannot be blank"},
+                status=status.HTTP_400_BAD_REQUEST,
+            )
+
+        cp = instance.custom_properties or {}
+        program = cp.get("program") or {}
+
+        if clean_title:
+            program["title"] = clean_title
+        if clean_desc:
+            program["description"] = clean_desc
+        program["user_edited"] = True
+
+        cp["program"] = program
+        instance.custom_properties = cp
+        instance.save(update_fields=["custom_properties"])
+
+        try:
+            from core.utils import send_websocket_update
+            send_websocket_update('updates', 'update', {
+                "success": True,
+                "type": "recording_updated",
+                "recording_id": instance.id,
+            })
+        except Exception:
+            pass
+
+        return Response({"success": True})
+
     def destroy(self, request, *args, **kwargs):
         """Delete the Recording and ensure any active DVR client connection is closed.
 
@@ -2581,7 +2727,6 @@ class RecordingViewSet(viewsets.ModelViewSet):
             _safe_remove(file_path)
             _safe_remove(temp_ts_path)
 
-            # Close the DB connection opened by this thread so it isn't leaked
             try:
                 from django.db import connection as _conn
                 _conn.close()
@@ -2718,16 +2863,44 @@ class DeleteSeriesRuleAPIView(APIView):
 
     @extend_schema(
         summary="Delete a series rule",
-        description="Remove a series recording rule by TVG ID. This does not remove already scheduled recordings.",
+        description="Remove a series recording rule by TVG ID and clean up future scheduled recordings.",
         parameters=[
             OpenApiParameter('tvg_id', str, OpenApiParameter.PATH, required=True, description='Channel TVG ID'),
         ],
     )
     def delete(self, request, tvg_id):
         tvg_id = unquote(str(tvg_id or ""))
-        rules = [r for r in CoreSettings.get_dvr_series_rules() if str(r.get("tvg_id")) != tvg_id]
-        CoreSettings.set_dvr_series_rules(rules)
-        return Response({"success": True, "rules": rules})
+
+        # Find the rule before removing to retain the title for cleanup
+        rules = CoreSettings.get_dvr_series_rules()
+        deleted_rule = next((r for r in rules if str(r.get("tvg_id")) == tvg_id), None)
+        remaining = [r for r in rules if str(r.get("tvg_id")) != tvg_id]
+        CoreSettings.set_dvr_series_rules(remaining)
+
+        # Delete only FUTURE recordings — preserve previously recorded episodes
+        removed = 0
+        if deleted_rule:
+            from .models import Recording
+            qs = Recording.objects.filter(
+                start_time__gte=timezone.now(),
+                custom_properties__program__tvg_id=tvg_id,
+            )
+            title = deleted_rule.get("title")
+            if title:
+                qs = qs.filter(custom_properties__program__title=title)
+            removed = qs.count()
+            qs.delete()
+
+        # Notify frontend to refresh recordings list
+        try:
+            from core.utils import send_websocket_update
+            send_websocket_update('updates', 'update', {
+                "success": True, "type": "recordings_refreshed", "removed": removed,
+            })
+        except Exception:
+            pass
+
+        return Response({"success": True, "rules": remaining, "removed": removed})
 
 
 class EvaluateSeriesRulesAPIView(APIView):
diff --git a/apps/channels/tasks.py b/apps/channels/tasks.py
index 4bbd082c..73e60441 100755
--- a/apps/channels/tasks.py
+++ b/apps/channels/tasks.py
@@ -20,24 +20,134 @@ from apps.channels.models import Channel
 from apps.epg.models import EPGData
 from core.models import CoreSettings
 
+from django.db import OperationalError, close_old_connections
 from channels.layers import get_channel_layer
 from asgiref.sync import async_to_sync
-
-from asgiref.sync import async_to_sync
-from channels.layers import get_channel_layer
 import tempfile
 from urllib.parse import quote
 
 logger = logging.getLogger(__name__)
 
 
+_url_validation_cache = {}
+_URL_CACHE_TTL = 300  # seconds
+
+
+def _validate_url(url, timeout=4):
+    """Validate that an HTTP(S) URL is reachable via HEAD request.
+    Returns True for non-HTTP URLs (skip validation) or 2xx/3xx responses.
+    Results are cached per-worker for 5 minutes to avoid redundant requests
+    when multiple recordings reference the same dead URL.
+    """
+    if not url or not isinstance(url, str):
+        return False
+    if not url.startswith(("http://", "https://")):
+        return True
+
+    now = time.monotonic()
+    cached = _url_validation_cache.get(url)
+    if cached is not None and (now - cached[1]) < _URL_CACHE_TTL:
+        return cached[0]
+
+    try:
+        resp = requests.head(url, timeout=timeout, allow_redirects=True)
+        if resp.status_code == 405:
+            # Server doesn't support HEAD; fall back to ranged GET
+            resp = requests.get(
+                url, timeout=timeout, allow_redirects=True,
+                headers={"Range": "bytes=0-0"}, stream=True,
+            )
+            resp.close()
+        result = resp.status_code < 400
+    except Exception:
+        result = False
+
+    _url_validation_cache[url] = (result, now)
+
+    # Evict expired entries when cache grows large
+    if len(_url_validation_cache) > 512:
+        cutoff = now - _URL_CACHE_TTL
+        expired = [k for k, v in _url_validation_cache.items() if v[1] < cutoff]
+        for k in expired:
+            del _url_validation_cache[k]
+
+    return result
+
+
+def _pick_best_image_from_epg_props(epg_props):
+    """Select the highest-quality poster/cover image from EPG custom_properties."""
+    try:
+        images = epg_props.get("images") or []
+        if not isinstance(images, list):
+            return None
+        size_order = {"xxl": 6, "xl": 5, "l": 4, "m": 3, "s": 2, "xs": 1}
+        def score(img):
+            t = (img.get("type") or "").lower()
+            size = (img.get("size") or "").lower()
+            return (2 if t in ("poster", "cover") else 1, size_order.get(size, 0))
+        best = None
+        for im in images:
+            if not isinstance(im, dict):
+                continue
+            url = im.get("url")
+            if not url:
+                continue
+            if best is None or score(im) > score(best):
+                best = im
+        return best.get("url") if best else None
+    except Exception:
+        return None
+
+
+def _match_epg_program_by_timeslot(channel_epg_data, rec_start, rec_end):
+    """Find an EPG program that covers at least 80% of the recording window.
+
+    Queries all programs overlapping the recording, calculates overlap for
+    each, and returns the best match only if it covers >= 80% of the
+    recording duration.  Recordings spanning multiple programs with no
+    dominant show return None (displayed as "Custom Recording").
+    Returns a dict with id, title, sub_title, and description, or None.
+    """
+    if not channel_epg_data or not rec_start or not rec_end:
+        return None
+    try:
+        candidates = channel_epg_data.programs.filter(
+            start_time__lt=rec_end,
+            end_time__gt=rec_start,
+        ).only("id", "title", "sub_title", "description", "start_time", "end_time")
+
+        rec_duration = (rec_end - rec_start).total_seconds()
+        if rec_duration <= 0:
+            return None
+
+        best = None
+        best_overlap = 0
+        for prog in candidates:
+            overlap_start = max(rec_start, prog.start_time)
+            overlap_end = min(rec_end, prog.end_time)
+            overlap = (overlap_end - overlap_start).total_seconds()
+            if overlap > best_overlap:
+                best_overlap = overlap
+                best = prog
+
+        if best and (best_overlap / rec_duration) >= 0.8:
+            return {
+                "id": best.id,
+                "title": best.title or "",
+                "sub_title": best.sub_title or "",
+                "description": best.description or "",
+            }
+    except Exception:
+        pass
+    return None
+
+
 def _db_retry(fn, max_retries=3, base_interval=1, label="DB operation"):
     """Execute fn() with exponential backoff retry on transient DB errors.
 
     Follows the same backoff pattern as RedisClient.get_client().
     Resets stale connections between attempts so the ORM reconnects.
     """
-    from django.db import OperationalError, close_old_connections
     for attempt in range(max_retries):
         try:
             return fn()
@@ -1006,7 +1116,7 @@ def evaluate_series_rules_impl(tvg_id: str | None = None):
 
         programs_qs = ProgramData.objects.filter(
                 epg=epg,
-                start_time__gte=now,
+                end_time__gt=now,
                 start_time__lte=horizon,
             )
         if series_title:
@@ -1016,7 +1126,7 @@ def evaluate_series_rules_impl(tvg_id: str | None = None):
         if series_title and not programs:
             all_progs = ProgramData.objects.filter(
                 epg=epg,
-                start_time__gte=now,
+                end_time__gt=now,
                 start_time__lte=horizon,
             ).only("id", "title", "start_time", "end_time", "custom_properties", "tvg_id")
             programs = [p for p in all_progs if normalize_name(p.title) == norm_series]
@@ -1608,229 +1718,26 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
 
         # Determine program info (may include id for deeper details)
         program = cp.get("program") or {}
+
+        # Enrich empty program dicts (manual recordings) from EPG time-slot data.
+        if isinstance(program, dict) and not program.get("user_edited") and not program.get("id") and not program.get("title"):
+            epg_match = _match_epg_program_by_timeslot(
+                channel.epg_data, recording_obj.start_time, recording_obj.end_time,
+            )
+            if epg_match:
+                program.update(epg_match)
+                cp["program"] = program
+
         final_path, temp_ts_path, filename = _build_output_paths(channel, program, start_time, end_time)
         cp["file_name"] = filename
         cp["file_path"] = final_path
         cp["_temp_file_path"] = temp_ts_path
 
-        # Resolve poster the same way VODs do:
-        # 1) Prefer image(s) from EPG Program custom_properties (images/icon)
-        # 2) Otherwise reuse an existing VOD logo matching title (Movie/Series)
-        # 3) Otherwise save any direct poster URL from provided program fields
-        program = (cp.get("program") or {}) if isinstance(cp, dict) else {}
-
-        def pick_best_image_from_epg_props(epg_props):
-            try:
-                images = epg_props.get("images") or []
-                if not isinstance(images, list):
-                    return None
-                # Prefer poster/cover and larger sizes
-                size_order = {"xxl": 6, "xl": 5, "l": 4, "m": 3, "s": 2, "xs": 1}
-                def score(img):
-                    t = (img.get("type") or "").lower()
-                    size = (img.get("size") or "").lower()
-                    return (
-                        2 if t in ("poster", "cover") else 1,
-                        size_order.get(size, 0)
-                    )
-                best = None
-                for im in images:
-                    if not isinstance(im, dict):
-                        continue
-                    url = im.get("url")
-                    if not url:
-                        continue
-                    if best is None or score(im) > score(best):
-                        best = im
-                return best.get("url") if best else None
-            except Exception:
-                return None
-
-        poster_logo_id = None
-        poster_url = None
-
-        # Try EPG Program custom_properties by ID
-        try:
-            from apps.epg.models import ProgramData
-            prog_id = program.get("id")
-            if prog_id:
-                epg_program = ProgramData.objects.filter(id=prog_id).only("custom_properties").first()
-                if epg_program and epg_program.custom_properties:
-                    epg_props = epg_program.custom_properties or {}
-                    poster_url = pick_best_image_from_epg_props(epg_props)
-                    if not poster_url:
-                        icon = epg_props.get("icon")
-                        if isinstance(icon, str) and icon:
-                            poster_url = icon
-        except Exception as e:
-            logger.debug(f"EPG image lookup failed: {e}")
-
-        # Fallback: reuse VOD Logo by matching title
-        if not poster_url and not poster_logo_id:
-            try:
-                from apps.vod.models import Movie, Series
-                title = program.get("title") or channel.name
-                vod_logo = None
-                movie = Movie.objects.filter(name__iexact=title).select_related("logo").first()
-                if movie and movie.logo:
-                    vod_logo = movie.logo
-                if not vod_logo:
-                    series = Series.objects.filter(name__iexact=title).select_related("logo").first()
-                    if series and series.logo:
-                        vod_logo = series.logo
-                if vod_logo:
-                    poster_logo_id = vod_logo.id
-            except Exception as e:
-                logger.debug(f"VOD logo fallback failed: {e}")
-
-        # External metadata lookups (TMDB/OMDb) when EPG/VOD didn't provide an image
-        if not poster_url and not poster_logo_id:
-            try:
-                tmdb_key = os.environ.get('TMDB_API_KEY')
-                omdb_key = os.environ.get('OMDB_API_KEY')
-                title = (program.get('title') or channel.name or '').strip()
-                year = None
-                imdb_id = None
-
-                # Try to derive year and imdb from EPG program custom_properties
-                try:
-                    from apps.epg.models import ProgramData
-                    prog_id = program.get('id')
-                    epg_program = ProgramData.objects.filter(id=prog_id).only('custom_properties').first() if prog_id else None
-                    if epg_program and epg_program.custom_properties:
-                        d = epg_program.custom_properties.get('date')
-                        if d and len(str(d)) >= 4:
-                            year = str(d)[:4]
-                        imdb_id = epg_program.custom_properties.get('imdb.com_id') or imdb_id
-                except Exception:
-                    pass
-
-                # TMDB: by IMDb ID
-                if not poster_url and tmdb_key and imdb_id:
-                    try:
-                        url = f"https://api.themoviedb.org/3/find/{quote(imdb_id)}?api_key={tmdb_key}&external_source=imdb_id"
-                        resp = requests.get(url, timeout=5)
-                        if resp.ok:
-                            data = resp.json() or {}
-                            picks = []
-                            for k in ('movie_results', 'tv_results', 'tv_episode_results', 'tv_season_results'):
-                                lst = data.get(k) or []
-                                picks.extend(lst)
-                            poster_path = None
-                            for item in picks:
-                                if item.get('poster_path'):
-                                    poster_path = item['poster_path']
-                                    break
-                            if poster_path:
-                                poster_url = f"https://image.tmdb.org/t/p/w780{poster_path}"
-                    except Exception:
-                        pass
-
-                # TMDB: by title (and year if available)
-                if not poster_url and tmdb_key and title:
-                    try:
-                        q = quote(title)
-                        extra = f"&year={year}" if year else ""
-                        url = f"https://api.themoviedb.org/3/search/multi?api_key={tmdb_key}&query={q}{extra}"
-                        resp = requests.get(url, timeout=5)
-                        if resp.ok:
-                            data = resp.json() or {}
-                            results = data.get('results') or []
-                            results.sort(key=lambda x: float(x.get('popularity') or 0), reverse=True)
-                            for item in results:
-                                if item.get('poster_path'):
-                                    poster_url = f"https://image.tmdb.org/t/p/w780{item['poster_path']}"
-                                    break
-                    except Exception:
-                        pass
-
-                # OMDb fallback
-                if not poster_url and omdb_key:
-                    try:
-                        if imdb_id:
-                            url = f"https://www.omdbapi.com/?apikey={omdb_key}&i={quote(imdb_id)}"
-                        elif title:
-                            yy = f"&y={year}" if year else ""
-                            url = f"https://www.omdbapi.com/?apikey={omdb_key}&t={quote(title)}{yy}"
-                        else:
-                            url = None
-                        if url:
-                            resp = requests.get(url, timeout=5)
-                            if resp.ok:
-                                data = resp.json() or {}
-                                p = data.get('Poster')
-                                if p and p != 'N/A':
-                                    poster_url = p
-                    except Exception:
-                        pass
-            except Exception as e:
-                logger.debug(f"External poster lookup failed: {e}")
-
-        # Keyless fallback providers (no API keys required) — only search when
-        # real program title is present; channel names return unrelated results.
-        _program_title = (program.get('title') or '').strip()
-        if not poster_url and not poster_logo_id and _program_title:
-            try:
-                title = _program_title
-                if title:
-                    # 1) TVMaze (TV shows) - singlesearch by title
-                    try:
-                        url = f"https://api.tvmaze.com/singlesearch/shows?q={quote(title)}"
-                        resp = requests.get(url, timeout=5)
-                        if resp.ok:
-                            data = resp.json() or {}
-                            img = (data.get('image') or {})
-                            p = img.get('original') or img.get('medium')
-                            if p:
-                                poster_url = p
-                    except Exception:
-                        pass
-
-                    # 2) iTunes Search API (movies or tv shows)
-                    if not poster_url:
-                        try:
-                            for media in ('movie', 'tvShow'):
-                                url = f"https://itunes.apple.com/search?term={quote(title)}&media={media}&limit=1"
-                                resp = requests.get(url, timeout=5)
-                                if resp.ok:
-                                    data = resp.json() or {}
-                                    results = data.get('results') or []
-                                    if results:
-                                        art = results[0].get('artworkUrl100')
-                                        if art:
-                                            # Scale up to 600x600 by convention
-                                            poster_url = art.replace('100x100', '600x600')
-                                            break
-                        except Exception:
-                            pass
-            except Exception as e:
-                logger.debug(f"Keyless poster lookup failed: {e}")
-
-        # Last: check direct fields on provided program object
-        if not poster_url and not poster_logo_id:
-            for key in ("poster", "cover", "cover_big", "image", "icon"):
-                val = program.get(key)
-                if isinstance(val, dict):
-                    candidate = val.get("url")
-                    if candidate:
-                        poster_url = candidate
-                        break
-                elif isinstance(val, str) and val:
-                    poster_url = val
-                    break
-
-        # Create or assign Logo
-        if not poster_logo_id and poster_url and len(poster_url) <= 1000:
-            try:
-                logo, _ = Logo.objects.get_or_create(url=poster_url, defaults={"name": program.get("title") or channel.name})
-                poster_logo_id = logo.id
-            except Exception as e:
-                logger.debug(f"Unable to persist poster to Logo: {e}")
-
-        # Fall back to the channel's own logo so the card always has an image
-        if not poster_logo_id and not poster_url and channel.logo_id:
-            poster_logo_id = channel.logo_id
-
+        # Resolve poster art via the shared pipeline (EPG → VOD → TMDB/OMDb →
+        # TVMaze/iTunes → direct program fields → Logo table → channel logo).
+        poster_logo_id, poster_url = _resolve_poster_for_program(
+            channel.name, program, channel_logo_id=channel.logo_id,
+        )
         if poster_logo_id:
             cp["poster_logo_id"] = poster_logo_id
         if poster_url and "poster_url" not in cp:
@@ -1860,7 +1767,7 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
         # Merge only the keys explicitly set into the fresh copy
         for key in ("status", "started_at", "file_url", "output_file_url",
                      "file_name", "file_path", "_temp_file_path",
-                     "poster_logo_id", "poster_url"):
+                     "program", "poster_logo_id", "poster_url"):
             if key in cp:
                 fresh_cp[key] = cp[key]
         recording_obj.custom_properties = fresh_cp
@@ -1933,6 +1840,7 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     _dvr_db_retry_interval = 1  # seconds (base for exponential backoff)
     # FFmpeg remux retry: covers transient I/O errors.
     _dvr_remux_max_retries = 2
+    _dvr_remux_retry_interval = 2  # seconds (base for exponential backoff)
 
     for base in candidates:
         test_url = f"{base.rstrip('/')}/proxy/ts/stream/{channel.uuid}"
@@ -2308,7 +2216,7 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
                         pass
             break  # Completed (success or deterministic failure)
 
-        except Exception as e:
+        except (OSError, subprocess.SubprocessError) as e:
             # Clean up partial output before potential retry
             try:
                 if os.path.exists(final_path):
@@ -2316,7 +2224,7 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
             except Exception:
                 pass
             if _remux_attempt + 1 < _dvr_remux_max_retries:
-                _wait = _dvr_db_retry_interval * (2 ** _remux_attempt)
+                _wait = _dvr_remux_retry_interval * (2 ** _remux_attempt)
                 logger.warning(
                     f"DVR recording {recording_id}: remux failed "
                     f"({type(e).__name__}), retrying in {_wait}s "
@@ -2422,8 +2330,6 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
             )
             return
 
-        recording_obj.custom_properties = cp
-
         def _save_final_metadata():
             recording_obj.custom_properties = cp
             recording_obj.save(update_fields=["custom_properties"])
@@ -2525,18 +2431,27 @@ def recover_recordings_on_startup():
         # Ensure future recordings are scheduled.
         # With ClockedSchedule, PeriodicTasks survive restarts in the DB.
         # Only recreate if the PeriodicTask is missing (safety net).
+        from django_celery_beat.models import PeriodicTask as _PT
+        from apps.channels.signals import _dvr_task_name
+
         upcoming = _db_retry(
             lambda: list(Recording.objects.filter(
                 start_time__gt=now, end_time__gt=now
             )),
             label="DVR recovery: fetching upcoming recordings",
         )
+
+        # Batch-fetch existing PeriodicTask names to avoid N+1 queries
+        task_names = {_dvr_task_name(r.id) for r in upcoming}
+        existing_tasks = set(_db_retry(
+            lambda: list(_PT.objects.filter(name__in=task_names).values_list("name", flat=True)),
+            label="DVR recovery: fetching existing periodic tasks",
+        )) if task_names else set()
+
         for rec in upcoming:
             try:
-                from django_celery_beat.models import PeriodicTask as _PT
-                from apps.channels.signals import _dvr_task_name
                 task_name = _dvr_task_name(rec.id)
-                if _PT.objects.filter(name=task_name).exists():
+                if task_name in existing_tasks:
                     if rec.task_id != task_name:
                         rec.task_id = task_name
                         _db_retry(
@@ -2809,14 +2724,23 @@ def comskip_process_recording(recording_id: int):
         _persist_custom_properties()
         _ws('error', {"reason": str(e)})
         return f"error:{e}"
-def _resolve_poster_for_program(channel_name, program):
-    """Internal helper that attempts to resolve a poster URL and/or Logo id.
+def _resolve_poster_for_program(channel_name, program, channel_logo_id=None):
+    """Resolve poster URL and/or Logo id for a recording program.
+
+    Callers should enrich the program dict via _match_epg_program_by_timeslot
+    before invoking this function so that EPG data is already available.
+
+    Pipeline: EPG images → VOD logo → TMDB/OMDb → TVMaze/iTunes →
+    direct program fields → Logo table → Logo creation → channel logo.
     Returns (poster_logo_id, poster_url) where either may be None.
     """
     poster_logo_id = None
     poster_url = None
+    epg_props = None
 
-    # Try EPG Program images first
+    _title = ((program.get("title") if isinstance(program, dict) else None) or "").strip() or None
+
+    # Stage 1: EPG Program images/icon (with URL validation)
     try:
         from apps.epg.models import ProgramData
         prog_id = program.get("id") if isinstance(program, dict) else None
@@ -2824,46 +2748,26 @@ def _resolve_poster_for_program(channel_name, program):
             epg_program = ProgramData.objects.filter(id=prog_id).only("custom_properties").first()
             if epg_program and epg_program.custom_properties:
                 epg_props = epg_program.custom_properties or {}
-
-                def pick_best_image_from_epg_props(epg_props):
-                    images = epg_props.get("images") or []
-                    if not isinstance(images, list):
-                        return None
-                    size_order = {"xxl": 6, "xl": 5, "l": 4, "m": 3, "s": 2, "xs": 1}
-                    def score(img):
-                        t = (img.get("type") or "").lower()
-                        size = (img.get("size") or "").lower()
-                        return (2 if t in ("poster", "cover") else 1, size_order.get(size, 0))
-                    best = None
-                    for im in images:
-                        if not isinstance(im, dict):
-                            continue
-                        url = im.get("url")
-                        if not url:
-                            continue
-                        if best is None or score(im) > score(best):
-                            best = im
-                    return best.get("url") if best else None
-
-                poster_url = pick_best_image_from_epg_props(epg_props)
+                poster_url = _pick_best_image_from_epg_props(epg_props)
+                if poster_url and not _validate_url(poster_url):
+                    poster_url = None
                 if not poster_url:
                     icon = epg_props.get("icon")
-                    if isinstance(icon, str) and icon:
+                    if isinstance(icon, str) and icon and _validate_url(icon):
                         poster_url = icon
     except Exception:
         pass
 
-    # VOD logo fallback by title
-    if not poster_url and not poster_logo_id:
+    # Stage 2: VOD logo fallback by title
+    if not poster_url and not poster_logo_id and _title:
         try:
             from apps.vod.models import Movie, Series
-            title = (program.get("title") if isinstance(program, dict) else None) or channel_name
             vod_logo = None
-            movie = Movie.objects.filter(name__iexact=title).select_related("logo").first()
+            movie = Movie.objects.filter(name__iexact=_title).select_related("logo").first()
             if movie and movie.logo:
                 vod_logo = movie.logo
             if not vod_logo:
-                series = Series.objects.filter(name__iexact=title).select_related("logo").first()
+                series = Series.objects.filter(name__iexact=_title).select_related("logo").first()
                 if series and series.logo:
                     vod_logo = series.logo
             if vod_logo:
@@ -2871,66 +2775,151 @@ def _resolve_poster_for_program(channel_name, program):
         except Exception:
             pass
 
-    # Keyless providers (TVMaze & iTunes) — only when real program title is 
-    # present. Searching by channel name returns unrelated art work from
-    # fuzzy API matching.
-    program_title = program.get('title') if isinstance(program, dict) else None
-    if not poster_url and not poster_logo_id and program_title:
+    # Stage 3: TMDB/OMDb (keyed APIs)
+    if not poster_url and not poster_logo_id and _title:
         try:
-            title = program_title
-            if title:
-                # TVMaze
+            tmdb_key = os.environ.get('TMDB_API_KEY')
+            omdb_key = os.environ.get('OMDB_API_KEY')
+            title = _title
+            year = None
+            imdb_id = None
+
+            # Derive year and imdb_id from cached EPG data
+            if epg_props:
+                d = epg_props.get('date')
+                if d and len(str(d)) >= 4:
+                    year = str(d)[:4]
+                imdb_id = epg_props.get('imdb.com_id')
+
+            # TMDB: by IMDb ID
+            if not poster_url and tmdb_key and imdb_id:
                 try:
-                    url = f"https://api.tvmaze.com/singlesearch/shows?q={quote(title)}"
+                    url = f"https://api.themoviedb.org/3/find/{quote(imdb_id)}?api_key={tmdb_key}&external_source=imdb_id"
                     resp = requests.get(url, timeout=5)
                     if resp.ok:
                         data = resp.json() or {}
-                        img = (data.get('image') or {})
-                        p = img.get('original') or img.get('medium')
-                        if p:
-                            poster_url = p
+                        picks = []
+                        for k in ('movie_results', 'tv_results', 'tv_episode_results', 'tv_season_results'):
+                            picks.extend(data.get(k) or [])
+                        for item in picks:
+                            if item.get('poster_path'):
+                                poster_url = f"https://image.tmdb.org/t/p/w780{item['poster_path']}"
+                                break
+                except Exception:
+                    pass
+
+            # TMDB: by title (and year if available)
+            if not poster_url and tmdb_key and title:
+                try:
+                    q = quote(title)
+                    extra = f"&year={year}" if year else ""
+                    url = f"https://api.themoviedb.org/3/search/multi?api_key={tmdb_key}&query={q}{extra}"
+                    resp = requests.get(url, timeout=5)
+                    if resp.ok:
+                        data = resp.json() or {}
+                        results = data.get('results') or []
+                        results.sort(key=lambda x: float(x.get('popularity') or 0), reverse=True)
+                        for item in results:
+                            if item.get('poster_path'):
+                                poster_url = f"https://image.tmdb.org/t/p/w780{item['poster_path']}"
+                                break
+                except Exception:
+                    pass
+
+            # OMDb fallback
+            if not poster_url and omdb_key:
+                try:
+                    if imdb_id:
+                        url = f"https://www.omdbapi.com/?apikey={omdb_key}&i={quote(imdb_id)}"
+                    elif title:
+                        yy = f"&y={year}" if year else ""
+                        url = f"https://www.omdbapi.com/?apikey={omdb_key}&t={quote(title)}{yy}"
+                    else:
+                        url = None
+                    if url:
+                        resp = requests.get(url, timeout=5)
+                        if resp.ok:
+                            data = resp.json() or {}
+                            p = data.get('Poster')
+                            if p and p != 'N/A':
+                                poster_url = p
                 except Exception:
                     pass
-                # iTunes
-                if not poster_url:
-                    try:
-                        for media in ('movie', 'tvShow'):
-                            url = f"https://itunes.apple.com/search?term={quote(title)}&media={media}&limit=1"
-                            resp = requests.get(url, timeout=5)
-                            if resp.ok:
-                                data = resp.json() or {}
-                                results = data.get('results') or []
-                                if results:
-                                    art = results[0].get('artworkUrl100')
-                                    if art:
-                                        poster_url = art.replace('100x100', '600x600')
-                                        break
-                    except Exception:
-                        pass
         except Exception:
             pass
 
-    # Fallback: search existing Logo entries by name if we still have nothing
-    if not poster_logo_id and not poster_url:
+    # Stage 4: Keyless providers (TVMaze & iTunes)
+    if not poster_url and not poster_logo_id and _title:
+        try:
+            title = _title
+            # TVMaze
+            try:
+                url = f"https://api.tvmaze.com/singlesearch/shows?q={quote(title)}"
+                resp = requests.get(url, timeout=5)
+                if resp.ok:
+                    data = resp.json() or {}
+                    img = (data.get('image') or {})
+                    p = img.get('original') or img.get('medium')
+                    if p:
+                        poster_url = p
+            except Exception:
+                pass
+            # iTunes
+            if not poster_url:
+                try:
+                    for media in ('movie', 'tvShow'):
+                        url = f"https://itunes.apple.com/search?term={quote(title)}&media={media}&limit=1"
+                        resp = requests.get(url, timeout=5)
+                        if resp.ok:
+                            data = resp.json() or {}
+                            results = data.get('results') or []
+                            if results:
+                                art = results[0].get('artworkUrl100')
+                                if art:
+                                    poster_url = art.replace('100x100', '600x600')
+                                    break
+                except Exception:
+                    pass
+        except Exception:
+            pass
+
+    # Stage 5: Direct fields on program object (with URL validation)
+    if not poster_url and not poster_logo_id and isinstance(program, dict):
+        for key in ("poster", "cover", "cover_big", "image", "icon"):
+            val = program.get(key)
+            if isinstance(val, dict):
+                candidate = val.get("url")
+                if candidate and _validate_url(candidate):
+                    poster_url = candidate
+                    break
+            elif isinstance(val, str) and val and _validate_url(val):
+                poster_url = val
+                break
+
+    # Stage 6: Search existing Logo entries by program title
+    if not poster_logo_id and not poster_url and _title:
         try:
             from .models import Logo
-            title = (program.get("title") if isinstance(program, dict) else None) or channel_name
-            existing = Logo.objects.filter(name__iexact=title).first()
+            existing = Logo.objects.filter(name__iexact=_title).first()
             if existing:
                 poster_logo_id = existing.id
                 poster_url = existing.url
         except Exception:
             pass
 
-    # Save to Logo if URL available
+    # Stage 7: Persist to Logo table if URL available
     if not poster_logo_id and poster_url and len(poster_url) <= 1000:
         try:
             from .models import Logo
-            logo, _ = Logo.objects.get_or_create(url=poster_url, defaults={"name": (program.get("title") if isinstance(program, dict) else None) or channel_name})
+            logo, _ = Logo.objects.get_or_create(url=poster_url, defaults={"name": _title or channel_name})
             poster_logo_id = logo.id
         except Exception:
             pass
 
+    # Stage 8: Fall back to channel logo
+    if not poster_logo_id and not poster_url and channel_logo_id:
+        poster_logo_id = channel_logo_id
+
     return poster_logo_id, poster_url
 
 
@@ -2949,11 +2938,20 @@ def prefetch_recording_artwork(recording_id):
             return "skipped: status is " + current_status
 
         program = cp.get("program") or {}
-        poster_logo_id, poster_url = _resolve_poster_for_program(rec.channel.name, program)
-        # Fall back to the channel's own logo so the card always has an image
-        # rather than relying on the frontend's async channelsById lookup.
-        if not poster_logo_id and not poster_url and rec.channel.logo_id:
-            poster_logo_id = rec.channel.logo_id
+
+        # Enrich empty program dicts (manual recordings) from EPG time-slot data.
+        # Persists matched title/description for display in the recording card.
+        if isinstance(program, dict) and not program.get("user_edited") and not program.get("id") and not program.get("title"):
+            epg_match = _match_epg_program_by_timeslot(
+                rec.channel.epg_data, rec.start_time, rec.end_time,
+            )
+            if epg_match:
+                program.update(epg_match)
+                cp["program"] = program
+
+        poster_logo_id, poster_url = _resolve_poster_for_program(
+            rec.channel.name, program, channel_logo_id=rec.channel.logo_id,
+        )
         updated = False
         if poster_logo_id and cp.get("poster_logo_id") != poster_logo_id:
             cp["poster_logo_id"] = poster_logo_id
@@ -2996,8 +2994,8 @@ def prefetch_recording_artwork(recording_id):
             # the stop endpoint or run_recording's final metadata save.
             rec.refresh_from_db()
             fresh_cp = rec.custom_properties or {}
-            for key in ("poster_logo_id", "poster_url", "rating", "rating_system",
-                        "season", "episode", "onscreen_episode"):
+            for key in ("program", "poster_logo_id", "poster_url", "rating",
+                        "rating_system", "season", "episode", "onscreen_episode"):
                 if key in cp:
                     fresh_cp[key] = cp[key]
             rec.custom_properties = fresh_cp
diff --git a/apps/channels/tests/test_epg_matching.py b/apps/channels/tests/test_epg_matching.py
new file mode 100644
index 00000000..fee7eeab
--- /dev/null
+++ b/apps/channels/tests/test_epg_matching.py
@@ -0,0 +1,180 @@
+"""Tests for the _match_epg_program_by_timeslot() helper in tasks.py.
+
+Covers:
+  - Exact time-slot match returns program dict
+  - 80% overlap threshold: at boundary, above, and below
+  - Multiple overlapping programs: dominant vs. evenly split
+  - Edge cases: None inputs, zero-duration recording, no EPG data
+  - Returned dict structure (id, title, sub_title, description)
+"""
+from datetime import timedelta
+
+from django.test import TestCase
+from django.utils import timezone
+
+from apps.channels.models import Channel
+from apps.epg.models import EPGSource, EPGData, ProgramData
+from apps.channels.tasks import _match_epg_program_by_timeslot
+
+
+class EpgMatchingSetupMixin:
+    """Shared setup for EPG matching tests."""
+
+    def setUp(self):
+        self.source = EPGSource.objects.create(name="Test Source")
+        self.epg = EPGData.objects.create(
+            tvg_id="test.channel", name="Test Channel EPG", epg_source=self.source,
+        )
+        self.channel = Channel.objects.create(
+            channel_number=50, name="EPG Match Channel", epg_data=self.epg,
+        )
+        self.base = timezone.now().replace(second=0, microsecond=0)
+
+    def _prog(self, offset_min, duration_min, title="Test Show", **kwargs):
+        """Create a ProgramData starting offset_min from self.base."""
+        start = self.base + timedelta(minutes=offset_min)
+        end = start + timedelta(minutes=duration_min)
+        return ProgramData.objects.create(
+            epg=self.epg, start_time=start, end_time=end, title=title, **kwargs,
+        )
+
+
+class ExactMatchTests(EpgMatchingSetupMixin, TestCase):
+    """Recording window exactly matches an EPG program."""
+
+    def test_exact_match_returns_program_dict(self):
+        prog = self._prog(0, 60, title="News at 9", sub_title="Top Stories",
+                          description="Evening news broadcast")
+        result = _match_epg_program_by_timeslot(
+            self.epg, prog.start_time, prog.end_time,
+        )
+        self.assertIsNotNone(result)
+        self.assertEqual(result["id"], prog.id)
+        self.assertEqual(result["title"], "News at 9")
+        self.assertEqual(result["sub_title"], "Top Stories")
+        self.assertEqual(result["description"], "Evening news broadcast")
+
+    def test_missing_optional_fields_returned_as_empty_strings(self):
+        prog = self._prog(0, 30, title="Minimal Show")
+        result = _match_epg_program_by_timeslot(
+            self.epg, prog.start_time, prog.end_time,
+        )
+        self.assertIsNotNone(result)
+        self.assertEqual(result["sub_title"], "")
+        self.assertEqual(result["description"], "")
+
+
+class OverlapThresholdTests(EpgMatchingSetupMixin, TestCase):
+    """80% overlap threshold boundary tests."""
+
+    def test_exactly_80_percent_overlap_returns_match(self):
+        """Program covers exactly 80% of the recording window."""
+        # Program: 0-60min, Recording: 0-75min → overlap = 60/75 = 80%
+        prog = self._prog(0, 60, title="Borderline Show")
+        rec_start = self.base
+        rec_end = self.base + timedelta(minutes=75)
+        result = _match_epg_program_by_timeslot(self.epg, rec_start, rec_end)
+        self.assertIsNotNone(result)
+        self.assertEqual(result["title"], "Borderline Show")
+
+    def test_below_80_percent_returns_none(self):
+        """Program covers 79% of the recording — below threshold."""
+        # Program: 0-60min, Recording: 0-76min → overlap = 60/76 ≈ 78.9%
+        prog = self._prog(0, 60, title="Too Short")
+        rec_start = self.base
+        rec_end = self.base + timedelta(minutes=76)
+        result = _match_epg_program_by_timeslot(self.epg, rec_start, rec_end)
+        self.assertIsNone(result)
+
+    def test_above_80_percent_returns_match(self):
+        """Program covers 90% of the recording."""
+        # Program: 0-60min, Recording: 0-66min → overlap = 60/66 ≈ 90.9%
+        prog = self._prog(0, 60, title="Good Match")
+        rec_start = self.base
+        rec_end = self.base + timedelta(minutes=66)
+        result = _match_epg_program_by_timeslot(self.epg, rec_start, rec_end)
+        self.assertIsNotNone(result)
+        self.assertEqual(result["title"], "Good Match")
+
+
+class MultipleProgramTests(EpgMatchingSetupMixin, TestCase):
+    """Recording spans multiple EPG programs."""
+
+    def test_dominant_program_returned(self):
+        """Recording spans 2 programs; one covers 85%, the other 15%."""
+        # Show A: 0-60min, Show B: 60-120min
+        # Recording: 9-69min → A overlap=51/60=85%, B overlap=9/60=15%
+        self._prog(0, 60, title="Show A")
+        self._prog(60, 60, title="Show B")
+        rec_start = self.base + timedelta(minutes=9)
+        rec_end = self.base + timedelta(minutes=69)
+        result = _match_epg_program_by_timeslot(self.epg, rec_start, rec_end)
+        self.assertIsNotNone(result)
+        self.assertEqual(result["title"], "Show A")
+
+    def test_evenly_split_returns_none(self):
+        """Recording spans 2 equal programs — neither reaches 80%."""
+        # Show A: 0-60min, Show B: 60-120min
+        # Recording: 30-90min → each covers 50%
+        self._prog(0, 60, title="Show A")
+        self._prog(60, 60, title="Show B")
+        rec_start = self.base + timedelta(minutes=30)
+        rec_end = self.base + timedelta(minutes=90)
+        result = _match_epg_program_by_timeslot(self.epg, rec_start, rec_end)
+        self.assertIsNone(result)
+
+    def test_three_programs_one_dominant(self):
+        """Recording spans 3 programs; middle one is dominant."""
+        # A: 0-30min, B: 30-90min, C: 90-120min
+        # Recording: 25-95min (70min window) → B overlap=60/70≈85.7%
+        self._prog(0, 30, title="Show A")
+        self._prog(30, 60, title="Show B")
+        self._prog(90, 30, title="Show C")
+        rec_start = self.base + timedelta(minutes=25)
+        rec_end = self.base + timedelta(minutes=95)
+        result = _match_epg_program_by_timeslot(self.epg, rec_start, rec_end)
+        self.assertIsNotNone(result)
+        self.assertEqual(result["title"], "Show B")
+
+
+class EdgeCaseTests(EpgMatchingSetupMixin, TestCase):
+    """Edge cases and error handling."""
+
+    def test_none_epg_data_returns_none(self):
+        result = _match_epg_program_by_timeslot(None, self.base, self.base + timedelta(hours=1))
+        self.assertIsNone(result)
+
+    def test_none_start_time_returns_none(self):
+        result = _match_epg_program_by_timeslot(self.epg, None, self.base + timedelta(hours=1))
+        self.assertIsNone(result)
+
+    def test_none_end_time_returns_none(self):
+        result = _match_epg_program_by_timeslot(self.epg, self.base, None)
+        self.assertIsNone(result)
+
+    def test_zero_duration_returns_none(self):
+        """Recording with start == end should return None."""
+        result = _match_epg_program_by_timeslot(self.epg, self.base, self.base)
+        self.assertIsNone(result)
+
+    def test_negative_duration_returns_none(self):
+        """Recording with end before start should return None."""
+        result = _match_epg_program_by_timeslot(
+            self.epg, self.base + timedelta(hours=1), self.base,
+        )
+        self.assertIsNone(result)
+
+    def test_no_overlapping_programs_returns_none(self):
+        """No EPG programs in the recording window."""
+        self._prog(0, 60, title="Earlier Show")
+        rec_start = self.base + timedelta(hours=5)
+        rec_end = rec_start + timedelta(hours=1)
+        result = _match_epg_program_by_timeslot(self.epg, rec_start, rec_end)
+        self.assertIsNone(result)
+
+    def test_empty_epg_no_programs_returns_none(self):
+        """EPGData exists but has no programs."""
+        result = _match_epg_program_by_timeslot(
+            self.epg, self.base, self.base + timedelta(hours=1),
+        )
+        self.assertIsNone(result)
diff --git a/apps/channels/tests/test_recording_metadata.py b/apps/channels/tests/test_recording_metadata.py
new file mode 100644
index 00000000..d4db6a2d
--- /dev/null
+++ b/apps/channels/tests/test_recording_metadata.py
@@ -0,0 +1,379 @@
+"""Tests for recording metadata endpoints and logo proxy negative cache.
+
+Covers:
+  - update_metadata endpoint: title/description, user_edited flag, validation
+  - refresh_artwork endpoint: returns immediately, background thread behavior
+  - Logo proxy negative cache: cache hit/miss, expiry, eviction, success clears
+"""
+import time as time_mod
+from datetime import timedelta
+from unittest.mock import MagicMock, patch
+
+from django.test import TestCase
+from django.utils import timezone
+from rest_framework.test import APIRequestFactory, force_authenticate
+
+from apps.channels.models import Channel, Recording, Logo
+from apps.channels.api_views import RecordingViewSet, LogoViewSet
+
+
+# ---------------------------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------------------------
+
+def _make_admin():
+    from django.contrib.auth import get_user_model
+    User = get_user_model()
+    u, _ = User.objects.get_or_create(
+        username="metadata_test_admin",
+        defaults={"user_level": User.UserLevel.ADMIN},
+    )
+    u.set_password("pass")
+    u.save()
+    return u
+
+
+# ---------------------------------------------------------------------------
+# update_metadata endpoint
+# ---------------------------------------------------------------------------
+
+class UpdateMetadataTests(TestCase):
+    """Tests for POST /api/channels/recordings/{id}/update-metadata/"""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(channel_number=70, name="Meta Test Channel")
+        self.user = _make_admin()
+        self.factory = APIRequestFactory()
+
+    def _update(self, rec, data):
+        request = self.factory.post(
+            f"/api/channels/recordings/{rec.id}/update-metadata/",
+            data, format="json",
+        )
+        force_authenticate(request, user=self.user)
+        view = RecordingViewSet.as_view({"post": "update_metadata"})
+        return view(request, pk=rec.id)
+
+    def _make_rec(self, custom_properties=None):
+        now = timezone.now()
+        return Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(hours=1),
+            end_time=now + timedelta(hours=1),
+            custom_properties=custom_properties or {},
+        )
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_update_title_only(self, _ws):
+        rec = self._make_rec()
+        response = self._update(rec, {"title": "My Show"})
+        self.assertEqual(response.status_code, 200)
+        rec.refresh_from_db()
+        program = rec.custom_properties["program"]
+        self.assertEqual(program["title"], "My Show")
+        self.assertTrue(program["user_edited"])
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_update_description_only(self, _ws):
+        rec = self._make_rec({"program": {"title": "Existing Title"}})
+        response = self._update(rec, {"description": "A great episode"})
+        self.assertEqual(response.status_code, 200)
+        rec.refresh_from_db()
+        program = rec.custom_properties["program"]
+        self.assertEqual(program["description"], "A great episode")
+        self.assertEqual(program["title"], "Existing Title")
+        self.assertTrue(program["user_edited"])
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_update_both_fields(self, _ws):
+        rec = self._make_rec()
+        response = self._update(rec, {"title": "New Title", "description": "New Desc"})
+        self.assertEqual(response.status_code, 200)
+        rec.refresh_from_db()
+        program = rec.custom_properties["program"]
+        self.assertEqual(program["title"], "New Title")
+        self.assertEqual(program["description"], "New Desc")
+        self.assertTrue(program["user_edited"])
+
+    def test_no_fields_returns_400(self):
+        rec = self._make_rec()
+        response = self._update(rec, {})
+        self.assertEqual(response.status_code, 400)
+        self.assertFalse(response.data.get("success"))
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_whitespace_trimmed(self, _ws):
+        rec = self._make_rec()
+        response = self._update(rec, {"title": "  Padded Title  "})
+        self.assertEqual(response.status_code, 200)
+        rec.refresh_from_db()
+        self.assertEqual(rec.custom_properties["program"]["title"], "Padded Title")
+
+    def test_whitespace_only_title_returns_400(self):
+        """Whitespace-only title and description should be rejected."""
+        rec = self._make_rec()
+        response = self._update(rec, {"title": "   ", "description": "   "})
+        self.assertEqual(response.status_code, 400)
+        self.assertFalse(response.data.get("success"))
+
+    def test_whitespace_only_title_with_valid_description(self):
+        """Whitespace-only title is ignored; valid description is accepted."""
+        rec = self._make_rec({"program": {"title": "Original"}})
+        with patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None):
+            response = self._update(rec, {"title": "   ", "description": "Valid desc"})
+        self.assertEqual(response.status_code, 200)
+        rec.refresh_from_db()
+        # Title should remain unchanged since the whitespace-only value is not applied
+        self.assertEqual(rec.custom_properties["program"]["title"], "Original")
+        self.assertEqual(rec.custom_properties["program"]["description"], "Valid desc")
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_creates_program_dict_when_absent(self, _ws):
+        """Recording with no program dict gets one created."""
+        rec = self._make_rec({"status": "completed"})
+        response = self._update(rec, {"title": "Brand New"})
+        self.assertEqual(response.status_code, 200)
+        rec.refresh_from_db()
+        self.assertIn("program", rec.custom_properties)
+        self.assertEqual(rec.custom_properties["program"]["title"], "Brand New")
+
+    def test_returns_404_for_nonexistent(self):
+        request = self.factory.post(
+            "/api/channels/recordings/99999/update-metadata/",
+            {"title": "Ghost"}, format="json",
+        )
+        force_authenticate(request, user=self.user)
+        view = RecordingViewSet.as_view({"post": "update_metadata"})
+        self.assertEqual(view(request, pk=99999).status_code, 404)
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_sends_websocket_event(self, mock_ws):
+        rec = self._make_rec()
+        self._update(rec, {"title": "WS Test"})
+        mock_ws.assert_called_once()
+        payload = mock_ws.call_args[0][2]
+        self.assertEqual(payload["type"], "recording_updated")
+        self.assertEqual(payload["recording_id"], rec.id)
+
+    @patch("core.utils.send_websocket_update", side_effect=Exception("WS down"))
+    def test_ws_failure_does_not_fail_request(self, _ws):
+        """WebSocket errors are silenced — the save still succeeds."""
+        rec = self._make_rec()
+        response = self._update(rec, {"title": "Resilient"})
+        self.assertEqual(response.status_code, 200)
+        rec.refresh_from_db()
+        self.assertEqual(rec.custom_properties["program"]["title"], "Resilient")
+
+
+# ---------------------------------------------------------------------------
+# refresh_artwork endpoint
+# ---------------------------------------------------------------------------
+
+class RefreshArtworkTests(TestCase):
+    """Tests for POST /api/channels/recordings/{id}/refresh-artwork/"""
+
+    def setUp(self):
+        self.channel = Channel.objects.create(channel_number=71, name="Artwork Test Channel")
+        self.user = _make_admin()
+        self.factory = APIRequestFactory()
+
+    def _refresh(self, rec):
+        request = self.factory.post(f"/api/channels/recordings/{rec.id}/refresh-artwork/")
+        force_authenticate(request, user=self.user)
+        view = RecordingViewSet.as_view({"post": "refresh_artwork"})
+        return view(request, pk=rec.id)
+
+    def _make_rec(self, custom_properties=None):
+        now = timezone.now()
+        return Recording.objects.create(
+            channel=self.channel,
+            start_time=now - timedelta(hours=1),
+            end_time=now + timedelta(hours=1),
+            custom_properties=custom_properties or {},
+        )
+
+    @patch("threading.Thread")
+    def test_returns_200_immediately(self, mock_thread):
+        mock_thread.return_value.start = MagicMock()
+        rec = self._make_rec()
+        response = self._refresh(rec)
+        self.assertEqual(response.status_code, 200)
+        self.assertTrue(response.data.get("success"))
+
+    @patch("threading.Thread")
+    def test_spawns_background_thread(self, mock_thread):
+        mock_thread.return_value.start = MagicMock()
+        rec = self._make_rec()
+        self._refresh(rec)
+        mock_thread.assert_called_once()
+        self.assertTrue(mock_thread.call_args[1].get("daemon", False))
+        mock_thread.return_value.start.assert_called_once()
+
+    def test_returns_404_for_nonexistent(self):
+        request = self.factory.post("/api/channels/recordings/99999/refresh-artwork/")
+        force_authenticate(request, user=self.user)
+        view = RecordingViewSet.as_view({"post": "refresh_artwork"})
+        self.assertEqual(view(request, pk=99999).status_code, 404)
+
+    @patch("django.db.close_old_connections")
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_no_downgrade_to_channel_logo(self, _ws, _close):
+        """When the pipeline returns the channel's own logo, existing poster is preserved."""
+        logo = Logo.objects.create(name="Channel Logo", url="https://example.com/ch.png")
+        self.channel.logo = logo
+        self.channel.save()
+        rec = self._make_rec({
+            "poster_logo_id": 999,  # existing real poster
+            "poster_url": "https://tmdb.com/real-poster.jpg",
+        })
+
+        with patch("apps.channels.tasks._resolve_poster_for_program",
+                    return_value=(logo.id, None)):
+            request = self.factory.post(f"/api/channels/recordings/{rec.id}/refresh-artwork/")
+            force_authenticate(request, user=self.user)
+
+            # Run synchronously by intercepting the thread
+            captured_fn = None
+            def capture_thread(*args, **kwargs):
+                nonlocal captured_fn
+                captured_fn = kwargs.get("target") or args[0]
+                mock = MagicMock()
+                mock.start = lambda: captured_fn(*kwargs.get("args", ()))
+                return mock
+
+            with patch("threading.Thread", side_effect=capture_thread):
+                view = RecordingViewSet.as_view({"post": "refresh_artwork"})
+                view(request, pk=rec.id)
+
+        rec.refresh_from_db()
+        # Existing poster should be preserved — not downgraded to channel logo
+        self.assertEqual(rec.custom_properties.get("poster_logo_id"), 999)
+        self.assertEqual(rec.custom_properties.get("poster_url"), "https://tmdb.com/real-poster.jpg")
+
+    @patch("django.db.close_old_connections")
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_upgrade_from_no_poster(self, _ws, _close):
+        """When a recording has no poster and the pipeline finds one, it gets updated."""
+        rec = self._make_rec({"program": {"title": "Some Show", "id": 42}})
+
+        with patch("apps.channels.tasks._resolve_poster_for_program",
+                    return_value=(555, "https://tmdb.com/new-poster.jpg")):
+            captured_fn = None
+            def capture_thread(*args, **kwargs):
+                nonlocal captured_fn
+                captured_fn = kwargs.get("target") or args[0]
+                mock = MagicMock()
+                mock.start = lambda: captured_fn(*kwargs.get("args", ()))
+                return mock
+
+            with patch("threading.Thread", side_effect=capture_thread):
+                request = self.factory.post(f"/api/channels/recordings/{rec.id}/refresh-artwork/")
+                force_authenticate(request, user=self.user)
+                view = RecordingViewSet.as_view({"post": "refresh_artwork"})
+                view(request, pk=rec.id)
+
+        rec.refresh_from_db()
+        self.assertEqual(rec.custom_properties.get("poster_logo_id"), 555)
+        self.assertEqual(rec.custom_properties.get("poster_url"), "https://tmdb.com/new-poster.jpg")
+
+
+# ---------------------------------------------------------------------------
+# Logo proxy negative cache
+# ---------------------------------------------------------------------------
+
+class LogoNegativeCacheTests(TestCase):
+    """Tests for the _logo_fetch_failures negative cache in LogoViewSet.cache()."""
+
+    def setUp(self):
+        from apps.channels import api_views
+        self._failures = api_views._logo_fetch_failures
+        self._failures.clear()
+        self.factory = APIRequestFactory()
+        self.user = _make_admin()
+
+    def _fetch_logo(self, logo):
+        request = self.factory.get(f"/api/channels/logos/{logo.id}/cache/")
+        force_authenticate(request, user=self.user)
+        view = LogoViewSet.as_view({"get": "cache"})
+        return view(request, pk=logo.id)
+
+    def test_failed_url_cached_on_non_200(self):
+        """Non-200 response adds URL to negative cache."""
+        logo = Logo.objects.create(name="Dead Logo", url="https://dead-cdn.com/logo.png")
+        mock_resp = MagicMock(status_code=404)
+        with patch("apps.channels.api_views.requests.get", return_value=mock_resp), \
+             patch("apps.channels.api_views.CoreSettings.get_default_user_agent_id", return_value="1"), \
+             patch("apps.channels.api_views.UserAgent.objects.get", return_value=MagicMock(user_agent="Test/1.0")):
+            response = self._fetch_logo(logo)
+        self.assertEqual(response.status_code, 404)
+        self.assertIn("https://dead-cdn.com/logo.png", self._failures)
+
+    def test_cached_failure_returns_404_immediately(self):
+        """Subsequent request for a cached-failed URL returns 404 without making a request."""
+        logo = Logo.objects.create(name="Cached Fail", url="https://cached-fail.com/logo.png")
+        self._failures["https://cached-fail.com/logo.png"] = time_mod.monotonic() + 300
+
+        with patch("apps.channels.api_views.requests.get") as mock_get:
+            response = self._fetch_logo(logo)
+        self.assertEqual(response.status_code, 404)
+        mock_get.assert_not_called()
+
+    def test_expired_cache_entry_allows_retry(self):
+        """After TTL expires, a new request is made."""
+        logo = Logo.objects.create(name="Expired", url="https://expired.com/logo.png")
+        self._failures["https://expired.com/logo.png"] = time_mod.monotonic() - 1  # already expired
+
+        mock_resp = MagicMock(status_code=200)
+        mock_resp.headers = {"Content-Type": "image/png"}
+        mock_resp.iter_content = MagicMock(return_value=[b"img"])
+        with patch("apps.channels.api_views.requests.get", return_value=mock_resp), \
+             patch("apps.channels.api_views.CoreSettings.get_default_user_agent_id", return_value="1"), \
+             patch("apps.channels.api_views.UserAgent.objects.get", return_value=MagicMock(user_agent="Test/1.0")):
+            response = self._fetch_logo(logo)
+        self.assertEqual(response.status_code, 200)
+
+    def test_success_clears_previous_failure(self):
+        """A successful fetch removes the URL from the failure cache."""
+        url = "https://recovered.com/logo.png"
+        logo = Logo.objects.create(name="Recovered", url=url)
+        self._failures[url] = time_mod.monotonic() - 1  # expired
+
+        mock_resp = MagicMock(status_code=200)
+        mock_resp.headers = {"Content-Type": "image/png"}
+        mock_resp.iter_content = MagicMock(return_value=[b"img"])
+        with patch("apps.channels.api_views.requests.get", return_value=mock_resp), \
+             patch("apps.channels.api_views.CoreSettings.get_default_user_agent_id", return_value="1"), \
+             patch("apps.channels.api_views.UserAgent.objects.get", return_value=MagicMock(user_agent="Test/1.0")):
+            self._fetch_logo(logo)
+        self.assertNotIn(url, self._failures)
+
+    def test_request_exception_cached(self):
+        """Network errors are cached the same as non-200 responses."""
+        import requests
+        logo = Logo.objects.create(name="Timeout", url="https://timeout.com/logo.png")
+        with patch("apps.channels.api_views.requests.get", side_effect=requests.Timeout("timed out")), \
+             patch("apps.channels.api_views.CoreSettings.get_default_user_agent_id", return_value="1"), \
+             patch("apps.channels.api_views.UserAgent.objects.get", return_value=MagicMock(user_agent="Test/1.0")):
+            response = self._fetch_logo(logo)
+        self.assertEqual(response.status_code, 404)
+        self.assertIn("https://timeout.com/logo.png", self._failures)
+
+    def test_eviction_when_cache_exceeds_256(self):
+        """Stale entries are evicted when the cache grows past 256."""
+        now = time_mod.monotonic()
+        # Fill with 257 expired entries
+        for i in range(257):
+            self._failures[f"https://old-{i}.com/x.png"] = now - 1  # already expired
+
+        logo = Logo.objects.create(name="Trigger", url="https://trigger-evict.com/logo.png")
+        import requests
+        with patch("apps.channels.api_views.requests.get", side_effect=requests.ConnectionError("fail")), \
+             patch("apps.channels.api_views.CoreSettings.get_default_user_agent_id", return_value="1"), \
+             patch("apps.channels.api_views.UserAgent.objects.get", return_value=MagicMock(user_agent="Test/1.0")):
+            self._fetch_logo(logo)
+
+        # Expired entries should be evicted
+        old_entries = [k for k in self._failures if k.startswith("https://old-")]
+        self.assertEqual(len(old_entries), 0)
+        # New failure entry should exist
+        self.assertIn("https://trigger-evict.com/logo.png", self._failures)
diff --git a/apps/channels/tests/test_validate_url.py b/apps/channels/tests/test_validate_url.py
new file mode 100644
index 00000000..e52c79fe
--- /dev/null
+++ b/apps/channels/tests/test_validate_url.py
@@ -0,0 +1,169 @@
+"""Tests for the _validate_url() helper in tasks.py.
+
+Covers:
+  - Rejection of None, empty, and non-string inputs
+  - Non-HTTP URLs pass through without network requests
+  - HTTP(S) URLs validated via HEAD request (2xx/3xx pass, 4xx/5xx fail)
+  - Network errors (timeout, connection) treated as failures
+  - Per-worker result cache: hits, expiry, eviction
+"""
+from unittest.mock import patch, MagicMock
+
+from django.test import TestCase
+
+from apps.channels.tasks import _validate_url, _url_validation_cache, _URL_CACHE_TTL
+
+
+class ValidateUrlInputTests(TestCase):
+    """Input validation — no network requests should be made."""
+
+    def setUp(self):
+        _url_validation_cache.clear()
+
+    def test_none_returns_false(self):
+        self.assertFalse(_validate_url(None))
+
+    def test_empty_string_returns_false(self):
+        self.assertFalse(_validate_url(""))
+
+    def test_non_string_returns_false(self):
+        self.assertFalse(_validate_url(123))
+        self.assertFalse(_validate_url(["http://example.com"]))
+
+    @patch("apps.channels.tasks.requests.head")
+    def test_non_http_url_returns_true_without_request(self, mock_head):
+        """file:// and other non-HTTP schemes skip validation."""
+        self.assertTrue(_validate_url("file:///local/path.jpg"))
+        self.assertTrue(_validate_url("/data/images/poster.jpg"))
+        mock_head.assert_not_called()
+
+
+class ValidateUrlNetworkTests(TestCase):
+    """HTTP(S) URL validation via HEAD request."""
+
+    def setUp(self):
+        _url_validation_cache.clear()
+
+    @patch("apps.channels.tasks.requests.head")
+    def test_200_returns_true(self, mock_head):
+        mock_head.return_value = MagicMock(status_code=200)
+        self.assertTrue(_validate_url("https://example.com/poster.jpg"))
+
+    @patch("apps.channels.tasks.requests.head")
+    def test_302_redirect_returns_true(self, mock_head):
+        mock_head.return_value = MagicMock(status_code=302)
+        self.assertTrue(_validate_url("https://example.com/redirect"))
+
+    @patch("apps.channels.tasks.requests.head")
+    def test_404_returns_false(self, mock_head):
+        mock_head.return_value = MagicMock(status_code=404)
+        self.assertFalse(_validate_url("https://dead-cdn.com/missing.jpg"))
+
+    @patch("apps.channels.tasks.requests.head")
+    def test_500_returns_false(self, mock_head):
+        mock_head.return_value = MagicMock(status_code=500)
+        self.assertFalse(_validate_url("https://broken.com/error"))
+
+    @patch("apps.channels.tasks.requests.head")
+    def test_timeout_returns_false(self, mock_head):
+        import requests
+        mock_head.side_effect = requests.Timeout("timed out")
+        self.assertFalse(_validate_url("https://slow-cdn.com/poster.jpg"))
+
+    @patch("apps.channels.tasks.requests.head")
+    def test_connection_error_returns_false(self, mock_head):
+        import requests
+        mock_head.side_effect = requests.ConnectionError("refused")
+        self.assertFalse(_validate_url("https://unreachable.com/poster.jpg"))
+
+    @patch("apps.channels.tasks.requests.head")
+    def test_custom_timeout_passed_to_head(self, mock_head):
+        mock_head.return_value = MagicMock(status_code=200)
+        _validate_url("https://example.com/img.jpg", timeout=10)
+        mock_head.assert_called_once_with(
+            "https://example.com/img.jpg", timeout=10, allow_redirects=True
+        )
+
+    @patch("apps.channels.tasks.requests.get")
+    @patch("apps.channels.tasks.requests.head")
+    def test_405_falls_back_to_get(self, mock_head, mock_get):
+        """When HEAD returns 405, fall back to a ranged GET request."""
+        mock_head.return_value = MagicMock(status_code=405)
+        mock_resp = MagicMock(status_code=200)
+        mock_get.return_value = mock_resp
+        self.assertTrue(_validate_url("https://no-head.com/poster.jpg"))
+        mock_get.assert_called_once()
+        mock_resp.close.assert_called_once()
+
+    @patch("apps.channels.tasks.requests.get")
+    @patch("apps.channels.tasks.requests.head")
+    def test_405_fallback_get_also_fails(self, mock_head, mock_get):
+        """When HEAD returns 405 and GET also fails, return False."""
+        mock_head.return_value = MagicMock(status_code=405)
+        mock_get.return_value = MagicMock(status_code=403)
+        self.assertFalse(_validate_url("https://blocked.com/poster.jpg"))
+
+
+class ValidateUrlCacheTests(TestCase):
+    """Per-worker result caching."""
+
+    def setUp(self):
+        _url_validation_cache.clear()
+
+    @patch("apps.channels.tasks.requests.head")
+    def test_cache_hit_avoids_second_request(self, mock_head):
+        mock_head.return_value = MagicMock(status_code=200)
+        url = "https://cached.com/poster.jpg"
+        self.assertTrue(_validate_url(url))
+        self.assertTrue(_validate_url(url))
+        mock_head.assert_called_once()
+
+    @patch("apps.channels.tasks.requests.head")
+    def test_cache_hit_returns_false_for_failed_url(self, mock_head):
+        mock_head.return_value = MagicMock(status_code=404)
+        url = "https://dead.com/missing.jpg"
+        self.assertFalse(_validate_url(url))
+        self.assertFalse(_validate_url(url))
+        mock_head.assert_called_once()
+
+    @patch("apps.channels.tasks.time.monotonic")
+    @patch("apps.channels.tasks.requests.head")
+    def test_cache_expiry_triggers_new_request(self, mock_head, mock_time):
+        """After TTL expires, a new HEAD request is made."""
+        mock_head.return_value = MagicMock(status_code=200)
+        url = "https://expiring.com/poster.jpg"
+
+        mock_time.return_value = 1000.0
+        self.assertTrue(_validate_url(url))
+        self.assertEqual(mock_head.call_count, 1)
+
+        # Within TTL — cache hit
+        mock_time.return_value = 1000.0 + _URL_CACHE_TTL - 1
+        self.assertTrue(_validate_url(url))
+        self.assertEqual(mock_head.call_count, 1)
+
+        # Past TTL — new request
+        mock_time.return_value = 1000.0 + _URL_CACHE_TTL + 1
+        self.assertTrue(_validate_url(url))
+        self.assertEqual(mock_head.call_count, 2)
+
+    @patch("apps.channels.tasks.time.monotonic")
+    @patch("apps.channels.tasks.requests.head")
+    def test_eviction_when_cache_exceeds_limit(self, mock_head, mock_time):
+        """Expired entries are evicted when cache grows past 512 entries."""
+        mock_head.return_value = MagicMock(status_code=200)
+
+        # Fill cache with 513 entries at time 0
+        mock_time.return_value = 0.0
+        for i in range(513):
+            _url_validation_cache[f"https://fill-{i}.com/img.jpg"] = (True, 0.0)
+
+        # Advance past TTL and add one more — triggers eviction
+        mock_time.return_value = _URL_CACHE_TTL + 1
+        _validate_url("https://trigger-eviction.com/img.jpg")
+
+        # All 513 old entries expired and should be evicted
+        remaining = [k for k in _url_validation_cache if k.startswith("https://fill-")]
+        self.assertEqual(len(remaining), 0)
+        # The new entry should remain
+        self.assertIn("https://trigger-eviction.com/img.jpg", _url_validation_cache)
diff --git a/apps/proxy/ts_proxy/client_manager.py b/apps/proxy/ts_proxy/client_manager.py
index 3351cf56..836f719e 100644
--- a/apps/proxy/ts_proxy/client_manager.py
+++ b/apps/proxy/ts_proxy/client_manager.py
@@ -23,7 +23,7 @@ class ClientManager:
         self.channel_id = channel_id
         self.redis_client = redis_client
         self.clients = set()
-        self.lock = threading.Lock()
+        self.lock = threading.RLock()
         self.last_active_time = time.time()
         self.worker_id = worker_id  # Store worker ID as instance variable
         self._heartbeat_running = True  # Flag to control heartbeat thread
diff --git a/frontend/package-lock.json b/frontend/package-lock.json
index 82c718eb..5ba9fec3 100644
--- a/frontend/package-lock.json
+++ b/frontend/package-lock.json
@@ -5794,24 +5794,6 @@
       "dev": true,
       "license": "MIT"
     },
-    "node_modules/yaml": {
-      "version": "2.8.2",
-      "resolved": "https://registry.npmjs.org/yaml/-/yaml-2.8.2.tgz",
-      "integrity": "sha512-mplynKqc1C2hTVYxd0PU2xQAc22TI1vShAYGksCCfxbn/dFwnHTNi1bvYsBTkhdUNtGIf5xNOg938rrSSYvS9A==",
-      "dev": true,
-      "license": "ISC",
-      "optional": true,
-      "peer": true,
-      "bin": {
-        "yaml": "bin.mjs"
-      },
-      "engines": {
-        "node": ">= 14.6"
-      },
-      "funding": {
-        "url": "https://github.com/sponsors/eemeli"
-      }
-    },
     "node_modules/yocto-queue": {
       "version": "0.1.0",
       "resolved": "https://registry.npmjs.org/yocto-queue/-/yocto-queue-0.1.0.tgz",
diff --git a/frontend/src/api.js b/frontend/src/api.js
index 43aca702..35233204 100644
--- a/frontend/src/api.js
+++ b/frontend/src/api.js
@@ -2687,6 +2687,30 @@ export default class API {
     }
   }
 
+  static async refreshArtwork(id) {
+    try {
+      await request(`${host}/api/channels/recordings/${id}/refresh-artwork/`, {
+        method: 'POST',
+      });
+    } catch (e) {
+      errorNotification(`Failed to refresh artwork for recording ${id}`, e);
+      throw e;
+    }
+  }
+
+  static async updateRecordingMetadata(id, { title, description }) {
+    try {
+      await request(`${host}/api/channels/recordings/${id}/update-metadata/`, {
+        method: 'POST',
+        body: JSON.stringify({ title, description }),
+        headers: { 'Content-Type': 'application/json' },
+      });
+    } catch (e) {
+      errorNotification(`Failed to update recording metadata`, e);
+      throw e;
+    }
+  }
+
   static async runComskip(recordingId) {
     try {
       const resp = await request(
diff --git a/frontend/src/components/cards/RecordingCard.jsx b/frontend/src/components/cards/RecordingCard.jsx
index 073ec542..e8d836fa 100644
--- a/frontend/src/components/cards/RecordingCard.jsx
+++ b/frontend/src/components/cards/RecordingCard.jsx
@@ -23,6 +23,7 @@ import {
   Tooltip,
 } from '@mantine/core';
 import { AlertTriangle, Plus, Square, SquareX } from 'lucide-react';
+import defaultLogo from '../../images/logo.png';
 import RecordingSynopsis from '../RecordingSynopsis';
 import {
   deleteRecordingById,
@@ -39,6 +40,30 @@ import {
   stopRecordingById,
 } from './../../utils/cards/RecordingCardUtils.js';
 
+const areRecordingPropsEqual = (prev, next) => {
+  const pr = prev.recording;
+  const nr = next.recording;
+  if (!pr || !nr) return pr === nr;
+
+  const pcp = pr.custom_properties || {};
+  const ncp = nr.custom_properties || {};
+
+  return (
+    pr.id === nr.id &&
+    pr.start_time === nr.start_time &&
+    pr.end_time === nr.end_time &&
+    pr._group_count === nr._group_count &&
+    pcp.status === ncp.status &&
+    pcp.poster_logo_id === ncp.poster_logo_id &&
+    pcp.poster_url === ncp.poster_url &&
+    pcp.file_url === ncp.file_url &&
+    pcp.output_file_url === ncp.output_file_url &&
+    pcp.comskip?.status === ncp.comskip?.status &&
+    pcp.program?.title === ncp.program?.title &&
+    prev.channel?.id === next.channel?.id
+  );
+};
+
 const RecordingCard = ({
   recording,
   onOpenDetails,
@@ -61,12 +86,11 @@ const RecordingCard = ({
   const description = program.description || customProps.description || '';
   const isRecurringRule = customProps?.rule?.type === 'recurring';
 
-  // Poster or channel logo
+  // Poster or channel logo (getPosterUrl falls back to Dispatcharr default logo)
   const posterUrl = getPosterUrl(
     customProps.poster_logo_id,
     customProps,
-    getChannelLogoUrl(channel),
-    env_mode
+    getChannelLogoUrl(channel)
   );
 
   const start = toUserTime(recording.start_time);
@@ -283,6 +307,8 @@ const RecordingCard = ({
         backgroundColor: isInterrupted ? '#2b1f20' : '#27272A',
         borderColor: isInterrupted ? '#a33' : undefined,
         height: '100%',
+        display: 'flex',
+        flexDirection: 'column',
         cursor: 'pointer',
       }}
       onClick={handleOnMainCardClick}
@@ -383,7 +409,7 @@ const RecordingCard = ({
         </Group>
       </Flex>
 
-      <Flex gap="sm" align="center">
+      <Flex gap="sm" align="flex-start" style={{ flex: 1 }}>
         <Image
           src={posterUrl}
           w={64}
@@ -391,9 +417,9 @@ const RecordingCard = ({
           fit="contain"
           radius="sm"
           alt={recordingName}
-          fallbackSrc="/logo.png"
+          fallbackSrc={getChannelLogoUrl(channel) || defaultLogo}
         />
-        <Stack gap={6} flex={1}>
+        <Stack gap={6} flex={1} style={{ alignSelf: 'stretch' }}>
           {!isSeriesGroup && subTitle && (
             <Group justify="space-between">
               <Text size="sm" c="dimmed">
@@ -436,7 +462,7 @@ const RecordingCard = ({
             </Text>
           )}
 
-          <Group justify="flex-end" gap="xs" pt={4}>
+          <Group justify="flex-end" gap="xs" pt={4} style={{ marginTop: 'auto' }}>
             {isInProgress && <WatchLive />}
 
             {!isUpcoming && <WatchRecording />}
@@ -591,4 +617,4 @@ const RecordingCard = ({
   );
 };
 
-export default RecordingCard;
+export default React.memo(RecordingCard, areRecordingPropsEqual);
diff --git a/frontend/src/components/forms/RecordingDetailsModal.jsx b/frontend/src/components/forms/RecordingDetailsModal.jsx
index 65599e24..f3b52c96 100644
--- a/frontend/src/components/forms/RecordingDetailsModal.jsx
+++ b/frontend/src/components/forms/RecordingDetailsModal.jsx
@@ -5,7 +5,9 @@ import {
   useTimeHelpers,
 } from '../../utils/dateTimeUtils.js';
 import React from 'react';
+import { Pencil, RefreshCcw, Check, X } from 'lucide-react';
 import {
+  ActionIcon,
   Badge,
   Button,
   Card,
@@ -15,9 +17,12 @@ import {
   Modal,
   Stack,
   Text,
+  Textarea,
+  TextInput,
 } from '@mantine/core';
 import useVideoStore from '../../store/useVideoStore.jsx';
 import { notifications } from '@mantine/notifications';
+import defaultLogo from '../../images/logo.png';
 import {
   deleteRecordingById,
   getChannelLogoUrl,
@@ -53,11 +58,49 @@ const RecordingDetailsModal = ({
   const { timeFormat: timeformat, dateFormat: dateformat } =
     useDateTimeFormat();
 
-  const safeRecording = recording || {};
+  // Prefer the store version of the recording for live updates
+  // (e.g., after artwork refresh or metadata edit via WebSocket).
+  // Preserve _group_count from the categorized prop — the store version
+  // doesn't carry this client-side field, so without merging it back
+  // isSeriesGroup would always be false and the episode list hidden.
+  const safeRecording = React.useMemo(() => {
+    if (recording?.id && Array.isArray(allRecordings)) {
+      const found = allRecordings.find((r) => r.id === recording.id);
+      if (found) {
+        if (recording._group_count != null) {
+          return { ...found, _group_count: recording._group_count };
+        }
+        return found;
+      }
+    }
+    return recording || {};
+  }, [allRecordings, recording]);
   const customProps = safeRecording.custom_properties || {};
   const program = customProps.program || {};
-  const recordingName = program.title || 'Custom Recording';
-  const description = program.description || customProps.description || '';
+
+  // Derive poster URL from live store data instead of the stale prop snapshot.
+  const livePosterUrl = React.useMemo(
+    () => getPosterUrl(customProps.poster_logo_id, customProps, getChannelLogoUrl(channel)),
+    [customProps.poster_logo_id, customProps, channel]
+  );
+
+  // Optimistic overrides — show saved values immediately without waiting
+  // for the WebSocket round-trip to refresh the store.
+  const [savedTitle, setSavedTitle] = React.useState(null);
+  const [savedDescription, setSavedDescription] = React.useState(null);
+  const recordingName = savedTitle ?? (program.title || 'Custom Recording');
+  const description = savedDescription ?? (program.description || customProps.description || '');
+
+  const [editing, setEditing] = React.useState(false);
+  const [editTitle, setEditTitle] = React.useState('');
+  const [editDescription, setEditDescription] = React.useState('');
+
+  // Reset optimistic state when the recording changes
+  React.useEffect(() => {
+    setSavedTitle(null);
+    setSavedDescription(null);
+    setEditing(false);
+  }, [recording?.id]);
   const start = toUserTime(safeRecording.start_time);
   const end = toUserTime(safeRecording.end_time);
   const stats = customProps.stream_info || {};
@@ -165,6 +208,49 @@ const RecordingDetailsModal = ({
     });
   };
 
+  const startEditing = () => {
+    setEditTitle(recordingName === 'Custom Recording' ? '' : recordingName);
+    setEditDescription(description);
+    setEditing(true);
+  };
+
+  const cancelEditing = () => setEditing(false);
+
+  const saveMetadata = async () => {
+    try {
+      await API.updateRecordingMetadata(recording.id, {
+        title: editTitle || 'Custom Recording',
+        description: editDescription,
+      });
+      setSavedTitle(editTitle || 'Custom Recording');
+      setSavedDescription(editDescription);
+      setEditing(false);
+      notifications.show({
+        title: 'Saved',
+        message: 'Recording metadata updated',
+        color: 'green',
+        autoClose: 2000,
+      });
+    } catch (error) {
+      console.error('Failed to save metadata', error);
+    }
+  };
+
+  const handleRefreshArtwork = async (e) => {
+    e.stopPropagation?.();
+    try {
+      await API.refreshArtwork(recording.id);
+      notifications.show({
+        title: 'Refreshing artwork',
+        message: 'Poster resolution started',
+        color: 'blue.5',
+        autoClose: 2000,
+      });
+    } catch (error) {
+      console.error('Failed to refresh artwork', error);
+    }
+  };
+
   const handleRunComskip = async (e) => {
     e.stopPropagation?.();
     try {
@@ -193,7 +279,8 @@ const RecordingDetailsModal = ({
       cp.onscreen_episode ?? pr?.custom_properties?.onscreen_episode;
     const se = getSeasonLabel(season, episode, onscreen);
     const posterLogoId = cp.poster_logo_id;
-    const purl = getPosterUrl(posterLogoId, cp, posterUrl);
+    const purl = getPosterUrl(posterLogoId, cp, livePosterUrl);
+    const epChannel = channelsById[rec.channel] || (rec.channel === recording?.channel ? channel : null);
 
     const onRemove = async (e) => {
       e?.stopPropagation?.();
@@ -226,7 +313,7 @@ const RecordingDetailsModal = ({
             fit="contain"
             radius="sm"
             alt={pr.title || recordingName}
-            fallbackSrc="/logo.png"
+            fallbackSrc={getChannelLogoUrl(epChannel) || defaultLogo}
           />
           <Stack gap={4} flex={1}>
             <Group justify="space-between">
@@ -340,15 +427,27 @@ const RecordingDetailsModal = ({
   const Movie = () => {
     return (
       <Flex gap="lg" align="flex-start">
-        <Image
-          src={posterUrl}
-          w={180}
-          h={240}
-          fit="contain"
-          radius="sm"
-          alt={recordingName}
-          fallbackSrc="/logo.png"
-        />
+        <Stack gap={4} align="center">
+          <Image
+            src={livePosterUrl}
+            w={180}
+            h={240}
+            fit="contain"
+            radius="sm"
+            alt={recordingName}
+            fallbackSrc={getChannelLogoUrl(channel) || defaultLogo}
+          />
+          <Button
+            size="compact-xs"
+            variant="subtle"
+            color="dimmed"
+            leftSection={<RefreshCcw size={12} />}
+            onClick={handleRefreshArtwork}
+            styles={{ root: { fontWeight: 400 } }}
+          >
+            Refresh artwork
+          </Button>
+        </Stack>
         <Stack gap={8} style={{ flex: 1 }}>
           <Group justify="space-between" align="center">
             <Text c="dimmed" size="sm">
@@ -385,11 +484,20 @@ const RecordingDetailsModal = ({
               </Badge>
             </Group>
           )}
-          {description && (
+          {editing ? (
+            <Textarea
+              value={editDescription}
+              onChange={(e) => setEditDescription(e.currentTarget.value)}
+              placeholder="Description (optional)"
+              size="sm"
+              minRows={2}
+              autosize
+            />
+          ) : description ? (
             <Text size="sm" style={{ whiteSpace: 'pre-wrap' }}>
               {description}
             </Text>
-          )}
+          ) : null}
           {statRows.length > 0 && (
             <Stack gap={4} pt={6}>
               <Text fw={600} size="sm">
@@ -415,9 +523,38 @@ const RecordingDetailsModal = ({
       opened={opened}
       onClose={onClose}
       title={
-        isSeriesGroup
-          ? `Series: ${recordingName}`
-          : `${recordingName}${program.sub_title ? ` - ${program.sub_title}` : ''}`
+        editing ? (
+          <Group gap={8} align="center" wrap="nowrap" style={{ flex: 1 }}>
+            <TextInput
+              value={editTitle}
+              onChange={(e) => setEditTitle(e.currentTarget.value)}
+              placeholder="Recording title"
+              size="sm"
+              style={{ flex: 1 }}
+              autoFocus
+              onKeyDown={(e) => { if (e.key === 'Enter') saveMetadata(); if (e.key === 'Escape') cancelEditing(); }}
+            />
+            <ActionIcon size="sm" variant="subtle" color="green" onClick={saveMetadata}>
+              <Check size={14} />
+            </ActionIcon>
+            <ActionIcon size="sm" variant="subtle" color="gray" onClick={cancelEditing}>
+              <X size={14} />
+            </ActionIcon>
+          </Group>
+        ) : (
+          <Group gap={8} align="center">
+            <span>
+              {isSeriesGroup
+                ? `Series: ${recordingName}`
+                : `${recordingName}${program.sub_title ? ` - ${program.sub_title}` : ''}`}
+            </span>
+            {!isSeriesGroup && (
+              <ActionIcon size="sm" variant="subtle" color="dimmed" onClick={startEditing}>
+                <Pencil size={14} />
+              </ActionIcon>
+            )}
+          </Group>
+        )
       }
       size="lg"
       centered
diff --git a/frontend/src/components/forms/RecurringRuleModal.jsx b/frontend/src/components/forms/RecurringRuleModal.jsx
index 630dbee6..34f49b66 100644
--- a/frontend/src/components/forms/RecurringRuleModal.jsx
+++ b/frontend/src/components/forms/RecurringRuleModal.jsx
@@ -33,7 +33,7 @@ import {
   updateRecurringRuleEnabled,
 } from '../../utils/forms/RecurringRuleModalUtils.js';
 
-const RecurringRuleModal = ({ opened, onClose, ruleId, onEditOccurrence }) => {
+const RecurringRuleModal = ({ opened, onClose, ruleId, recording: sourceRecording, onEditOccurrence }) => {
   const channels = useChannelsStore((s) => s.channels);
   const recurringRules = useChannelsStore((s) => s.recurringRules);
   const fetchRecurringRules = useChannelsStore((s) => s.fetchRecurringRules);
@@ -202,7 +202,45 @@ const RecurringRuleModal = ({ opened, onClose, ruleId, onEditOccurrence }) => {
   if (!rule) {
     return (
       <Modal opened={opened} onClose={onClose} title="Recurring Rule" centered>
-        <Text size="sm">Recurring rule not found.</Text>
+        <Stack gap="md">
+          <Text size="sm">
+            The recurring rule for this recording no longer exists.
+          </Text>
+          {sourceRecording && (
+            <>
+              <Text size="sm" c="dimmed">
+                Would you like to delete this recording?
+              </Text>
+              <Group justify="flex-end">
+                <Button variant="default" onClick={onClose}>
+                  Cancel
+                </Button>
+                <Button
+                  color="red"
+                  loading={deleting}
+                  onClick={async () => {
+                    setDeleting(true);
+                    try {
+                      await deleteRecordingById(sourceRecording.id);
+                      notifications.show({
+                        title: 'Recording deleted',
+                        color: 'green',
+                        autoClose: 2500,
+                      });
+                      onClose();
+                    } catch (e) {
+                      console.error('Failed to delete orphaned recording', e);
+                    } finally {
+                      setDeleting(false);
+                    }
+                  }}
+                >
+                  Delete Recording
+                </Button>
+              </Group>
+            </>
+          )}
+        </Stack>
       </Modal>
     );
   }
diff --git a/frontend/src/pages/DVR.jsx b/frontend/src/pages/DVR.jsx
index 0c04a51c..5289127d 100644
--- a/frontend/src/pages/DVR.jsx
+++ b/frontend/src/pages/DVR.jsx
@@ -76,7 +76,7 @@ const DVRPage = () => {
   };
   const closeDetails = () => setDetailsOpen(false);
 
-  const openRuleModal = (recording) => {
+  const openRuleModal = (recording, isDelete) => {
     const ruleId = recording?.custom_properties?.rule?.id;
     if (!ruleId) {
       openDetails(recording);
@@ -85,7 +85,7 @@ const DVRPage = () => {
     setDetailsOpen(false);
     setDetailsRecording(null);
     setEditRecording(null);
-    setRuleModal({ open: true, ruleId });
+    setRuleModal({ open: true, ruleId, recording, isDelete: isDelete || false });
   };
 
   const closeRuleModal = () => setRuleModal({ open: false, ruleId: null });
@@ -285,6 +285,7 @@ const DVRPage = () => {
         opened={ruleModal.open}
         onClose={closeRuleModal}
         ruleId={ruleModal.ruleId}
+        recording={ruleModal.recording}
         onEditOccurrence={(occ) => {
           setRuleModal({ open: false, ruleId: null });
           setEditRecording(occ);
diff --git a/frontend/src/pages/__tests__/DVR.test.jsx b/frontend/src/pages/__tests__/DVR.test.jsx
index 83e3e03f..15c68f5d 100644
--- a/frontend/src/pages/__tests__/DVR.test.jsx
+++ b/frontend/src/pages/__tests__/DVR.test.jsx
@@ -128,11 +128,15 @@ vi.mock('../../utils/dateTimeUtils.js', async (importActual) => {
     useTimeHelpers: vi.fn(),
   };
 });
-vi.mock('../../utils/cards/RecordingCardUtils.js', () => ({
-  getPosterUrl: vi.fn(),
-  getRecordingUrl: vi.fn(),
-  getShowVideoUrl: vi.fn(),
-}));
+vi.mock('../../utils/cards/RecordingCardUtils.js', async (importOriginal) => {
+  const actual = await importOriginal();
+  return {
+    ...actual,
+    getPosterUrl: vi.fn(),
+    getRecordingUrl: vi.fn(),
+    getShowVideoUrl: vi.fn(),
+  };
+});
 vi.mock('../../utils/pages/DVRUtils.js', async (importActual) => {
   const actual = await importActual();
   return {
diff --git a/frontend/src/utils/cards/RecordingCardUtils.js b/frontend/src/utils/cards/RecordingCardUtils.js
index 216494b8..c390f28b 100644
--- a/frontend/src/utils/cards/RecordingCardUtils.js
+++ b/frontend/src/utils/cards/RecordingCardUtils.js
@@ -1,5 +1,6 @@
 import API from '../../api.js';
 import useChannelsStore from '../../store/channels.jsx';
+import defaultLogo from '../../images/logo.png';
 
 export const removeRecording = (id) => {
   // Optimistically remove immediately from UI
@@ -32,17 +33,17 @@ export const getChannelLogoUrl = (channel) => {
 export const getPosterUrl = (posterLogoId, customProperties, posterUrl) => {
   let purl = posterLogoId
     ? `/api/channels/logos/${posterLogoId}/cache/`
-    : customProperties?.poster_url || posterUrl || '/logo.png';
+    : customProperties?.poster_url || posterUrl || null;
   if (
+    purl &&
     typeof import.meta !== 'undefined' &&
     import.meta.env &&
     import.meta.env.DEV &&
-    purl &&
     purl.startsWith('/')
   ) {
     purl = `${window.location.protocol}//${window.location.hostname}:5656${purl}`;
   }
-  return purl;
+  return purl || defaultLogo;
 };
 
 export const getShowVideoUrl = (channel, env_mode) => {
diff --git a/frontend/src/utils/cards/__tests__/RecordingCardUtils.test.js b/frontend/src/utils/cards/__tests__/RecordingCardUtils.test.js
index 882b0fb1..4dad5900 100644
--- a/frontend/src/utils/cards/__tests__/RecordingCardUtils.test.js
+++ b/frontend/src/utils/cards/__tests__/RecordingCardUtils.test.js
@@ -126,7 +126,9 @@ describe('RecordingCardUtils', () => {
       vi.stubEnv('DEV', false);
       const result = getPosterUrl(null, {}, '');
 
-      expect(result).toBe('/logo.png');
+      // Falls back to the imported default Dispatcharr logo asset
+      expect(result).toBeTruthy();
+      expect(result).toContain('logo');
     });
 
     it('prepends dev server URL in dev mode for relative paths', () => {
diff --git a/frontend/src/utils/forms/RecordingDetailsModalUtils.js b/frontend/src/utils/forms/RecordingDetailsModalUtils.js
index e3ea2496..c81c3b53 100644
--- a/frontend/src/utils/forms/RecordingDetailsModalUtils.js
+++ b/frontend/src/utils/forms/RecordingDetailsModalUtils.js
@@ -30,8 +30,9 @@ const filterByUpcoming = (arr, tvid, titleKey, toUserTime, userNow) => {
 
     if ((pr.tvg_id || '') !== tvid) return false;
     if ((pr.title || '').toLowerCase() !== titleKey) return false;
-    const st = toUserTime(r.start_time);
-    return st.isAfter(userNow());
+    // Include episodes that haven't ended yet (currently-airing + future)
+    const et = toUserTime(r.end_time);
+    return et.isAfter(userNow());
   });
 };
 

From bd14e48f8b0d3c67625f33163adfc2906cf8a8e2 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Mon, 2 Mar 2026 22:00:49 -0600
Subject: [PATCH 40/58] =?UTF-8?q?fix(dvr):=20harden=20recording=20pipeline?=
 =?UTF-8?q?=20=E2=80=94=20idempotency=20guard,=20remux=20sanity=20checks,?=
 =?UTF-8?q?=20timezone=20fix?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- Idempotency guard: fail closed on DB errors instead of proceeding, preventing duplicate tasks from overwriting valid recordings (#641)
- Remux sanity checks: reject MKV output that is <50% of a previous MKV (duplicate-task overwrite) or <10% of the source TS (corrupt first attempt); preserve .ts for manual recovery on all failure paths (#619)
- Recurring rule timezone fix: use user's configured timezone for date window calculation instead of UTC, fixing silent recording loss for users in UTC-negative timezones after 4pm local time (#1042)
- Include .ts path in final remux failure log for easier recovery
---
 apps/channels/tasks.py | 63 ++++++++++++++++++++++++++++++++++++++----
 1 file changed, 57 insertions(+), 6 deletions(-)

diff --git a/apps/channels/tasks.py b/apps/channels/tasks.py
index 73e60441..9548d659 100755
--- a/apps/channels/tasks.py
+++ b/apps/channels/tasks.py
@@ -1404,14 +1404,15 @@ def sync_recurring_rule_impl(rule_id: int, drop_existing: bool = True, horizon_d
     except Exception:
         logger.warning("Invalid or unsupported time zone '%s'; falling back to Server default", tz_name)
         tz = timezone.get_current_timezone()
-    start_limit = rule.start_date or now.date()
+    local_today = now.astimezone(tz).date()
+    start_limit = rule.start_date or local_today
     end_limit = rule.end_date
     horizon = now + timedelta(days=horizon_days)
-    start_window = max(start_limit, now.date())
+    start_window = max(start_limit, local_today)
     if drop_existing and end_limit:
         end_window = end_limit
     else:
-        end_window = horizon.date()
+        end_window = horizon.astimezone(tz).date()
         if end_limit and end_limit < end_window:
             end_window = end_limit
     if end_window < start_window:
@@ -1633,6 +1634,8 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     from .models import Recording, Logo
 
     # --- Idempotency guard (prevents duplicate recordings from task redelivery) ---
+    # Fail closed: if the DB is unreachable, abort rather than risk a duplicate
+    # task overwriting a valid recording.
     try:
         rec_check = Recording.objects.filter(id=recording_id).only("custom_properties").first()
         if not rec_check:
@@ -1648,7 +1651,11 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
             )
             return
     except Exception as e:
-        logger.debug(f"Idempotency guard check failed (proceeding anyway): {e}")
+        logger.error(
+            f"Idempotency guard DB check failed for recording {recording_id} "
+            f"({type(e).__name__}: {e}) — aborting to prevent potential duplicate."
+        )
+        return
 
     # --- Clean up the one-off PeriodicTask that dispatched this task ---
     try:
@@ -2114,7 +2121,7 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     recording_cancelled = not Recording.objects.filter(id=recording_id).exists()
     if recording_cancelled:
         logger.info(f"Recording {recording_id} was cancelled — skipping remux and metadata.")
-        # Clean up any remaining temp or stub output files
+        # Clean up all artifacts for the cancelled recording
         for _cleanup_path in [temp_ts_path, final_path]:
             if not _cleanup_path:
                 continue
@@ -2129,6 +2136,12 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
 
     # Remux TS to MKV container with retry for transient I/O errors
     remux_success = False
+    existing_mkv_size = 0
+    try:
+        if final_path and os.path.exists(final_path):
+            existing_mkv_size = os.path.getsize(final_path)
+    except OSError:
+        pass
     for _remux_attempt in range(_dvr_remux_max_retries):
         try:
             if temp_ts_path and os.path.exists(temp_ts_path):
@@ -2197,6 +2210,43 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
                     else:
                         logger.error(f"TS→MP4 conversion failed (return code: {result_mp4.returncode})")
 
+                # Sanity-check the remuxed file.  Two checks:
+                # 1. If a pre-existing MKV was overwritten, reject a
+                #    file that is drastically smaller (duplicate-task
+                #    overwrite protection).
+                # 2. If the MKV is smaller than the .ts source, the
+                #    remux likely produced a corrupt or truncated file.
+                if remux_success:
+                    try:
+                        new_size = os.path.getsize(final_path)
+                        ts_size = os.path.getsize(temp_ts_path) if temp_ts_path and os.path.exists(temp_ts_path) else 0
+                        reject = False
+                        if existing_mkv_size > 0 and new_size < existing_mkv_size * 0.5:
+                            logger.error(
+                                f"DVR recording {recording_id}: new MKV "
+                                f"({new_size:,} bytes) is less than 50%% of "
+                                f"the previous MKV ({existing_mkv_size:,} bytes) "
+                                f"— refusing to overwrite. Keeping .ts for "
+                                f"manual recovery."
+                            )
+                            reject = True
+                        elif ts_size > 0 and new_size < ts_size * 0.1:
+                            logger.error(
+                                f"DVR recording {recording_id}: remuxed MKV "
+                                f"({new_size:,} bytes) is less than 10%% of "
+                                f"the source TS ({ts_size:,} bytes) — likely "
+                                f"corrupt. Keeping .ts for manual recovery."
+                            )
+                            reject = True
+                        if reject:
+                            remux_success = False
+                            try:
+                                os.remove(final_path)
+                            except OSError:
+                                pass
+                    except OSError:
+                        pass
+
                 # Clean up temp TS file only on successful remux
                 if remux_success:
                     try:
@@ -2234,7 +2284,8 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
             else:
                 logger.warning(
                     f"DVR recording {recording_id}: remux failed "
-                    f"after {_dvr_remux_max_retries} attempts: {e}"
+                    f"after {_dvr_remux_max_retries} attempts: {e}. "
+                    f"Keeping .ts for manual recovery: {temp_ts_path}"
                 )
 
     # Persist final metadata to Recording (status, ended_at, and stream stats if available)

From 9d08305dcb9ed5973385e29d33c09e815b0928b8 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Tue, 3 Mar 2026 13:59:51 -0600
Subject: [PATCH 41/58] fix(dvr): fix recording modal crash, harden recovery
 pipeline, improve tests
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Frontend:
- RecordingDetailsModal: fix ReferenceError crash (editing state used before declaration in useMemo — TDZ in production bundles broke all tile clicks)
- RecordingDetailsModal: fix focus-stealing on description edit by calling Movie()/Series() as plain functions instead of <Movie />/<Series />(inline component refs change every render, causing unmount/remount)
- RecordingDetailsModal: suppress sub_title in modal title after custom save
- ErrorBoundary: log caught errors and show message for easier debugging
- guideUtils: filter terminal recordings from mapRecordingsByProgramId so Guide red dots clear after recording finishes or rule is deleted
- RecordingDetailsModalUtils.test: add missing end_time to test fixtures (filterByUpcoming now uses end_time instead of start_time)

Backend:
- tasks.py: output file collision avoidance (check .mkv/.ts before writing)
- tasks.py: pre-restart TS segment concatenation (direct concat→MKV remux)
- tasks.py: recovery pipeline — recover crashed "recording" status, remux expired recordings, revoke stale tasks, deterministic task IDs, early lock release, 10-min lock TTL for large remux operations
- tasks.py: poster resolver skips external API when title matches channel name
- celery.py: trigger recover_recordings_on_startup via worker_ready signal

Tests:
- Renamed test_dvr_fixes.py → test_recording_pipeline.py (clarity)
- Renamed test_dvr_retry.py → test_db_retry.py (clarity), fixed mock paths for close_old_connections (resolved 3 test failures), replaced synthetic test
- test_recording_stop_cancel: replaced brittle time.sleep(0.5) with deterministic thread-capture pattern
---
 apps/channels/tasks.py                        | 246 +++++++-
 .../{test_dvr_retry.py => test_db_retry.py}   |  58 +-
 .../channels/tests/test_recording_pipeline.py | 524 ++++++++++++++++++
 .../tests/test_recording_stop_cancel.py       |  25 +-
 dispatcharr/celery.py                         |   9 +-
 frontend/src/components/ErrorBoundary.jsx     |   8 +-
 .../forms/RecordingDetailsModal.jsx           |   9 +-
 frontend/src/pages/guideUtils.js              |   6 +-
 .../RecordingDetailsModalUtils.test.js        |   4 +
 9 files changed, 826 insertions(+), 63 deletions(-)
 rename apps/channels/tests/{test_dvr_retry.py => test_db_retry.py} (84%)
 create mode 100644 apps/channels/tests/test_recording_pipeline.py

diff --git a/apps/channels/tasks.py b/apps/channels/tasks.py
index 9548d659..29fe5fc2 100755
--- a/apps/channels/tasks.py
+++ b/apps/channels/tasks.py
@@ -1611,6 +1611,28 @@ def _build_output_paths(channel, program, start_time, end_time):
         rel_path = rel_path[2:]
     final_path = rel_path if rel_path.startswith('/') else os.path.join(library_root, rel_path)
     final_path = os.path.normpath(final_path)
+
+    # Avoid overwriting an existing file from a different recording.
+    # Check BOTH .mkv and .ts — a pre-restart TS segment may exist at
+    # the same base name even when the MKV is a 0-byte placeholder.
+    base, ext = os.path.splitext(final_path)
+    counter = 1
+    while True:
+        candidate_base = final_path[:-len(ext)]  # strip extension
+        ts_candidate = candidate_base + '.ts'
+        try:
+            mkv_occupied = os.stat(final_path).st_size > 0
+        except OSError:
+            mkv_occupied = False
+        try:
+            ts_occupied = os.stat(ts_candidate).st_size > 0
+        except OSError:
+            ts_occupied = False
+        if not mkv_occupied and not ts_occupied:
+            break
+        counter += 1
+        final_path = f"{base}_{counter}{ext}"
+
     # Ensure directory exists
     os.makedirs(os.path.dirname(final_path), exist_ok=True)
 
@@ -2121,8 +2143,13 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     recording_cancelled = not Recording.objects.filter(id=recording_id).exists()
     if recording_cancelled:
         logger.info(f"Recording {recording_id} was cancelled — skipping remux and metadata.")
-        # Clean up all artifacts for the cancelled recording
-        for _cleanup_path in [temp_ts_path, final_path]:
+        # Clean up all artifacts for the cancelled recording,
+        # including any pre-restart .ts segments from server recovery.
+        # Use the in-memory recording_obj since the DB row is already deleted.
+        _cancel_cleanup = [temp_ts_path, final_path]
+        _cancel_cp = (recording_obj.custom_properties or {}) if recording_obj else {}
+        _cancel_cleanup.extend(_cancel_cp.get("_pre_restart_ts_paths", []))
+        for _cleanup_path in _cancel_cleanup:
             if not _cleanup_path:
                 continue
             try:
@@ -2134,8 +2161,93 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
                 pass
         return
 
+    # Concatenate pre-restart .ts segments with the current segment.
+    # Instead of creating an intermediate combined.ts and then remuxing to
+    # MKV (which loses timestamp boundary info and causes playback freezes
+    # at the splice point), go directly from the concat list → MKV.
+    # This lets ffmpeg's MKV muxer see each segment boundary and write
+    # correct cue points / clusters for seamless seeking.
+    _concat_did_remux = False
+    try:
+        _rec_obj_for_concat = Recording.objects.filter(id=recording_id).only("custom_properties").first()
+        _concat_cp = (_rec_obj_for_concat.custom_properties or {}) if _rec_obj_for_concat else {}
+        pre_restart_segments = _concat_cp.get("_pre_restart_ts_paths", [])
+        # Filter to segments that still exist on disk and have data
+        def _has_data(p):
+            try:
+                return os.stat(p).st_size > 0
+            except OSError:
+                return False
+        pre_restart_segments = [p for p in pre_restart_segments if p and _has_data(p)]
+        if pre_restart_segments and temp_ts_path and os.path.exists(temp_ts_path):
+            all_segments = pre_restart_segments + [temp_ts_path]
+            concat_list_path = temp_ts_path + ".concat.txt"
+            try:
+                with open(concat_list_path, "w") as cl:
+                    for seg in all_segments:
+                        cl.write(f"file '{seg}'\n")
+
+                # Direct concat → MKV in a single pass.
+                # -reset_timestamps 1 tells the concat demuxer to reset
+                # timestamps at each segment boundary, eliminating the
+                # discontinuity that causes playback to freeze at the
+                # splice point.
+                concat_result = subprocess.run(
+                    [
+                        "ffmpeg", "-y",
+                        "-fflags", "+genpts+igndts+discardcorrupt",
+                        "-err_detect", "ignore_err",
+                        "-f", "concat", "-safe", "0",
+                        "-segment_time_metadata", "1",
+                        "-i", concat_list_path,
+                        "-reset_timestamps", "1",
+                        "-map", "0",
+                        "-c", "copy",
+                        final_path,
+                    ],
+                    stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True,
+                )
+                if concat_result.returncode == 0 and os.path.exists(final_path) and os.path.getsize(final_path) > 0:
+                    _concat_did_remux = True
+                    # Clean up individual TS segments (including current)
+                    for seg in all_segments:
+                        try:
+                            os.remove(seg)
+                        except OSError:
+                            pass
+                    logger.info(
+                        f"DVR recording {recording_id}: concat→MKV succeeded — "
+                        f"{len(all_segments)} segments → {os.path.basename(final_path)} "
+                        f"({os.path.getsize(final_path):,} bytes)"
+                    )
+                else:
+                    logger.warning(
+                        f"DVR recording {recording_id}: direct concat→MKV failed "
+                        f"(rc={concat_result.returncode}), falling back to "
+                        f"normal remux with current segment only. "
+                        f"stderr: {(concat_result.stderr or '')[:500]}"
+                    )
+            finally:
+                try:
+                    os.remove(concat_list_path)
+                except OSError:
+                    pass
+            # Clear the pre-restart paths from custom_properties
+            if _rec_obj_for_concat:
+                _ccp = _rec_obj_for_concat.custom_properties or {}
+                _ccp.pop("_pre_restart_ts_paths", None)
+                _ccp.pop("interrupted_reason", None)
+                _rec_obj_for_concat.custom_properties = _ccp
+                _rec_obj_for_concat.save(update_fields=["custom_properties"])
+    except Exception as e:
+        logger.warning(
+            f"DVR recording {recording_id}: segment concatenation error "
+            f"({type(e).__name__}: {e}), proceeding with current segment only."
+        )
+
     # Remux TS to MKV container with retry for transient I/O errors
-    remux_success = False
+    # (Skip if concat already produced the final MKV directly.)
+    remux_success = _concat_did_remux
     existing_mkv_size = 0
     try:
         if final_path and os.path.exists(final_path):
@@ -2143,6 +2255,8 @@ def run_recording(recording_id, channel_id, start_time_str, end_time_str):
     except OSError:
         pass
     for _remux_attempt in range(_dvr_remux_max_retries):
+        if remux_success:
+            break
         try:
             if temp_ts_path and os.path.exists(temp_ts_path):
                 # First attempt: Direct TS to MKV remux
@@ -2432,8 +2546,9 @@ def recover_recordings_on_startup():
         redis = RedisClient.get_client()
         if redis:
             lock_key = "dvr:recover_lock"
-            # Set lock with 60s TTL; only first winner proceeds
-            if not redis.set(lock_key, "1", ex=60, nx=True):
+            # Set lock with 10-minute TTL; must be long enough for Phase 2
+            # ffmpeg remux operations on large files.
+            if not redis.set(lock_key, "1", ex=600, nx=True):
                 return "Recovery already in progress"
 
         now = timezone.now()
@@ -2451,12 +2566,15 @@ def recover_recordings_on_startup():
                 cp = rec.custom_properties or {}
                 current_status = cp.get("status", "")
 
-                # Skip recordings that are already in a terminal or active state.
+                # Skip recordings that are already in a terminal state.
                 # "completed" / "stopped" — user stopped or it finished normally; do NOT
                 # overwrite the status and re-schedule (that would cause the
                 # Interrupted → In-Progress → Previously-Recorded ghost cycle).
-                # "recording" — a worker is already streaming; leave it alone.
-                if current_status in ("completed", "stopped", "recording"):
+                # NOTE: "recording" is NOT skipped — this function runs on
+                # worker_ready, meaning all previous workers are dead.  A
+                # recording stuck in "recording" status is from a crashed
+                # worker and must be recovered.
+                if current_status in ("completed", "stopped"):
                     logger.info(
                         f"recover_recordings_on_startup: skipping recording {rec.id} "
                         f"(status={current_status!r}, already in terminal/active state)."
@@ -2466,19 +2584,104 @@ def recover_recordings_on_startup():
                 # Mark interrupted due to restart; will flip to 'recording' when task starts
                 cp["status"] = "interrupted"
                 cp["interrupted_reason"] = "server_restarted"
+
+                # Preserve the pre-restart .ts segment path so run_recording
+                # can concatenate it with the resumed segment later.
+                old_ts = cp.get("_temp_file_path")
+                if old_ts and os.path.exists(old_ts) and os.path.getsize(old_ts) > 0:
+                    prior_segments = cp.get("_pre_restart_ts_paths", [])
+                    prior_segments.append(old_ts)
+                    cp["_pre_restart_ts_paths"] = prior_segments
+                    logger.info(
+                        f"recover_recordings_on_startup: recording {rec.id} — "
+                        f"preserving pre-restart TS segment: {old_ts}"
+                    )
+
                 rec.custom_properties = cp
                 _db_retry(
                     lambda r=rec: r.save(update_fields=["custom_properties"]),
                     label=f"DVR recovery: recording {rec.id} status update",
                 )
 
-                # Start recording for remaining window
+                # Revoke the old PeriodicTask so Celery Beat doesn't also
+                # fire run_recording for this recording (would be a duplicate).
+                old_task_id = rec.task_id
+                if old_task_id:
+                    try:
+                        revoke_task(old_task_id)
+                    except Exception:
+                        pass
+
+                # Start recording for remaining window.  Use a deterministic
+                # task_id so duplicate dispatches (e.g. from a second recovery
+                # attempt) are deduplicated by Celery/Redis.
+                recovery_task_id = f"dvr-recover-{rec.id}"
                 run_recording.apply_async(
-                    args=[rec.id, rec.channel_id, str(now), str(rec.end_time)], eta=now
+                    args=[rec.id, rec.channel_id, str(now), str(rec.end_time)],
+                    eta=now,
+                    task_id=recovery_task_id,
                 )
             except Exception as e:
                 logger.warning(f"Failed to resume recording {rec.id}: {e}")
 
+        # Finalize expired recordings that were active when the server crashed
+        # but whose end_time has now passed.  Remux the partial .ts and mark
+        # as interrupted so the user can watch whatever was captured.
+        expired = _db_retry(
+            lambda: list(Recording.objects.filter(
+                end_time__lte=now,
+                custom_properties__status="recording",
+            )),
+            label="DVR recovery: fetching expired recordings",
+        )
+        for rec in expired:
+            try:
+                cp = rec.custom_properties or {}
+                ts_path = cp.get("_temp_file_path")
+                mkv_path = cp.get("file_path")
+
+                if ts_path and os.path.exists(ts_path) and os.path.getsize(ts_path) > 0 and mkv_path:
+                    logger.info(
+                        f"recover_recordings_on_startup: recording {rec.id} expired "
+                        f"during downtime — remuxing partial TS ({os.path.getsize(ts_path):,} bytes)"
+                    )
+                    os.makedirs(os.path.dirname(mkv_path), exist_ok=True)
+                    result = subprocess.run(
+                        [
+                            "ffmpeg", "-y",
+                            "-fflags", "+genpts+igndts+discardcorrupt",
+                            "-err_detect", "ignore_err",
+                            "-i", ts_path, "-map", "0", "-c", "copy", mkv_path,
+                        ],
+                        stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True,
+                    )
+                    if result.returncode == 0 and os.path.exists(mkv_path) and os.path.getsize(mkv_path) > 0:
+                        cp["status"] = "interrupted"
+                        cp["interrupted_reason"] = "server_restarted_after_end"
+                        cp["remux_success"] = True
+                        try:
+                            os.remove(ts_path)
+                        except OSError:
+                            pass
+                        logger.info(f"recover_recordings_on_startup: recording {rec.id} remuxed successfully")
+                    else:
+                        cp["status"] = "interrupted"
+                        cp["interrupted_reason"] = "server_restarted_after_end"
+                        cp["remux_success"] = False
+                        logger.warning(f"recover_recordings_on_startup: recording {rec.id} remux failed, keeping .ts")
+                else:
+                    cp["status"] = "interrupted"
+                    cp["interrupted_reason"] = "server_restarted_after_end"
+                    cp["remux_success"] = False
+
+                rec.custom_properties = cp
+                _db_retry(
+                    lambda r=rec: r.save(update_fields=["custom_properties"]),
+                    label=f"DVR recovery: recording {rec.id} expired status update",
+                )
+            except Exception as e:
+                logger.warning(f"Failed to finalize expired recording {rec.id}: {e}")
+
         # Ensure future recordings are scheduled.
         # With ClockedSchedule, PeriodicTasks survive restarts in the DB.
         # Only recreate if the PeriodicTask is missing (safety net).
@@ -2521,6 +2724,12 @@ def recover_recordings_on_startup():
             except Exception as e:
                 logger.warning(f"Failed to schedule recording {rec.id}: {e}")
 
+        # Release the lock early so a subsequent restart can recover
+        # immediately.  The 10-minute TTL is only a safety net in case
+        # recovery itself crashes before reaching this point.
+        if redis:
+            redis.delete(lock_key)
+
         return "Recovery complete"
     except Exception as e:
         logger.error(f"Error during DVR recovery: {e}")
@@ -2791,6 +3000,15 @@ def _resolve_poster_for_program(channel_name, program, channel_logo_id=None):
 
     _title = ((program.get("title") if isinstance(program, dict) else None) or "").strip() or None
 
+    # Guard: if the "title" is really just the channel name (common when EPG
+    # has no real program data), don't use it for external API searches —
+    # those queries produce false-positive artwork from unrelated shows.
+    _title_is_channel_name = False
+    if _title and channel_name:
+        def _norm_channel(s):
+            return s.lower().replace("*", "").replace("-", " ").strip()
+        _title_is_channel_name = _norm_channel(_title) == _norm_channel(channel_name)
+
     # Stage 1: EPG Program images/icon (with URL validation)
     try:
         from apps.epg.models import ProgramData
@@ -2810,7 +3028,7 @@ def _resolve_poster_for_program(channel_name, program, channel_logo_id=None):
         pass
 
     # Stage 2: VOD logo fallback by title
-    if not poster_url and not poster_logo_id and _title:
+    if not poster_url and not poster_logo_id and _title and not _title_is_channel_name:
         try:
             from apps.vod.models import Movie, Series
             vod_logo = None
@@ -2827,7 +3045,7 @@ def _resolve_poster_for_program(channel_name, program, channel_logo_id=None):
             pass
 
     # Stage 3: TMDB/OMDb (keyed APIs)
-    if not poster_url and not poster_logo_id and _title:
+    if not poster_url and not poster_logo_id and _title and not _title_is_channel_name:
         try:
             tmdb_key = os.environ.get('TMDB_API_KEY')
             omdb_key = os.environ.get('OMDB_API_KEY')
@@ -2900,7 +3118,7 @@ def _resolve_poster_for_program(channel_name, program, channel_logo_id=None):
             pass
 
     # Stage 4: Keyless providers (TVMaze & iTunes)
-    if not poster_url and not poster_logo_id and _title:
+    if not poster_url and not poster_logo_id and _title and not _title_is_channel_name:
         try:
             title = _title
             # TVMaze
@@ -2948,7 +3166,7 @@ def _resolve_poster_for_program(channel_name, program, channel_logo_id=None):
                 break
 
     # Stage 6: Search existing Logo entries by program title
-    if not poster_logo_id and not poster_url and _title:
+    if not poster_logo_id and not poster_url and _title and not _title_is_channel_name:
         try:
             from .models import Logo
             existing = Logo.objects.filter(name__iexact=_title).first()
diff --git a/apps/channels/tests/test_dvr_retry.py b/apps/channels/tests/test_db_retry.py
similarity index 84%
rename from apps/channels/tests/test_dvr_retry.py
rename to apps/channels/tests/test_db_retry.py
index 7d6b3a97..e65d929a 100644
--- a/apps/channels/tests/test_dvr_retry.py
+++ b/apps/channels/tests/test_db_retry.py
@@ -25,7 +25,7 @@ class DbRetryTests(TestCase):
     """Tests for the _db_retry() exponential backoff helper."""
 
     @patch("apps.channels.tasks.time.sleep")
-    @patch("apps.channels.tasks.close_old_connections", create=True)
+    @patch("apps.channels.tasks.close_old_connections")
     def test_succeeds_on_first_attempt(self, _close, _sleep):
         """No retry needed when fn succeeds immediately."""
         result = _db_retry(lambda: "ok", max_retries=3)
@@ -33,7 +33,7 @@ class DbRetryTests(TestCase):
         _sleep.assert_not_called()
 
     @patch("apps.channels.tasks.time.sleep")
-    @patch("django.db.close_old_connections")
+    @patch("apps.channels.tasks.close_old_connections")
     def test_retries_on_operational_error_then_succeeds(self, mock_close, mock_sleep):
         """Retry succeeds on second attempt after OperationalError."""
         call_count = {"n": 0}
@@ -49,7 +49,7 @@ class DbRetryTests(TestCase):
         self.assertEqual(call_count["n"], 2)
 
     @patch("apps.channels.tasks.time.sleep")
-    @patch("django.db.close_old_connections")
+    @patch("apps.channels.tasks.close_old_connections")
     def test_raises_after_max_retries_exhausted(self, mock_close, mock_sleep):
         """Raises OperationalError after all retries fail."""
         def always_fail():
@@ -59,7 +59,7 @@ class DbRetryTests(TestCase):
             _db_retry(always_fail, max_retries=3, base_interval=1)
 
     @patch("apps.channels.tasks.time.sleep")
-    @patch("django.db.close_old_connections")
+    @patch("apps.channels.tasks.close_old_connections")
     def test_exponential_backoff_timing(self, mock_close, mock_sleep):
         """Sleep durations follow exponential backoff: 1s, 2s, 4s."""
         call_count = {"n": 0}
@@ -74,7 +74,7 @@ class DbRetryTests(TestCase):
         mock_sleep.assert_has_calls([call(1), call(2)])
 
     @patch("apps.channels.tasks.time.sleep")
-    @patch("django.db.close_old_connections")
+    @patch("apps.channels.tasks.close_old_connections")
     def test_close_old_connections_called_between_retries(self, mock_close, mock_sleep):
         """Stale DB connections are reset before each retry attempt."""
         call_count = {"n": 0}
@@ -89,7 +89,7 @@ class DbRetryTests(TestCase):
         mock_close.assert_called_once()
 
     @patch("apps.channels.tasks.time.sleep")
-    @patch("django.db.close_old_connections")
+    @patch("apps.channels.tasks.close_old_connections")
     def test_non_operational_error_not_retried(self, mock_close, mock_sleep):
         """Non-OperationalError exceptions propagate immediately."""
         def raise_value_error():
@@ -100,14 +100,14 @@ class DbRetryTests(TestCase):
         mock_sleep.assert_not_called()
 
     @patch("apps.channels.tasks.time.sleep")
-    @patch("django.db.close_old_connections")
+    @patch("apps.channels.tasks.close_old_connections")
     def test_returns_fn_return_value(self, mock_close, mock_sleep):
         """Return value of fn() is passed through."""
         result = _db_retry(lambda: {"key": "value"}, max_retries=3)
         self.assertEqual(result, {"key": "value"})
 
     @patch("apps.channels.tasks.time.sleep")
-    @patch("django.db.close_old_connections")
+    @patch("apps.channels.tasks.close_old_connections")
     def test_single_retry_allowed(self, mock_close, mock_sleep):
         """max_retries=1 means no retry — fail immediately."""
         with self.assertRaises(OperationalError):
@@ -177,7 +177,7 @@ class FinalMetadataSaveRetryTests(TestCase):
 
         with patch.object(rec, "save", side_effect=patched_save):
             with patch("apps.channels.tasks.time.sleep"):
-                with patch("django.db.close_old_connections"):
+                with patch("apps.channels.tasks.close_old_connections"):
                     def _save():
                         rec.custom_properties = cp
                         rec.save(update_fields=["custom_properties"])
@@ -192,35 +192,19 @@ class FinalMetadataSaveRetryTests(TestCase):
 # ---------------------------------------------------------------------------
 
 class InitialConnectionRetryTests(TestCase):
-    """When a candidate base URL fails with a retriable error before data flows,
-    the DVR task must retry the same base before moving to the next candidate."""
+    """Verify that the DVR task's reconnection logic retries the same
+    base URL before falling back to the next candidate."""
 
-    def setUp(self):
-        self.channel = Channel.objects.create(
-            channel_number=96, name="Connection Retry Channel"
-        )
+    def test_reconnect_max_constant_exists_in_run_recording(self):
+        """run_recording must define a max-reconnect limit to prevent
+        infinite retries on the same broken base URL."""
+        import inspect
+        from apps.channels.tasks import run_recording
+        source = inspect.getsource(run_recording)
 
-    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
-    def test_retriable_error_increments_reconnect_counter(self, _ws):
-        """Verify the reconnection counter logic: same base retried before fallback."""
-        # This tests the logic structure directly rather than running the full task.
-        # The reconnection counter starts at 0, increments on each retriable error,
-        # and the loop 'continue's (retries same base) until max is exceeded.
-        max_reconnects = 5
-        reconnects = 0
-        attempts_on_base = 0
-
-        # Simulate the inner reconnection loop logic
-        while True:
-            attempts_on_base += 1
-            # Simulate retriable error with no data
-            reconnects += 1
-            if reconnects <= max_reconnects:
-                continue  # Would retry same base
-            break  # Would move to next base
-
-        self.assertEqual(attempts_on_base, max_reconnects + 1)
-        self.assertEqual(reconnects, max_reconnects + 1)
+        # The reconnection counter pattern must be present
+        self.assertIn("reconnect", source.lower(),
+                       "run_recording must contain reconnection logic")
 
 
 # ---------------------------------------------------------------------------
@@ -263,7 +247,7 @@ class RecoveryRetryTests(TestCase):
 
         with patch.object(Recording, "save", patched_save):
             with patch("apps.channels.tasks.time.sleep"):
-                with patch("django.db.close_old_connections"):
+                with patch("apps.channels.tasks.close_old_connections"):
                     _db_retry(
                         lambda: rec.save(update_fields=["custom_properties"]),
                         max_retries=3,
diff --git a/apps/channels/tests/test_recording_pipeline.py b/apps/channels/tests/test_recording_pipeline.py
new file mode 100644
index 00000000..b1362422
--- /dev/null
+++ b/apps/channels/tests/test_recording_pipeline.py
@@ -0,0 +1,524 @@
+"""Tests for recent DVR fixes.
+
+Covers:
+  1. Collision avoidance: _build_output_paths checks both .mkv and .ts files
+  2. Logo guard: _resolve_poster_for_program skips external APIs when title ≈ channel name
+  3. Recording status lifecycle: status transitions visible via API
+  4. Concat flags: error-tolerant ffmpeg flags used for segment concatenation
+  5. Recovery skip-list: "recording" status NOT in terminal skip list
+"""
+import os
+from datetime import timedelta
+from unittest.mock import MagicMock, patch
+
+from django.test import TestCase
+from django.utils import timezone
+from rest_framework.test import APIRequestFactory, force_authenticate
+
+from apps.channels.models import Channel, Recording
+
+
+# ---------------------------------------------------------------------------
+# Helpers
+# ---------------------------------------------------------------------------
+
+def _make_admin():
+    from django.contrib.auth import get_user_model
+    User = get_user_model()
+    u, _ = User.objects.get_or_create(
+        username="dvr_fixes_admin",
+        defaults={"user_level": User.UserLevel.ADMIN},
+    )
+    u.set_password("pass")
+    u.save()
+    return u
+
+
+def _make_channel(name="Test Channel", number=100):
+    return Channel.objects.create(channel_number=number, name=name)
+
+
+def _make_recording(channel, **overrides):
+    now = timezone.now()
+    defaults = {
+        "channel": channel,
+        "start_time": now - timedelta(hours=1),
+        "end_time": now + timedelta(hours=1),
+        "custom_properties": {},
+    }
+    defaults.update(overrides)
+    return Recording.objects.create(**defaults)
+
+
+# =========================================================================
+# 1. Collision avoidance — _build_output_paths
+# =========================================================================
+
+class CollisionAvoidanceTests(TestCase):
+    """_build_output_paths must increment the filename counter when
+    EITHER the .mkv OR the .ts file already exists with size > 0."""
+
+    def _call(self, channel, program, start, end):
+        from apps.channels.tasks import _build_output_paths
+        return _build_output_paths(channel, program, start, end)
+
+    @patch("apps.channels.tasks.CoreSettings.get_dvr_tv_fallback_template",
+           return_value="TV/{show}/{start}.mkv")
+    @patch("apps.channels.tasks.CoreSettings.get_dvr_tv_template",
+           return_value="TV/{show}/S{season:02d}E{episode:02d}.mkv")
+    def test_no_collision_when_nothing_exists(self, _tv, _fb):
+        """Fresh path — no files exist, counter stays at 1."""
+        ch = MagicMock(name="TestCh")
+        ch.name = "TestCh"
+        program = {"title": "My Show"}
+        now = timezone.now()
+
+        def mock_stat(path):
+            raise OSError("No such file")
+
+        with patch("os.stat", side_effect=mock_stat), \
+             patch("os.makedirs"):
+            final, ts, fname = self._call(ch, program, now, now + timedelta(hours=1))
+
+        # Should NOT have a _2 suffix
+        self.assertNotIn("_2", final)
+        self.assertTrue(final.endswith(".mkv"))
+
+    @patch("apps.channels.tasks.CoreSettings.get_dvr_tv_fallback_template",
+           return_value="TV/{show}/{start}.mkv")
+    @patch("apps.channels.tasks.CoreSettings.get_dvr_tv_template",
+           return_value="TV/{show}/S{season:02d}E{episode:02d}.mkv")
+    def test_collision_when_ts_exists_but_mkv_is_zero_bytes(self, _tv, _fb):
+        """Pre-restart scenario: MKV is 0-byte placeholder, TS has real data.
+        The old code only checked MKV size, so it would reuse the path.
+        The fix also checks TS, so it must increment."""
+        ch = MagicMock(name="TestCh")
+        ch.name = "TestCh"
+        program = {"title": "My Show"}
+        now = timezone.now()
+
+        def mock_stat(path):
+            if "_2" in path:
+                raise OSError("No such file")
+            result = MagicMock()
+            if path.endswith('.mkv'):
+                result.st_size = 0       # MKV is 0-byte placeholder
+            elif path.endswith('.ts'):
+                result.st_size = 5000000  # TS has real data from pre-restart
+            else:
+                result.st_size = 0
+            return result
+
+        with patch("os.stat", side_effect=mock_stat), \
+             patch("os.makedirs"):
+            final, ts, fname = self._call(ch, program, now, now + timedelta(hours=1))
+
+        # Must have incremented to _2
+        self.assertIn("_2", final, "Should increment counter when TS file has data")
+
+    @patch("apps.channels.tasks.CoreSettings.get_dvr_tv_fallback_template",
+           return_value="TV/{show}/{start}.mkv")
+    @patch("apps.channels.tasks.CoreSettings.get_dvr_tv_template",
+           return_value="TV/{show}/S{season:02d}E{episode:02d}.mkv")
+    def test_collision_when_mkv_has_data(self, _tv, _fb):
+        """Standard collision: MKV file has data, should increment."""
+        ch = MagicMock(name="TestCh")
+        ch.name = "TestCh"
+        program = {"title": "My Show"}
+        now = timezone.now()
+
+        def mock_stat(path):
+            if "_2" in path:
+                raise OSError("No such file")
+            result = MagicMock()
+            if path.endswith('.mkv'):
+                result.st_size = 1000000  # MKV has data
+            else:
+                result.st_size = 0
+            return result
+
+        with patch("os.stat", side_effect=mock_stat), \
+             patch("os.makedirs"):
+            final, ts, fname = self._call(ch, program, now, now + timedelta(hours=1))
+
+        self.assertIn("_2", final, "Should increment counter when MKV file has data")
+
+    @patch("apps.channels.tasks.CoreSettings.get_dvr_tv_fallback_template",
+           return_value="TV/{show}/{start}.mkv")
+    @patch("apps.channels.tasks.CoreSettings.get_dvr_tv_template",
+           return_value="TV/{show}/S{season:02d}E{episode:02d}.mkv")
+    def test_no_collision_when_both_zero_bytes(self, _tv, _fb):
+        """Both MKV and TS exist but are 0 bytes — no collision."""
+        ch = MagicMock(name="TestCh")
+        ch.name = "TestCh"
+        program = {"title": "My Show"}
+        now = timezone.now()
+
+        def mock_stat(path):
+            result = MagicMock()
+            result.st_size = 0  # All files empty
+            return result
+
+        with patch("os.stat", side_effect=mock_stat), \
+             patch("os.makedirs"):
+            final, ts, fname = self._call(ch, program, now, now + timedelta(hours=1))
+
+        self.assertNotIn("_2", final, "Should NOT increment when all files are empty")
+
+    @patch("apps.channels.tasks.CoreSettings.get_dvr_tv_fallback_template",
+           return_value="TV/{show}/{start}.mkv")
+    @patch("apps.channels.tasks.CoreSettings.get_dvr_tv_template",
+           return_value="TV/{show}/S{season:02d}E{episode:02d}.mkv")
+    def test_collision_increments_to_3_when_2_also_occupied(self, _tv, _fb):
+        """When both base and _2 are occupied, should go to _3."""
+        ch = MagicMock(name="TestCh")
+        ch.name = "TestCh"
+        program = {"title": "My Show"}
+        now = timezone.now()
+
+        def mock_stat(path):
+            if "_3" in path:
+                raise OSError("No such file")
+            result = MagicMock()
+            if path.endswith('.ts'):
+                result.st_size = 5000000
+            else:
+                result.st_size = 0
+            return result
+
+        with patch("os.stat", side_effect=mock_stat), \
+             patch("os.makedirs"):
+            final, ts, fname = self._call(ch, program, now, now + timedelta(hours=1))
+
+        self.assertIn("_3", final, "Should increment to _3 when base and _2 are occupied")
+
+
+# =========================================================================
+# 2. Logo guard — _resolve_poster_for_program
+# =========================================================================
+
+class LogoGuardTests(TestCase):
+    """When the program title matches the channel name, external API
+    searches (VOD, TMDB, OMDb, TVMaze, iTunes) must be skipped."""
+
+    def _call(self, channel_name, program, channel_logo_id=None):
+        from apps.channels.tasks import _resolve_poster_for_program
+        return _resolve_poster_for_program(channel_name, program, channel_logo_id)
+
+    @patch("apps.channels.tasks.requests.get")
+    def test_channel_name_as_title_skips_external_apis(self, mock_get):
+        """Title = 'USA A&E SD*', channel = 'USA A&E SD*' → no external calls."""
+        program = {"title": "USA A&E SD*"}
+        logo_id, url = self._call("USA A&E SD*", program, channel_logo_id=42)
+
+        # Should NOT have called any external APIs
+        mock_get.assert_not_called()
+        # Should fall back to channel logo
+        self.assertEqual(logo_id, 42)
+        self.assertIsNone(url)
+
+    @patch("apps.channels.tasks.requests.get")
+    def test_channel_name_normalized_match(self, mock_get):
+        """Title = 'fox news', channel = 'FOX-News*' → normalized match, skip APIs."""
+        program = {"title": "fox news"}
+        logo_id, url = self._call("FOX-News*", program, channel_logo_id=99)
+
+        mock_get.assert_not_called()
+        self.assertEqual(logo_id, 99)
+
+    @patch("apps.channels.tasks.requests.get")
+    def test_real_title_still_searched(self, mock_get):
+        """Title = 'Breaking Bad' on channel 'AMC' → should try external APIs."""
+        # Mock TVMaze returning a result
+        mock_resp = MagicMock(ok=True, status_code=200)
+        mock_resp.json.return_value = {
+            "image": {"original": "https://tvmaze.com/breaking-bad.jpg"}
+        }
+        mock_get.return_value = mock_resp
+
+        program = {"title": "Breaking Bad"}
+        logo_id, url = self._call("AMC", program)
+
+        # Should have made at least one external API call
+        self.assertTrue(mock_get.called, "Should search external APIs for real titles")
+        self.assertIsNotNone(url)
+
+    @patch("apps.channels.tasks.requests.get")
+    def test_no_title_skips_to_channel_logo(self, mock_get):
+        """No title at all → falls through to channel logo, no API calls."""
+        program = {}
+        logo_id, url = self._call("SomeChannel", program, channel_logo_id=55)
+
+        mock_get.assert_not_called()
+        self.assertEqual(logo_id, 55)
+
+    @patch("apps.channels.tasks.requests.get")
+    def test_epg_image_still_used_even_when_title_is_channel_name(self, mock_get):
+        """Even when title = channel name, Stage 1 (EPG images) should still work."""
+        from apps.epg.models import ProgramData, EPGSource, EPGData
+
+        # Create an EPG source + EPGData entry + program with an icon URL
+        epg_source = EPGSource.objects.create(source_type="xmltv", name="Test EPG")
+        epg_data = EPGData.objects.create(tvg_id="test.ch", epg_source=epg_source)
+        prog = ProgramData.objects.create(
+            epg=epg_data,
+            title="Test Channel HD",
+            start_time=timezone.now() - timedelta(hours=1),
+            end_time=timezone.now() + timedelta(hours=1),
+            custom_properties={"icon": "https://epg-cdn.com/test-icon.png"},
+        )
+
+        program = {"title": "Test Channel HD", "id": prog.id}
+
+        # Mock _validate_url to return True for the icon URL
+        with patch("apps.channels.tasks._validate_url", return_value=True):
+            logo_id, url = self._call("Test Channel HD", program, channel_logo_id=10)
+
+        # EPG icon should still be used (Stage 1 doesn't depend on title guard)
+        self.assertEqual(url, "https://epg-cdn.com/test-icon.png")
+        mock_get.assert_not_called()
+
+
+# =========================================================================
+# 3. Recording status lifecycle via API
+# =========================================================================
+
+class RecordingStatusLifecycleTests(TestCase):
+    """Verify recording status transitions and that terminal recordings
+    are properly filterable (supports the red-dot fix in guideUtils)."""
+
+    def setUp(self):
+        self.channel = _make_channel("Status Test Channel", 200)
+        self.user = _make_admin()
+        self.factory = APIRequestFactory()
+
+    def _list_recordings(self):
+        from apps.channels.api_views import RecordingViewSet
+        request = self.factory.get("/api/channels/recordings/")
+        force_authenticate(request, user=self.user)
+        view = RecordingViewSet.as_view({"get": "list"})
+        return view(request)
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_stopped_recording_has_terminal_status(self, _ws):
+        """After stop, custom_properties.status = 'stopped'."""
+        from apps.channels.api_views import RecordingViewSet
+
+        rec = _make_recording(self.channel, custom_properties={
+            "status": "recording",
+            "program": {"id": 1, "title": "Live Show"},
+        })
+
+        request = self.factory.post(f"/api/channels/recordings/{rec.id}/stop/")
+        force_authenticate(request, user=self.user)
+        view = RecordingViewSet.as_view({"post": "stop"})
+
+        with patch("apps.channels.signals.revoke_task"):
+            response = view(request, pk=rec.id)
+
+        self.assertIn(response.status_code, [200, 204])
+        rec.refresh_from_db()
+        self.assertEqual(rec.custom_properties.get("status"), "stopped")
+
+    def test_listing_includes_status_in_custom_properties(self):
+        """API listing returns custom_properties with status field."""
+        _make_recording(self.channel, custom_properties={
+            "status": "recording",
+            "program": {"id": 1, "title": "Recording Show"},
+        })
+        _make_recording(self.channel, custom_properties={
+            "status": "stopped",
+            "program": {"id": 2, "title": "Stopped Show"},
+        })
+
+        response = self._list_recordings()
+        self.assertEqual(response.status_code, 200)
+
+        statuses = [r["custom_properties"].get("status") for r in response.data]
+        self.assertIn("recording", statuses)
+        self.assertIn("stopped", statuses)
+
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_delete_recording_removes_from_listing(self, _ws):
+        """Deleting a recording removes it from the listing entirely."""
+        from apps.channels.api_views import RecordingViewSet
+
+        rec = _make_recording(self.channel, custom_properties={
+            "status": "stopped",
+            "program": {"id": 3, "title": "To Delete"},
+        })
+        rec_id = rec.id
+
+        request = self.factory.delete(f"/api/channels/recordings/{rec_id}/")
+        force_authenticate(request, user=self.user)
+        view = RecordingViewSet.as_view({"delete": "destroy"})
+
+        with patch("apps.channels.signals.revoke_task"):
+            response = view(request, pk=rec_id)
+
+        self.assertIn(response.status_code, [200, 204])
+        self.assertFalse(Recording.objects.filter(id=rec_id).exists())
+
+
+# =========================================================================
+# 4. Concat flags — error-tolerant ffmpeg
+# =========================================================================
+
+class ConcatFlagsTests(TestCase):
+    """Verify that the finalize phase uses error-tolerant ffmpeg flags
+    when concatenating pre-restart segments."""
+
+    def test_concat_command_includes_error_tolerant_flags(self):
+        """Inspect the source code to confirm error-tolerant flags are present.
+        This is a static analysis test — no ffmpeg execution needed."""
+        import inspect
+        from apps.channels.tasks import run_recording
+        source = inspect.getsource(run_recording)
+
+        # The concat subprocess.run call must include these flags
+        self.assertIn("+genpts+igndts+discardcorrupt", source,
+                       "Concat must use +genpts+igndts+discardcorrupt fflags")
+        self.assertIn("ignore_err", source,
+                       "Concat must use -err_detect ignore_err")
+        self.assertIn("-f", source)
+        self.assertIn("concat", source)
+
+    def test_concat_goes_directly_to_mkv(self):
+        """Concat must produce MKV directly (not intermediate .ts) to
+        preserve timestamp boundaries and avoid playback freeze at splice."""
+        import inspect
+        from apps.channels.tasks import run_recording
+        source = inspect.getsource(run_recording)
+
+        # Must contain reset_timestamps for proper segment boundary handling
+        self.assertIn("reset_timestamps", source,
+                       "Concat must use -reset_timestamps 1 for seamless seeking")
+        # Must write directly to final_path (MKV), not an intermediate .ts
+        self.assertIn("_concat_did_remux", source,
+                       "Concat path must set flag to skip separate remux step")
+
+    def test_segment_time_metadata_present(self):
+        """Verify concat uses -segment_time_metadata for boundary awareness."""
+        import inspect
+        from apps.channels.tasks import run_recording
+        source = inspect.getsource(run_recording)
+
+        self.assertIn("segment_time_metadata", source,
+                       "Concat must use -segment_time_metadata 1 for segment boundary handling")
+
+
+# =========================================================================
+# 5. Recovery skip-list
+# =========================================================================
+
+class RecoverySkipListTests(TestCase):
+    """Verify that the recovery function does NOT skip 'recording' status,
+    since that's the exact status recordings have when the server crashes."""
+
+    def test_recording_status_not_in_skip_list(self):
+        """Inspect recover_recordings_on_startup to ensure 'recording' is
+        NOT treated as a terminal/skip state."""
+        import inspect
+        from apps.channels.tasks import recover_recordings_on_startup
+        source = inspect.getsource(recover_recordings_on_startup)
+
+        # Find the skip condition line
+        # It should be: if current_status in ("completed", "stopped"):
+        # NOT: if current_status in ("completed", "stopped", "recording"):
+        lines = source.split('\n')
+        skip_line = None
+        for line in lines:
+            if 'current_status in' in line and ('completed' in line or 'stopped' in line):
+                skip_line = line.strip()
+                break
+
+        self.assertIsNotNone(skip_line, "Should find the skip-list condition")
+        self.assertNotIn('"recording"', skip_line,
+                          "Skip list must NOT contain 'recording' — "
+                          "that's the status of crashed mid-stream recordings that need recovery")
+
+    @patch("core.utils.RedisClient")
+    @patch("apps.channels.tasks.run_recording")
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_recovery_processes_recording_status(self, _ws, mock_run, mock_redis_cls):
+        """A recording with status='recording' should be recovered, not skipped."""
+        mock_redis_conn = MagicMock()
+        mock_redis_conn.set.return_value = True  # Acquire lock
+        mock_redis_cls.get_client.return_value = mock_redis_conn
+
+        channel = _make_channel("Recovery Test", 300)
+        now = timezone.now()
+        rec = _make_recording(channel, custom_properties={
+            "status": "recording",
+            "program": {"title": "Crashed Show"},
+        }, end_time=now + timedelta(hours=2))
+
+        from apps.channels.tasks import recover_recordings_on_startup
+
+        with patch("apps.channels.signals.revoke_task"):
+            result = recover_recordings_on_startup()
+
+        # The recording should have been dispatched for recovery
+        self.assertTrue(mock_run.apply_async.called,
+                        "Recording with status='recording' should be dispatched for recovery")
+
+    @patch("core.utils.RedisClient")
+    @patch("apps.channels.tasks.run_recording")
+    @patch("core.utils.send_websocket_update", side_effect=lambda *a, **kw: None)
+    def test_recovery_skips_stopped_recordings(self, _ws, mock_run, mock_redis_cls):
+        """A recording with status='stopped' should be skipped by recovery."""
+        mock_redis_conn = MagicMock()
+        mock_redis_conn.set.return_value = True
+        mock_redis_cls.get_client.return_value = mock_redis_conn
+
+        channel = _make_channel("Recovery Skip Test", 301)
+        now = timezone.now()
+        rec = _make_recording(channel, custom_properties={
+            "status": "stopped",
+            "program": {"title": "Finished Show"},
+        }, end_time=now + timedelta(hours=2))
+
+        from apps.channels.tasks import recover_recordings_on_startup
+        with patch("apps.channels.signals.revoke_task"):
+            recover_recordings_on_startup()
+
+        # Should NOT have dispatched a recovery task
+        mock_run.apply_async.assert_not_called()
+
+
+# =========================================================================
+# 6. Frontend red-dot filter (guideUtils.mapRecordingsByProgramId)
+# =========================================================================
+
+class MapRecordingsByProgramIdTests(TestCase):
+    """These test the BACKEND side — confirming that recording status
+    is preserved in the API response so the frontend can filter on it.
+
+    The actual frontend filtering is covered by frontend/src/pages/__tests__/DVR.test.jsx
+    and the guideUtils code, but we verify the data contract here."""
+
+    def test_recording_custom_properties_status_persisted(self):
+        """Recording status in custom_properties survives save/load cycle."""
+        channel = _make_channel("Red Dot Test", 400)
+        rec = _make_recording(channel, custom_properties={
+            "status": "stopped",
+            "program": {"id": 42, "title": "A Show"},
+        })
+
+        rec.refresh_from_db()
+        self.assertEqual(rec.custom_properties["status"], "stopped")
+
+    def test_terminal_statuses_are_well_defined(self):
+        """Verify the terminal status set matches what the frontend uses."""
+        # These are the statuses that should NOT show a red dot in the Guide
+        terminal = {"stopped", "completed", "interrupted", "failed"}
+        channel = _make_channel("Terminal Status Test", 410)
+
+        # Verify each status is a valid recording status
+        for status in terminal:
+            rec = _make_recording(channel, custom_properties={
+                "status": status,
+                "program": {"id": 100, "title": "Test"},
+            })
+            rec.refresh_from_db()
+            self.assertEqual(rec.custom_properties["status"], status)
diff --git a/apps/channels/tests/test_recording_stop_cancel.py b/apps/channels/tests/test_recording_stop_cancel.py
index 962160ee..e4b22be4 100644
--- a/apps/channels/tests/test_recording_stop_cancel.py
+++ b/apps/channels/tests/test_recording_stop_cancel.py
@@ -7,7 +7,6 @@ Covers:
   - run_recording race guard before status write
   - _stop_dvr_clients() DVR client isolation
 """
-import time
 from datetime import timedelta
 from unittest.mock import MagicMock, AsyncMock, patch
 
@@ -82,13 +81,29 @@ class StopEndpointTests(TestCase):
         self.assertIn("stopped_at", rec.custom_properties)
 
     def test_stop_calls_stop_dvr_clients_in_background(self):
+        """stop() spawns a background thread whose target calls _stop_dvr_clients."""
         rec = self._make_rec()
+
         with patch("apps.channels.api_views._stop_dvr_clients", return_value=1) as mock_stop, \
-             patch("core.utils.send_websocket_update"):
+             patch("core.utils.send_websocket_update"), \
+             patch("threading.Thread") as mock_thread:
+            mock_thread.return_value.start = MagicMock()
             self._stop(rec)
-            time.sleep(0.5)
-        self.assertTrue(mock_stop.called)
-        args, kwargs = mock_stop.call_args
+
+        # Verify a daemon thread was spawned
+        mock_thread.assert_called_once()
+        thread_kwargs = mock_thread.call_args[1]
+        self.assertTrue(thread_kwargs.get("daemon"), "Thread must be daemon")
+
+        # Execute the captured target with DB connection close patched out
+        target = thread_kwargs["target"]
+        with patch("apps.channels.api_views._stop_dvr_clients", return_value=1) as mock_stop2, \
+             patch("apps.channels.signals.revoke_task", side_effect=Exception("skip")), \
+             patch("django.db.connection") as mock_conn:
+            target()
+
+        self.assertTrue(mock_stop2.called)
+        args, kwargs = mock_stop2.call_args
         actual_rec_id = kwargs.get("recording_id") or (args[1] if len(args) > 1 else None)
         self.assertEqual(actual_rec_id, rec.id)
 
diff --git a/dispatcharr/celery.py b/dispatcharr/celery.py
index c845dafe..436ffc35 100644
--- a/dispatcharr/celery.py
+++ b/dispatcharr/celery.py
@@ -2,7 +2,7 @@
 import os
 from celery import Celery
 import logging
-from celery.signals import task_postrun  # Add import for signals
+from celery.signals import task_postrun, worker_ready
 
 # Initialize with defaults before Django settings are loaded
 DEFAULT_LOG_LEVEL = 'DEBUG'
@@ -149,3 +149,10 @@ def setup_celery_logging(**kwargs):
         except (AttributeError, TypeError):
             # If the log level string is invalid, default to DEBUG
             logger.setLevel(logging.DEBUG)
+
+
+@worker_ready.connect
+def on_worker_ready(**kwargs):
+    """Resume or finalize interrupted DVR recordings after a worker restart."""
+    from apps.channels.tasks import recover_recordings_on_startup
+    recover_recordings_on_startup.delay()
diff --git a/frontend/src/components/ErrorBoundary.jsx b/frontend/src/components/ErrorBoundary.jsx
index 83fa4de7..b2de253d 100644
--- a/frontend/src/components/ErrorBoundary.jsx
+++ b/frontend/src/components/ErrorBoundary.jsx
@@ -4,12 +4,16 @@ class ErrorBoundary extends React.Component {
   state = { hasError: false };
 
   static getDerivedStateFromError(error) {
-    return { hasError: true };
+    return { hasError: true, error };
+  }
+
+  componentDidCatch(error, errorInfo) {
+    console.error('ErrorBoundary caught:', error, errorInfo);
   }
 
   render() {
     if (this.state.hasError) {
-      return <div>Something went wrong</div>;
+      return <div>Something went wrong: {this.state.error?.message}</div>;
     }
     return this.props.children;
   }
diff --git a/frontend/src/components/forms/RecordingDetailsModal.jsx b/frontend/src/components/forms/RecordingDetailsModal.jsx
index f3b52c96..4b2610c3 100644
--- a/frontend/src/components/forms/RecordingDetailsModal.jsx
+++ b/frontend/src/components/forms/RecordingDetailsModal.jsx
@@ -58,6 +58,8 @@ const RecordingDetailsModal = ({
   const { timeFormat: timeformat, dateFormat: dateformat } =
     useDateTimeFormat();
 
+  const [editing, setEditing] = React.useState(false);
+
   // Prefer the store version of the recording for live updates
   // (e.g., after artwork refresh or metadata edit via WebSocket).
   // Preserve _group_count from the categorized prop — the store version
@@ -91,7 +93,6 @@ const RecordingDetailsModal = ({
   const recordingName = savedTitle ?? (program.title || 'Custom Recording');
   const description = savedDescription ?? (program.description || customProps.description || '');
 
-  const [editing, setEditing] = React.useState(false);
   const [editTitle, setEditTitle] = React.useState('');
   const [editDescription, setEditDescription] = React.useState('');
 
@@ -546,7 +547,9 @@ const RecordingDetailsModal = ({
             <span>
               {isSeriesGroup
                 ? `Series: ${recordingName}`
-                : `${recordingName}${program.sub_title ? ` - ${program.sub_title}` : ''}`}
+                : savedTitle
+                  ? recordingName
+                  : `${recordingName}${program.sub_title ? ` - ${program.sub_title}` : ''}`}
             </span>
             {!isSeriesGroup && (
               <ActionIcon size="sm" variant="subtle" color="dimmed" onClick={startEditing}>
@@ -567,7 +570,7 @@ const RecordingDetailsModal = ({
         title: { color: 'white' },
       }}
     >
-      {isSeriesGroup ? <Series /> : <Movie />}
+      {isSeriesGroup ? Series() : Movie()}
     </Modal>
   );
 };
diff --git a/frontend/src/pages/guideUtils.js b/frontend/src/pages/guideUtils.js
index 98dc8e35..239adcfa 100644
--- a/frontend/src/pages/guideUtils.js
+++ b/frontend/src/pages/guideUtils.js
@@ -214,11 +214,15 @@ export const mapChannelsById = (guideChannels) => {
   return map;
 };
 
+const _terminalStatuses = new Set(['stopped', 'completed', 'interrupted', 'failed']);
+
 export const mapRecordingsByProgramId = (recordings) => {
   const map = new Map();
   (recordings || []).forEach((recording) => {
     const programId = recording?.custom_properties?.program?.id;
-    if (programId != null) {
+    const status = recording?.custom_properties?.status;
+    // Only show indicator for pending/active recordings, not terminal ones
+    if (programId != null && !_terminalStatuses.has(status)) {
       map.set(programId, recording);
     }
   });
diff --git a/frontend/src/utils/forms/__tests__/RecordingDetailsModalUtils.test.js b/frontend/src/utils/forms/__tests__/RecordingDetailsModalUtils.test.js
index ca689ff7..c86da2a8 100644
--- a/frontend/src/utils/forms/__tests__/RecordingDetailsModalUtils.test.js
+++ b/frontend/src/utils/forms/__tests__/RecordingDetailsModalUtils.test.js
@@ -282,6 +282,7 @@ describe('RecordingDetailsModalUtils', () => {
       const recordings = [
         {
           start_time: '2024-01-02T12:00:00',
+          end_time: '2024-01-02T13:00:00',
           channel: 'ch1',
           custom_properties: {
             program: { tvg_id: 'show1', title: 'Test Show' },
@@ -289,6 +290,7 @@ describe('RecordingDetailsModalUtils', () => {
         },
         {
           start_time: '2024-01-02T13:00:00',
+          end_time: '2024-01-02T14:00:00',
           channel: 'ch1',
           custom_properties: {
             program: { tvg_id: 'show2', title: 'Other Show' },
@@ -313,6 +315,7 @@ describe('RecordingDetailsModalUtils', () => {
       const recordings = [
         {
           start_time: '2023-12-31T12:00:00',
+          end_time: '2023-12-31T13:00:00',
           channel: 'ch1',
           custom_properties: {
             program: { tvg_id: 'show1', title: 'Test Show' },
@@ -320,6 +323,7 @@ describe('RecordingDetailsModalUtils', () => {
         },
         {
           start_time: '2024-01-02T12:00:00',
+          end_time: '2024-01-02T13:00:00',
           channel: 'ch1',
           custom_properties: {
             program: { tvg_id: 'show1', title: 'Test Show' },

From 1deea473ea9c0500a894e273f16fcc84fa2fa409 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Tue, 3 Mar 2026 19:59:00 -0600
Subject: [PATCH 42/58] fix users.test.jsx

Allow for ErrorBoundary message rendering so that the test does not fail expecting exact match
---
 frontend/src/pages/__tests__/Users.test.jsx | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/frontend/src/pages/__tests__/Users.test.jsx b/frontend/src/pages/__tests__/Users.test.jsx
index b7ab4ae2..94bf5552 100644
--- a/frontend/src/pages/__tests__/Users.test.jsx
+++ b/frontend/src/pages/__tests__/Users.test.jsx
@@ -21,7 +21,7 @@ describe('UsersPage', () => {
 
     const { container } = render(<UsersPage />);
 
-    expect(screen.getByText('Something went wrong')).toBeInTheDocument();
+    expect(screen.getByText(/Something went wrong/)).toBeInTheDocument();
     expect(screen.queryByTestId('users-table')).not.toBeInTheDocument();
   });
 
@@ -39,7 +39,7 @@ describe('UsersPage', () => {
     const { container } = render(<UsersPage />);
 
     // id: 0 is falsy, so should render empty
-    expect(screen.getByText('Something went wrong')).toBeInTheDocument();
+    expect(screen.getByText(/Something went wrong/)).toBeInTheDocument();
   });
 
   it('switches from unauthenticated to authenticated state', () => {
@@ -47,7 +47,7 @@ describe('UsersPage', () => {
 
     render(<UsersPage />);
 
-    expect(screen.getByText('Something went wrong')).toBeInTheDocument();
+    expect(screen.getByText(/Something went wrong/)).toBeInTheDocument();
 
     useAuthStore.mockReturnValue({ id: 1 });
 

From 2dec38773defac31ae3fe00f55bfda0bafef0961 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Tue, 3 Mar 2026 21:50:08 -0600
Subject: [PATCH 43/58] fix(dvr): fix recording card S/E overlap and hidden
 metadata

Move Season/Episode label from the header badge row to a dedicated metadata row in the card body, preventing action buttons (extend, stop, delete) from obscuring it on in-progress recordings with long titles. Also show episode subtitle and S/E info on series group cards, which were previously hidden behind !isSeriesGroup guards.
---
 frontend/src/components/cards/RecordingCard.jsx | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)

diff --git a/frontend/src/components/cards/RecordingCard.jsx b/frontend/src/components/cards/RecordingCard.jsx
index e8d836fa..903bb7dd 100644
--- a/frontend/src/components/cards/RecordingCard.jsx
+++ b/frontend/src/components/cards/RecordingCard.jsx
@@ -350,11 +350,6 @@ const RecordingCard = ({
                   Recurring
                 </Badge>
               )}
-              {seLabel && !isSeriesGroup && (
-                <Badge color="gray" variant="light">
-                  {seLabel}
-                </Badge>
-              )}
             </Group>
           </Stack>
         </Group>
@@ -420,7 +415,7 @@ const RecordingCard = ({
           fallbackSrc={getChannelLogoUrl(channel) || defaultLogo}
         />
         <Stack gap={6} flex={1} style={{ alignSelf: 'stretch' }}>
-          {!isSeriesGroup && subTitle && (
+          {subTitle && (
             <Group justify="space-between">
               <Text size="sm" c="dimmed">
                 Episode
@@ -430,6 +425,16 @@ const RecordingCard = ({
               </Text>
             </Group>
           )}
+          {seLabel && (
+            <Group justify="space-between">
+              <Text size="sm" c="dimmed">
+                Season/Episode
+              </Text>
+              <Text size="sm" fw={700}>
+                {seLabel}
+              </Text>
+            </Group>
+          )}
           <Group justify="space-between">
             <Text size="sm" c="dimmed">
               Channel

From 68de1b07231fadf0fd6076330070a7c898e59ae8 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Wed, 4 Mar 2026 10:17:26 -0600
Subject: [PATCH 44/58] fix(dvr): add DVR search/filters, fix in-progress
 scheduling, prevent duplicate recordings

- Add search, channel, and status filter controls to DVR page (matching Guide/VODs pattern)
- Fix post_save signal to schedule recordings for currently-playing programs immediately instead of skipping them
- Remove duplicate evaluate_series_rules.delay() call from series rule creation endpoint to prevent race condition that created two recordings for the same program
---
 apps/channels/api_views.py                    |   9 +-
 apps/channels/signals.py                      |  24 +--
 .../tests/test_recording_scheduling.py        |  37 ++++
 frontend/src/pages/DVR.jsx                    | 178 +++++++++++++++---
 frontend/src/pages/__tests__/DVR.test.jsx     |  25 +++
 frontend/src/utils/pages/DVRUtils.js          |  52 +++++
 .../utils/pages/__tests__/DVRUtils.test.js    | 158 ++++++++++++++++
 7 files changed, 436 insertions(+), 47 deletions(-)

diff --git a/apps/channels/api_views.py b/apps/channels/api_views.py
index 2d4e5077..4460c277 100644
--- a/apps/channels/api_views.py
+++ b/apps/channels/api_views.py
@@ -49,7 +49,6 @@ from .serializers import (
 )
 from .tasks import (
     match_epg_channels,
-    evaluate_series_rules,
     evaluate_series_rules_impl,
     match_single_channel_epg,
     match_selected_channels_epg,
@@ -2846,11 +2845,9 @@ class SeriesRulesAPIView(APIView):
         else:
             rules.append({"tvg_id": tvg_id, "mode": mode, "title": title})
         CoreSettings.set_dvr_series_rules(rules)
-        # Evaluate immediately for this tvg_id (async)
-        try:
-            evaluate_series_rules.delay(tvg_id)
-        except Exception:
-            pass
+        # Note: frontend calls the evaluate endpoint explicitly after creating
+        # the rule, so do NOT fire evaluate_series_rules.delay() here to
+        # avoid a race that creates duplicate recordings.
         return Response({"success": True, "rules": rules})
 
 
diff --git a/apps/channels/signals.py b/apps/channels/signals.py
index a42b8c04..dd2ac158 100644
--- a/apps/channels/signals.py
+++ b/apps/channels/signals.py
@@ -191,28 +191,30 @@ def schedule_task_on_save(sender, instance, created, **kwargs):
 
         if not instance.task_id:
             start_time = instance.start_time
+            end_time = instance.end_time
 
-            # Make both datetimes aware (in UTC)
+            # Make datetimes aware (in UTC)
             if not is_aware(start_time):
-                print("Start time was not aware, making aware")
                 start_time = make_aware(start_time)
+            if end_time and not is_aware(end_time):
+                end_time = make_aware(end_time)
 
             current_time = now()
 
-            # Debug log
-            print(f"Start time: {start_time}, Now: {current_time}")
-
-            # Optionally allow slight fudge factor (1 second) to ensure scheduling happens
             if start_time > current_time - timedelta(seconds=1):
-                print("Scheduling recording task!")
-                # Pass the corrected, timezone-aware start_time explicitly so
-                # schedule_recording_task uses it as the Celery ETA rather than
-                # re-reading instance.start_time which may still be naive.
+                # Future recording — schedule at start_time
+                logger.info(f"Recording {instance.id}: scheduling task at {start_time}")
                 task_id = schedule_recording_task(instance, eta=start_time)
                 instance.task_id = task_id
                 instance.save(update_fields=['task_id'])
+            elif end_time and end_time > current_time:
+                # Currently-playing — start immediately (e.g. series rule for in-progress program)
+                logger.info(f"Recording {instance.id}: start_time in past but end_time still future, scheduling immediately")
+                task_id = schedule_recording_task(instance, eta=current_time)
+                instance.task_id = task_id
+                instance.save(update_fields=['task_id'])
             else:
-                print("Start time is in the past. Not scheduling.")
+                logger.info(f"Recording {instance.id}: start_time and end_time both in past, not scheduling")
         # Kick off poster/artwork prefetch to enrich Upcoming cards.
         # Skip when the recording is already active or finished — run_recording
         # handles its own poster resolution, and scheduling artwork prefetch
diff --git a/apps/channels/tests/test_recording_scheduling.py b/apps/channels/tests/test_recording_scheduling.py
index e9daecd7..c615c151 100644
--- a/apps/channels/tests/test_recording_scheduling.py
+++ b/apps/channels/tests/test_recording_scheduling.py
@@ -275,6 +275,43 @@ class SignalIntegrationTests(TestCase):
             PeriodicTask.objects.filter(name__startswith="dvr-recording-").count(), 0
         )
 
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_post_save_schedules_currently_playing_recording(self, mock_artwork):
+        """A recording with past start_time but future end_time schedules immediately."""
+        mock_artwork.apply_async.return_value = MagicMock()
+
+        past_start = timezone.now() - timedelta(minutes=30)
+        future_end = timezone.now() + timedelta(minutes=30)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=past_start,
+            end_time=future_end,
+        )
+
+        rec.refresh_from_db()
+        task_name = f"dvr-recording-{rec.id}"
+        self.assertEqual(rec.task_id, task_name)
+        self.assertTrue(PeriodicTask.objects.filter(name=task_name).exists())
+
+    @patch("apps.channels.signals.prefetch_recording_artwork")
+    def test_post_save_skips_fully_past_recording(self, mock_artwork):
+        """A recording with both start_time and end_time in the past is not scheduled."""
+        mock_artwork.apply_async.return_value = MagicMock()
+
+        past_start = timezone.now() - timedelta(hours=2)
+        past_end = timezone.now() - timedelta(hours=1)
+        rec = Recording.objects.create(
+            channel=self.channel,
+            start_time=past_start,
+            end_time=past_end,
+        )
+
+        rec.refresh_from_db()
+        self.assertIsNone(rec.task_id)
+        self.assertFalse(
+            PeriodicTask.objects.filter(name=f"dvr-recording-{rec.id}").exists()
+        )
+
     @patch("apps.channels.signals.prefetch_recording_artwork")
     def test_pre_save_revokes_on_time_change(self, mock_artwork):
         """Changing a recording's start_time revokes the old task and creates a new one."""
diff --git a/frontend/src/pages/DVR.jsx b/frontend/src/pages/DVR.jsx
index 5289127d..a8e0d6ed 100644
--- a/frontend/src/pages/DVR.jsx
+++ b/frontend/src/pages/DVR.jsx
@@ -1,16 +1,20 @@
 import React, { useMemo, useState, useEffect, lazy, Suspense } from 'react';
 import {
+  ActionIcon,
   Box,
   Button,
   Badge,
+  Flex,
   Group,
+  Select,
   SimpleGrid,
   Stack,
   Text,
+  TextInput,
   Title,
   useMantineTheme,
 } from '@mantine/core';
-import { SquarePlus } from 'lucide-react';
+import { Search, SquarePlus, X } from 'lucide-react';
 import useChannelsStore from '../store/channels';
 import API from '../api';
 import useSettingsStore from '../store/settings';
@@ -22,7 +26,7 @@ const RecordingDetailsModal = lazy(
 );
 import RecurringRuleModal from '../components/forms/RecurringRuleModal.jsx';
 import RecordingCard from '../components/cards/RecordingCard.jsx';
-import { categorizeRecordings } from '../utils/pages/DVRUtils.js';
+import { categorizeRecordings, filterRecordings, buildChannelOptions } from '../utils/pages/DVRUtils.js';
 import {
   getChannelLogoUrl,
   getPosterUrl,
@@ -31,6 +35,13 @@ import {
 } from '../utils/cards/RecordingCardUtils.js';
 import ErrorBoundary from '../components/ErrorBoundary.jsx';
 
+const STATUS_OPTIONS = [
+  { value: 'recording', label: 'Recording' },
+  { value: 'scheduled', label: 'Scheduled' },
+  { value: 'completed', label: 'Completed' },
+  { value: 'interrupted', label: 'Interrupted' },
+];
+
 const RecordingList = ({
   list,
   onOpenDetails,
@@ -62,6 +73,11 @@ const DVRPage = () => {
   const [ruleModal, setRuleModal] = useState({ open: false, ruleId: null });
   const [editRecording, setEditRecording] = useState(null);
 
+  // Filter state
+  const [searchQuery, setSearchQuery] = useState('');
+  const [selectedChannelId, setSelectedChannelId] = useState(null);
+  const [selectedStatus, setSelectedStatus] = useState(null);
+
   const openRecordingModal = () => {
     setRecordingModalOpen(true);
   };
@@ -130,6 +146,43 @@ const DVRPage = () => {
     return categorizeRecordings(recordings, toUserTime, now);
   }, [recordings, now, toUserTime]);
 
+  // Channel options for filter dropdown (from raw unfiltered data)
+  const channelOptions = useMemo(() => {
+    return buildChannelOptions(channelsById, inProgress, upcoming, completed);
+  }, [channelsById, inProgress, upcoming, completed]);
+
+  // Filtered buckets
+  const hasActiveFilters = searchQuery !== '' || selectedChannelId !== null || selectedStatus !== null;
+
+  const filteredInProgress = useMemo(() => {
+    if (selectedStatus && selectedStatus !== 'recording') return [];
+    return filterRecordings(inProgress, searchQuery, selectedChannelId);
+  }, [inProgress, searchQuery, selectedChannelId, selectedStatus]);
+
+  const filteredUpcoming = useMemo(() => {
+    if (selectedStatus && selectedStatus !== 'scheduled') return [];
+    return filterRecordings(upcoming, searchQuery, selectedChannelId);
+  }, [upcoming, searchQuery, selectedChannelId, selectedStatus]);
+
+  const filteredCompleted = useMemo(() => {
+    if (selectedStatus && !['completed', 'interrupted'].includes(selectedStatus)) return [];
+    let filtered = filterRecordings(completed, searchQuery, selectedChannelId);
+    if (selectedStatus === 'interrupted') {
+      filtered = filtered.filter((rec) => rec.custom_properties?.status === 'interrupted');
+    } else if (selectedStatus === 'completed') {
+      // "Completed" includes both completed and stopped recordings
+      filtered = filtered.filter((rec) => rec.custom_properties?.status !== 'interrupted');
+    }
+    return filtered;
+  }, [completed, searchQuery, selectedChannelId, selectedStatus]);
+
+  // Filter handlers
+  const clearFilters = () => {
+    setSearchQuery('');
+    setSelectedChannelId(null);
+    setSelectedStatus(null);
+  };
+
   const handleOnWatchLive = () => {
     const rec = detailsRecording;
     const now = userNow();
@@ -166,26 +219,77 @@ const DVRPage = () => {
   };
   return (
     <Box p={10}>
-      <Button
-        leftSection={<SquarePlus size={18} />}
-        variant="light"
-        size="sm"
-        onClick={openRecordingModal}
-        p={5}
-        color={theme.tailwind.green[5]}
-        style={{
-          borderWidth: '1px',
-          borderColor: theme.tailwind.green[5],
-          color: 'white',
-        }}
-      >
-        New Recording
-      </Button>
-      <Stack gap="lg" pt={12}>
+      <Flex gap="md" align="center" wrap="wrap" mb={12}>
+        <Button
+          leftSection={<SquarePlus size={18} />}
+          variant="light"
+          size="sm"
+          onClick={openRecordingModal}
+          p={5}
+          color={theme.tailwind.green[5]}
+          style={{
+            borderWidth: '1px',
+            borderColor: theme.tailwind.green[5],
+            color: 'white',
+          }}
+        >
+          New Recording
+        </Button>
+
+        <TextInput
+          placeholder="Search recordings..."
+          value={searchQuery}
+          onChange={(e) => setSearchQuery(e.target.value)}
+          w={'250px'}
+          leftSection={<Search size={16} />}
+          rightSection={
+            searchQuery ? (
+              <ActionIcon
+                onClick={() => setSearchQuery('')}
+                variant="subtle"
+                color="gray"
+                size="sm"
+              >
+                <X size={14} />
+              </ActionIcon>
+            ) : null
+          }
+        />
+
+        <Select
+          placeholder="Filter by channel"
+          data={channelOptions}
+          value={selectedChannelId}
+          onChange={setSelectedChannelId}
+          w={'220px'}
+          clearable
+          searchable
+        />
+
+        <Select
+          placeholder="Filter by status"
+          data={STATUS_OPTIONS}
+          value={selectedStatus}
+          onChange={setSelectedStatus}
+          w={'180px'}
+          clearable
+        />
+
+        {hasActiveFilters && (
+          <Button variant="subtle" onClick={clearFilters} size="sm">
+            Clear Filters
+          </Button>
+        )}
+      </Flex>
+      <Stack gap="lg">
         <div>
           <Group justify="space-between" mb={8}>
             <Title order={4}>Currently Recording</Title>
-            <Badge color="red.6">{inProgress.length}</Badge>
+            <Badge color="red.6">
+              {hasActiveFilters
+                ? `${filteredInProgress.length} / ${inProgress.length}`
+                : inProgress.length}
+            </Badge>
           </Group>
           <SimpleGrid
             cols={3}
@@ -197,15 +301,17 @@ const DVRPage = () => {
           >
             {
               <RecordingList
-                list={inProgress}
+                list={filteredInProgress}
                 onOpenDetails={openDetails}
                 onOpenRecurring={openRuleModal}
                 channelsById={channelsById}
               />
             }
-            {inProgress.length === 0 && (
+            {filteredInProgress.length === 0 && (
               <Text size="sm" c="dimmed">
-                Nothing recording right now.
+                {hasActiveFilters
+                  ? 'No recordings match your filters.'
+                  : 'Nothing recording right now.'}
               </Text>
             )}
           </SimpleGrid>
@@ -214,7 +320,11 @@ const DVRPage = () => {
         <div>
           <Group justify="space-between" mb={8}>
             <Title order={4}>Upcoming Recordings</Title>
-            <Badge color="yellow.6">{upcoming.length}</Badge>
+            <Badge color="yellow.6">
+              {hasActiveFilters
+                ? `${filteredUpcoming.length} / ${upcoming.length}`
+                : upcoming.length}
+            </Badge>
           </Group>
           <SimpleGrid
             cols={3}
@@ -226,15 +336,17 @@ const DVRPage = () => {
           >
             {
               <RecordingList
-                list={upcoming}
+                list={filteredUpcoming}
                 onOpenDetails={openDetails}
                 onOpenRecurring={openRuleModal}
                 channelsById={channelsById}
               />
             }
-            {upcoming.length === 0 && (
+            {filteredUpcoming.length === 0 && (
               <Text size="sm" c="dimmed">
-                No upcoming recordings.
+                {hasActiveFilters
+                  ? 'No recordings match your filters.'
+                  : 'No upcoming recordings.'}
               </Text>
             )}
           </SimpleGrid>
@@ -243,7 +355,11 @@ const DVRPage = () => {
         <div>
           <Group justify="space-between" mb={8}>
             <Title order={4}>Previously Recorded</Title>
-            <Badge color="gray.6">{completed.length}</Badge>
+            <Badge color="gray.6">
+              {hasActiveFilters
+                ? `${filteredCompleted.length} / ${completed.length}`
+                : completed.length}
+            </Badge>
           </Group>
           <SimpleGrid
             cols={3}
@@ -255,15 +371,17 @@ const DVRPage = () => {
           >
             {
               <RecordingList
-                list={completed}
+                list={filteredCompleted}
                 onOpenDetails={openDetails}
                 onOpenRecurring={openRuleModal}
                 channelsById={channelsById}
               />
             }
-            {completed.length === 0 && (
+            {filteredCompleted.length === 0 && (
               <Text size="sm" c="dimmed">
-                No completed recordings yet.
+                {hasActiveFilters
+                  ? 'No recordings match your filters.'
+                  : 'No completed recordings yet.'}
               </Text>
             )}
           </SimpleGrid>
diff --git a/frontend/src/pages/__tests__/DVR.test.jsx b/frontend/src/pages/__tests__/DVR.test.jsx
index 15c68f5d..2f5e5021 100644
--- a/frontend/src/pages/__tests__/DVR.test.jsx
+++ b/frontend/src/pages/__tests__/DVR.test.jsx
@@ -33,10 +33,35 @@ vi.mock('../../api');
 
 // Mock Mantine components
 vi.mock('@mantine/core', () => ({
+  ActionIcon: ({ children, onClick }) => (
+    <button data-testid="action-icon" onClick={onClick}>{children}</button>
+  ),
   Box: ({ children }) => <div data-testid="box">{children}</div>,
   Container: ({ children }) => <div data-testid="container">{children}</div>,
+  Flex: ({ children }) => <div data-testid="flex">{children}</div>,
   Title: ({ children, order }) => <h1 data-order={order}>{children}</h1>,
   Text: ({ children }) => <p>{children}</p>,
+  TextInput: ({ placeholder, value, onChange, ...props }) => (
+    <input
+      data-testid="text-input"
+      placeholder={placeholder}
+      value={value}
+      onChange={onChange}
+    />
+  ),
+  Select: ({ placeholder, data, value, onChange, ...props }) => (
+    <select
+      data-testid="select"
+      value={value || ''}
+      onChange={(e) => onChange(e.target.value || null)}
+      aria-label={placeholder}
+    >
+      <option value="">{placeholder}</option>
+      {(data || []).map((opt) => (
+        <option key={opt.value} value={opt.value}>{opt.label}</option>
+      ))}
+    </select>
+  ),
   Button: ({ children, onClick, leftSection, loading, ...props }) => (
     <button onClick={onClick} disabled={loading} {...props}>
       {leftSection}
diff --git a/frontend/src/utils/pages/DVRUtils.js b/frontend/src/utils/pages/DVRUtils.js
index fb80dd15..e1082a07 100644
--- a/frontend/src/utils/pages/DVRUtils.js
+++ b/frontend/src/utils/pages/DVRUtils.js
@@ -88,3 +88,55 @@ export const categorizeRecordings = (recordings, toUserTime, now) => {
     completed,
   };
 };
+
+export const filterRecordings = (recordings, searchQuery, selectedChannelId) => {
+  return recordings.filter((rec) => {
+    if (searchQuery) {
+      const q = searchQuery.toLowerCase();
+      const cp = rec.custom_properties || {};
+      const program = cp.program || {};
+      const title = (program.title || '').toLowerCase();
+      const subTitle = (program.sub_title || '').toLowerCase();
+      const description = (program.description || cp.description || '').toLowerCase();
+
+      if (!title.includes(q) && !subTitle.includes(q) && !description.includes(q)) {
+        return false;
+      }
+    }
+
+    if (selectedChannelId) {
+      if (String(rec.channel) !== String(selectedChannelId)) return false;
+    }
+
+    return true;
+  });
+};
+
+export const buildChannelOptions = (channelsById, ...buckets) => {
+  const channelIds = new Set();
+  for (const bucket of buckets) {
+    for (const rec of bucket) {
+      if (rec.channel != null) channelIds.add(rec.channel);
+    }
+  }
+
+  const options = [];
+  for (const id of channelIds) {
+    const ch = channelsById[id];
+    if (ch) {
+      options.push({
+        value: String(ch.id),
+        label: ch.channel_number ? `${ch.channel_number} - ${ch.name}` : ch.name,
+      });
+    }
+  }
+
+  options.sort((a, b) => {
+    const aNum = parseInt(a.label, 10);
+    const bNum = parseInt(b.label, 10);
+    if (!isNaN(aNum) && !isNaN(bNum)) return aNum - bNum;
+    return a.label.localeCompare(b.label);
+  });
+
+  return options;
+};
diff --git a/frontend/src/utils/pages/__tests__/DVRUtils.test.js b/frontend/src/utils/pages/__tests__/DVRUtils.test.js
index dd76784a..beeb57ed 100644
--- a/frontend/src/utils/pages/__tests__/DVRUtils.test.js
+++ b/frontend/src/utils/pages/__tests__/DVRUtils.test.js
@@ -541,4 +541,162 @@ describe('DVRUtils', () => {
       expect(result.upcoming[0]._group_count).toBe(1);
     });
   });
+
+  describe('filterRecordings', () => {
+    const makeRec = (id, title, subTitle, description, channel) => ({
+      id,
+      channel,
+      custom_properties: {
+        program: { title, sub_title: subTitle, description },
+      },
+    });
+
+    it('returns all recordings when no filters active', () => {
+      const recs = [makeRec(1, 'News', 'Episode 1', 'Daily news', 5)];
+      expect(DVRUtils.filterRecordings(recs, '', null)).toEqual(recs);
+    });
+
+    it('filters by title case-insensitively', () => {
+      const recs = [
+        makeRec(1, 'Morning News', '', '', 1),
+        makeRec(2, 'Sports Center', '', '', 1),
+      ];
+      const result = DVRUtils.filterRecordings(recs, 'news', null);
+      expect(result).toHaveLength(1);
+      expect(result[0].id).toBe(1);
+    });
+
+    it('filters by sub_title', () => {
+      const recs = [
+        makeRec(1, 'Show A', 'Pilot Episode', '', 1),
+        makeRec(2, 'Show B', 'Finale', '', 1),
+      ];
+      const result = DVRUtils.filterRecordings(recs, 'pilot', null);
+      expect(result).toHaveLength(1);
+      expect(result[0].id).toBe(1);
+    });
+
+    it('filters by description', () => {
+      const recs = [
+        makeRec(1, 'Show A', '', 'A thriller about detectives', 1),
+        makeRec(2, 'Show B', '', 'A comedy special', 1),
+      ];
+      const result = DVRUtils.filterRecordings(recs, 'detective', null);
+      expect(result).toHaveLength(1);
+      expect(result[0].id).toBe(1);
+    });
+
+    it('filters by channel id', () => {
+      const recs = [
+        makeRec(1, 'Show A', '', '', 5),
+        makeRec(2, 'Show B', '', '', 10),
+      ];
+      const result = DVRUtils.filterRecordings(recs, '', '5');
+      expect(result).toHaveLength(1);
+      expect(result[0].id).toBe(1);
+    });
+
+    it('combines search and channel filter with AND logic', () => {
+      const recs = [
+        makeRec(1, 'News', '', '', 5),
+        makeRec(2, 'News', '', '', 10),
+        makeRec(3, 'Sports', '', '', 5),
+      ];
+      const result = DVRUtils.filterRecordings(recs, 'news', '5');
+      expect(result).toHaveLength(1);
+      expect(result[0].id).toBe(1);
+    });
+
+    it('handles recordings without custom_properties', () => {
+      const recs = [{ id: 1, channel: 5 }];
+      const result = DVRUtils.filterRecordings(recs, 'anything', null);
+      expect(result).toHaveLength(0);
+    });
+
+    it('handles recordings without custom_properties and no search', () => {
+      const recs = [{ id: 1, channel: 5 }];
+      const result = DVRUtils.filterRecordings(recs, '', null);
+      expect(result).toHaveLength(1);
+    });
+
+    it('handles empty array', () => {
+      expect(DVRUtils.filterRecordings([], 'test', '5')).toEqual([]);
+    });
+
+    it('falls back to custom_properties.description', () => {
+      const recs = [{
+        id: 1,
+        channel: 1,
+        custom_properties: {
+          description: 'A fallback description',
+          program: { title: 'Show' },
+        },
+      }];
+      const result = DVRUtils.filterRecordings(recs, 'fallback', null);
+      expect(result).toHaveLength(1);
+    });
+  });
+
+  describe('buildChannelOptions', () => {
+    const channelsById = {
+      5: { id: 5, name: 'CNN', channel_number: '5' },
+      10: { id: 10, name: 'ESPN', channel_number: '10' },
+      3: { id: 3, name: 'NBC', channel_number: '3' },
+      20: { id: 20, name: 'Local Access' },
+    };
+
+    it('returns only channels present in recordings', () => {
+      const bucket = [{ channel: 5 }, { channel: 10 }];
+      const result = DVRUtils.buildChannelOptions(channelsById, bucket);
+      expect(result).toHaveLength(2);
+      expect(result.map((o) => o.value)).toEqual(['5', '10']);
+    });
+
+    it('sorts by channel number numerically', () => {
+      const bucket = [{ channel: 10 }, { channel: 3 }, { channel: 5 }];
+      const result = DVRUtils.buildChannelOptions(channelsById, bucket);
+      expect(result[0].label).toBe('3 - NBC');
+      expect(result[1].label).toBe('5 - CNN');
+      expect(result[2].label).toBe('10 - ESPN');
+    });
+
+    it('handles multiple buckets', () => {
+      const b1 = [{ channel: 5 }];
+      const b2 = [{ channel: 10 }];
+      const b3 = [{ channel: 3 }];
+      const result = DVRUtils.buildChannelOptions(channelsById, b1, b2, b3);
+      expect(result).toHaveLength(3);
+    });
+
+    it('deduplicates channels across buckets', () => {
+      const b1 = [{ channel: 5 }, { channel: 10 }];
+      const b2 = [{ channel: 5 }, { channel: 3 }];
+      const result = DVRUtils.buildChannelOptions(channelsById, b1, b2);
+      expect(result).toHaveLength(3);
+    });
+
+    it('skips channels not in channelsById', () => {
+      const bucket = [{ channel: 5 }, { channel: 999 }];
+      const result = DVRUtils.buildChannelOptions(channelsById, bucket);
+      expect(result).toHaveLength(1);
+      expect(result[0].value).toBe('5');
+    });
+
+    it('formats label with channel number when available', () => {
+      const bucket = [{ channel: 5 }];
+      const result = DVRUtils.buildChannelOptions(channelsById, bucket);
+      expect(result[0].label).toBe('5 - CNN');
+    });
+
+    it('formats label without channel number when missing', () => {
+      const bucket = [{ channel: 20 }];
+      const result = DVRUtils.buildChannelOptions(channelsById, bucket);
+      expect(result[0].label).toBe('Local Access');
+    });
+
+    it('returns empty array for empty buckets', () => {
+      const result = DVRUtils.buildChannelOptions(channelsById, [], []);
+      expect(result).toEqual([]);
+    });
+  });
 });

From a21ec41f8de6b4dd29f0503353ad7e2d23f7661d Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Wed, 4 Mar 2026 15:05:45 -0600
Subject: [PATCH 45/58] fix(guide): show recording indicator for
 completed/stopped recordings

The red dot on the TV guide was filtering out all terminal recordings (completed, stopped, interrupted, failed). Now only interrupted and failed recordings are excluded, so users can see at a glance which past shows were successfully recorded.
---
 .../src/pages/__tests__/guideUtils.test.js    | 26 +++++++++++++++++++
 frontend/src/pages/guideUtils.js              |  6 ++---
 2 files changed, 29 insertions(+), 3 deletions(-)

diff --git a/frontend/src/pages/__tests__/guideUtils.test.js b/frontend/src/pages/__tests__/guideUtils.test.js
index 8985be0d..4f1ef780 100644
--- a/frontend/src/pages/__tests__/guideUtils.test.js
+++ b/frontend/src/pages/__tests__/guideUtils.test.js
@@ -705,6 +705,32 @@ describe('guideUtils', () => {
 
       expect(result.size).toBe(0);
     });
+
+    it('should include completed and stopped recordings', () => {
+      const recordings = [
+        { id: 1, custom_properties: { program: { id: 'p1' }, status: 'completed' } },
+        { id: 2, custom_properties: { program: { id: 'p2' }, status: 'stopped' } },
+      ];
+
+      const result = guideUtils.mapRecordingsByProgramId(recordings);
+
+      expect(result.size).toBe(2);
+      expect(result.get('p1').id).toBe(1);
+      expect(result.get('p2').id).toBe(2);
+    });
+
+    it('should exclude interrupted and failed recordings', () => {
+      const recordings = [
+        { id: 1, custom_properties: { program: { id: 'p1' }, status: 'interrupted' } },
+        { id: 2, custom_properties: { program: { id: 'p2' }, status: 'failed' } },
+        { id: 3, custom_properties: { program: { id: 'p3' }, status: 'recording' } },
+      ];
+
+      const result = guideUtils.mapRecordingsByProgramId(recordings);
+
+      expect(result.size).toBe(1);
+      expect(result.get('p3').id).toBe(3);
+    });
   });
 
   describe('formatTime', () => {
diff --git a/frontend/src/pages/guideUtils.js b/frontend/src/pages/guideUtils.js
index 239adcfa..940c12b7 100644
--- a/frontend/src/pages/guideUtils.js
+++ b/frontend/src/pages/guideUtils.js
@@ -214,15 +214,15 @@ export const mapChannelsById = (guideChannels) => {
   return map;
 };
 
-const _terminalStatuses = new Set(['stopped', 'completed', 'interrupted', 'failed']);
+const _excludedStatuses = new Set(['interrupted', 'failed']);
 
 export const mapRecordingsByProgramId = (recordings) => {
   const map = new Map();
   (recordings || []).forEach((recording) => {
     const programId = recording?.custom_properties?.program?.id;
     const status = recording?.custom_properties?.status;
-    // Only show indicator for pending/active recordings, not terminal ones
-    if (programId != null && !_terminalStatuses.has(status)) {
+    // Show indicator for all recordings except interrupted/failed
+    if (programId != null && !_excludedStatuses.has(status)) {
       map.set(programId, recording);
     }
   });

From 5f06bd28ba9710d12d5cac2119ab1db9c9f6bd3f Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Wed, 4 Mar 2026 15:20:41 -0600
Subject: [PATCH 46/58] fix(guide): restrict recording indicator to
 pending/active recordings

Only show the red recording dot on the TV guide for scheduled and in-progress recordings. Completed, stopped, interrupted, and failed recordings are excluded. Including terminal recordings caused false indicators on unrelated channels due to recycled dummy program IDs for channels without real EPG data.
---
 .../src/pages/__tests__/guideUtils.test.js    | 23 ++++++-------------
 frontend/src/pages/guideUtils.js              |  6 ++---
 2 files changed, 10 insertions(+), 19 deletions(-)

diff --git a/frontend/src/pages/__tests__/guideUtils.test.js b/frontend/src/pages/__tests__/guideUtils.test.js
index 4f1ef780..59967372 100644
--- a/frontend/src/pages/__tests__/guideUtils.test.js
+++ b/frontend/src/pages/__tests__/guideUtils.test.js
@@ -706,30 +706,21 @@ describe('guideUtils', () => {
       expect(result.size).toBe(0);
     });
 
-    it('should include completed and stopped recordings', () => {
+    it('should exclude terminal status recordings', () => {
       const recordings = [
         { id: 1, custom_properties: { program: { id: 'p1' }, status: 'completed' } },
         { id: 2, custom_properties: { program: { id: 'p2' }, status: 'stopped' } },
+        { id: 3, custom_properties: { program: { id: 'p3' }, status: 'interrupted' } },
+        { id: 4, custom_properties: { program: { id: 'p4' }, status: 'failed' } },
+        { id: 5, custom_properties: { program: { id: 'p5' }, status: 'recording' } },
+        { id: 6, custom_properties: { program: { id: 'p6' } } },
       ];
 
       const result = guideUtils.mapRecordingsByProgramId(recordings);
 
       expect(result.size).toBe(2);
-      expect(result.get('p1').id).toBe(1);
-      expect(result.get('p2').id).toBe(2);
-    });
-
-    it('should exclude interrupted and failed recordings', () => {
-      const recordings = [
-        { id: 1, custom_properties: { program: { id: 'p1' }, status: 'interrupted' } },
-        { id: 2, custom_properties: { program: { id: 'p2' }, status: 'failed' } },
-        { id: 3, custom_properties: { program: { id: 'p3' }, status: 'recording' } },
-      ];
-
-      const result = guideUtils.mapRecordingsByProgramId(recordings);
-
-      expect(result.size).toBe(1);
-      expect(result.get('p3').id).toBe(3);
+      expect(result.get('p5').id).toBe(5);
+      expect(result.get('p6').id).toBe(6);
     });
   });
 
diff --git a/frontend/src/pages/guideUtils.js b/frontend/src/pages/guideUtils.js
index 940c12b7..239adcfa 100644
--- a/frontend/src/pages/guideUtils.js
+++ b/frontend/src/pages/guideUtils.js
@@ -214,15 +214,15 @@ export const mapChannelsById = (guideChannels) => {
   return map;
 };
 
-const _excludedStatuses = new Set(['interrupted', 'failed']);
+const _terminalStatuses = new Set(['stopped', 'completed', 'interrupted', 'failed']);
 
 export const mapRecordingsByProgramId = (recordings) => {
   const map = new Map();
   (recordings || []).forEach((recording) => {
     const programId = recording?.custom_properties?.program?.id;
     const status = recording?.custom_properties?.status;
-    // Show indicator for all recordings except interrupted/failed
-    if (programId != null && !_excludedStatuses.has(status)) {
+    // Only show indicator for pending/active recordings, not terminal ones
+    if (programId != null && !_terminalStatuses.has(status)) {
       map.set(programId, recording);
     }
   });

From 06f864ad6dfa6e4913b7624b9af767fd90bfd193 Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Wed, 4 Mar 2026 17:37:42 -0600
Subject: [PATCH 47/58] fix(stream_generator): optimize resource and health
 checks with throttling

---
 apps/proxy/ts_proxy/stream_generator.py | 101 +++++++++++++++---------
 1 file changed, 65 insertions(+), 36 deletions(-)

diff --git a/apps/proxy/ts_proxy/stream_generator.py b/apps/proxy/ts_proxy/stream_generator.py
index f66572d6..80329a46 100644
--- a/apps/proxy/ts_proxy/stream_generator.py
+++ b/apps/proxy/ts_proxy/stream_generator.py
@@ -58,6 +58,19 @@ class StreamGenerator:
         self.last_ttl_refresh = time.time()
         self.ttl_refresh_interval = 3  # Refresh TTL every 3 seconds of active streaming
 
+        # Cached proxy server reference
+        self.proxy_server = None
+
+        # Non-owner health check throttle: avoid Redis GET on every loop iteration
+        self._last_health_check_time = 0.0
+        self._last_health_check_result = False
+        self._health_check_interval = 2.0  # seconds
+
+        # Resource check throttle: Redis stop/state checks are expensive; throttle
+        # them while allowing cheap in-memory checks to run every iteration.
+        self._last_resource_check_time = 0.0
+        self._resource_check_interval = 1.0  # seconds
+
     def generate(self):
         """
         Generator function that produces the stream content for the client.
@@ -192,6 +205,7 @@ class StreamGenerator:
         self.local_index = max(0, current_buffer_index - initial_behind)
 
         # Store important objects as instance variables
+        self.proxy_server = proxy_server
         self.buffer = buffer
         self.stream_manager = stream_manager
         self.last_yield_time = time.time()
@@ -265,9 +279,7 @@ class StreamGenerator:
 
     def _check_resources(self):
         """Check if required resources still exist."""
-        proxy_server = ProxyServer.get_instance()
-
-        # Enhanced resource checks
+        proxy_server = self.proxy_server or ProxyServer.get_instance()
         if self.channel_id not in proxy_server.stream_buffers:
             logger.info(f"[{self.client_id}] Channel buffer no longer exists, terminating stream")
             return False
@@ -276,35 +288,43 @@ class StreamGenerator:
             logger.info(f"[{self.client_id}] Client manager no longer exists, terminating stream")
             return False
 
-        # Check if this specific client has been stopped (Redis keys, etc.)
-        if proxy_server.redis_client:
-            # Channel stop check - with extended key set
-            stop_key = RedisKeys.channel_stopping(self.channel_id)
-            if proxy_server.redis_client.exists(stop_key):
-                logger.info(f"[{self.client_id}] Detected channel stop signal, terminating stream")
+        client_manager = proxy_server.client_managers[self.channel_id]
+        if self.client_id not in client_manager.clients:
+            logger.info(f"[{self.client_id}] Client no longer in client manager, terminating stream")
+            return False
+
+        # --- Redis checks: throttled to _resource_check_interval (default 1s) ---
+        # 3 Redis round-trips on every iteration is expensive at stream rates;
+        # stop/state signals change infrequently so a 1-second poll is sufficient.
+        if not proxy_server.redis_client:
+            return True
+
+        now = time.time()
+        if now - self._last_resource_check_time < self._resource_check_interval:
+            return True
+
+        self._last_resource_check_time = now
+
+        # Channel stop check
+        stop_key = RedisKeys.channel_stopping(self.channel_id)
+        if proxy_server.redis_client.exists(stop_key):
+            logger.info(f"[{self.client_id}] Detected channel stop signal, terminating stream")
+            return False
+
+        # Channel state in metadata
+        metadata_key = RedisKeys.channel_metadata(self.channel_id)
+        metadata = proxy_server.redis_client.hgetall(metadata_key)
+        if metadata and b'state' in metadata:
+            state = metadata[b'state'].decode('utf-8')
+            if state in ['error', 'stopped', 'stopping']:
+                logger.info(f"[{self.client_id}] Channel in {state} state, terminating stream")
                 return False
 
-            # Also check channel state in metadata
-            metadata_key = RedisKeys.channel_metadata(self.channel_id)
-            metadata = proxy_server.redis_client.hgetall(metadata_key)
-            if metadata and b'state' in metadata:
-                state = metadata[b'state'].decode('utf-8')
-                if state in ['error', 'stopped', 'stopping']:
-                    logger.info(f"[{self.client_id}] Channel in {state} state, terminating stream")
-                    return False
-
-            # Client stop check
-            client_stop_key = RedisKeys.client_stop(self.channel_id, self.client_id)
-            if proxy_server.redis_client.exists(client_stop_key):
-                logger.info(f"[{self.client_id}] Detected client stop signal, terminating stream")
-                return False
-
-            # Also check if client has been removed from client_manager
-            if self.channel_id in proxy_server.client_managers:
-                client_manager = proxy_server.client_managers[self.channel_id]
-                if self.client_id not in client_manager.clients:
-                    logger.info(f"[{self.client_id}] Client no longer in client manager, terminating stream")
-                    return False
+        # Client stop check
+        client_stop_key = RedisKeys.client_stop(self.channel_id, self.client_id)
+        if proxy_server.redis_client.exists(client_stop_key):
+            logger.info(f"[{self.client_id}] Detected client stop signal, terminating stream")
+            return False
 
         return True
 
@@ -313,7 +333,7 @@ class StreamGenerator:
         # Process and send chunks
         total_size = sum(len(c) for c in chunks)
         logger.debug(f"[{self.client_id}] Retrieved {len(chunks)} chunks ({total_size} bytes) from index {self.local_index+1} to {next_index}")
-        proxy_server = ProxyServer.get_instance()
+        proxy_server = self.proxy_server or ProxyServer.get_instance()
 
         # Send the chunks to the client
         for chunk in chunks:
@@ -391,16 +411,25 @@ class StreamGenerator:
             # Non-owner worker: stream_manager only exists in the owner process.
             # Approximate health from the Redis last_data timestamp; if stale
             # beyond CONNECTION_TIMEOUT, send keepalives to prevent DVR timeout.
+            # Throttled: only re-query Redis every _health_check_interval seconds
+            # to avoid a Redis GET on every loop iteration during sustained waits.
+            now = time.time()
+            if now - self._last_health_check_time < self._health_check_interval:
+                return self._last_health_check_result
             try:
-                proxy_server = ProxyServer.get_instance()
+                proxy_server = self.proxy_server or ProxyServer.get_instance()
                 if proxy_server.redis_client:
                     raw = proxy_server.redis_client.get(RedisKeys.last_data(self.channel_id))
                     if raw:
-                        age = time.time() - float(raw)
+                        age = now - float(raw)
                         timeout_threshold = getattr(Config, 'CONNECTION_TIMEOUT', 10)
-                        return age >= timeout_threshold
-                    # No timestamp in Redis → key missing or expired → unhealthy
-                    return True
+                        result = age >= timeout_threshold
+                    else:
+                        # No timestamp in Redis → key missing or expired → unhealthy
+                        result = True
+                    self._last_health_check_time = now
+                    self._last_health_check_result = result
+                    return result
             except Exception:
                 pass
             return False

From 4dfdafd1d797be0089dbded8ea7d2849e1cc70c0 Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Thu, 5 Mar 2026 11:54:03 -0600
Subject: [PATCH 48/58] changelog: Update changelog for DVR PR

---
 CHANGELOG.md | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 88f6797a..c6fc1f9d 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -11,6 +11,14 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 - New Client Buffer proxy setting: new clients joining an active channel are now positioned a configurable number of seconds behind live rather than a fixed chunk count. The start position is determined by wall-clock chunk receive time (stored as a Redis sorted set alongside the buffer), so the buffer depth is consistent in seconds regardless of stream bitrate. Setting the value to `0` starts clients at live with no buffer. Defaults to 2 seconds. Existing chunk-count gating for the first client connecting to a channel is unchanged. The setting is exposed in Settings → Proxy as "New Client Buffer (seconds)".
 - Channel table filter for channels that have stale streams: A new "Has Stale Streams" filter option in the channel table header menu highlights and filters channels containing at least one stale stream. Channels with stale streams are visually distinguished with an orange tint. The filter is mutually exclusive with "Only Empty Channels". - Thanks [@JCBird1012](https://github.com/JCBird1012)
+- DVR enhancements — Thanks [@CodeBormen](https://github.com/CodeBormen)
+  - **Stop Recording**: A new Stop button (distinct from Cancel) cleanly ends an in-progress recording early and keeps the partial file available for playback. The API returns immediately; stream teardown and task revocation happen in a background thread to prevent 504 timeouts. When multiple recordings run simultaneously, stopping one only terminates that recording's proxy client by ID, leaving all others unaffected. (Closes #454)
+  - **Extend Recording**: In-progress recordings can be extended by 15, 30, or 60 minutes without interrupting the stream.
+  - **Inline metadata editing**: Title and description can now be edited directly in the recording details modal.
+  - **Refresh artwork button**: Manually re-run poster resolution on demand from the recording card.
+  - **Multi-source poster resolution**: Added pipeline querying EPG, VOD, TMDB, OMDb, TVMaze, and iTunes for richer recording artwork.
+  - **Series rules for currently-airing episodes**: Series rules now capture currently-airing episodes in addition to future scheduled ones. (Closes #473)
+  - **Search and filter controls**: Added search and filter controls to the recordings list.
 
 ### Fixed
 
@@ -31,6 +39,26 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   - **No recovery from expired ownership**: `get_channel_owner()` called `redis.get()` twice inside a lambda (TOCTOU race — key could expire between calls); `extend_ownership()` silently returned `False` on expiry with no re-acquisition; and the non-owner cleanup path unconditionally killed streams even when the worker held the `stream_manager`. Fixed with a single `GET` in `get_channel_owner()`, re-acquisition via atomic `SET NX EX` in `extend_ownership()`, and a re-acquisition attempt with client-aware cleanup deferral in the cleanup thread.
 - `get_instance()` deadlock: if `ProxyServer()` raised an exception during singleton construction, `_instance` was left permanently as the `_INITIALIZING` sentinel, causing all subsequent `get_instance()` callers to spin in an infinite `gevent.sleep()` loop. Construction is now wrapped in `try/except`; on failure `_instance` resets to `None` so the next call can retry.
 - Non-atomic ownership acquisition in `try_acquire_ownership()`: replaced the separate `setnx()` + `expire()` calls with a single atomic `SET NX EX`, eliminating the race window where a process crash between the two calls could leave an ownership key with no TTL (permanent ownership lock).
+- DVR bug fixes — Thanks [@CodeBormen](https://github.com/CodeBormen)
+  - **Duplicate recording execution**: Shared `ClockedSchedule` objects were firing `run_recording` multiple times for the same recording, producing corrupted output files. `run_recording` now exits immediately if the recording is already in progress, completed, or stopped (failing closed on DB errors to prevent duplicate tasks from overwriting valid recordings). `revoke_task()` also cleans up both the `PeriodicTask` and its `ClockedSchedule` on execution to eliminate orphaned schedule objects. (Fixes #641)
+  - **Long-range scheduling**: Recordings scheduled more than one hour out were handed off to multiple workers simultaneously because Redis' 3600 s timer limit triggered the job more than once, corrupting the output file. Migrated to `ClockedSchedule` for database-backed persistence that survives restarts and upgrades and avoids the multi-worker race entirely. (Fixes #940)
+  - **Stream reconnection resilience**: Recordings now survive transient network drops with automatic reconnection retrying up to 5 times and appending to the existing file. DB operations use exponential-backoff retry for transient database errors throughout the recording lifecycle.
+  - **Crash recovery pipeline**: On worker restart, recordings stuck in "recording" status have their segments concatenated and remuxed. Remux sanity checks reject MKV output that is less than 50% the size of a previous MKV (duplicate-task overwrite) or less than 10% of the source TS (corrupt first attempt); the source `.ts` is preserved for manual recovery on all failure paths. (Fixes #619, #624)
+  - **Output file collision**: Fixed collision when multiple tasks targeted the same filename.
+  - **WebSocket deadlock**: `send_websocket_update()` was deadlocking the gevent event loop, causing one recording's WebSocket events to block all other simultaneous recordings.
+  - **DVR client isolation**: Stop and Cancel operations now identify the target client by recording ID (via `User-Agent: Dispatcharr-DVR/recording-{id}`), ensuring only the correct proxy client is torn down and never affecting other recordings on the same channel.
+  - **Accidental stream termination on delete**: `destroy()` now only calls `_stop_dvr_clients()` for in-progress recordings, preventing stream termination when deleting a completed recording.
+  - **Recording card logos**: Logos were not displaying due to a channel summary API shape mismatch.
+  - **Logo fetch negative cache**: Added negative cache for failed remote logo fetches so dead CDNs no longer block Daphne workers on repeated requests.
+  - **Artwork fuzzy-match sanitisation**: Poster artwork fuzzy-matching against external APIs (TMDB, OMDb, etc.) was producing incorrect results for channels with names like "USA A&E SD\*"; channel-name strings are now sanitised before querying external sources.
+  - **Series modal "No upcoming episodes"**: Fixed due to a missing `_group_count` merge and an incorrect time filter.
+  - **Series rule cleanup**: Deleting a series rule left orphaned recordings and stale Guide indicators; rule deletion now cleans up all associated recordings. Orphaned recordings with no parent rule are also cleaned up automatically. (Fixes #1041)
+  - **Series rule timezone calculation**: Recurring rules silently dropped scheduled recordings for users in UTC-negative timezones after 4 pm local time. (Fixes #1042)
+  - **Recording modal TDZ crash**: Modal crashed on load in production bundles due to a Temporal Dead Zone error — editing state was referenced before its declaration in the minified bundle.
+  - **Description textarea focus loss**: The description textarea lost focus immediately when opened because the inline editing component was remounting on every render.
+  - **WebSocket-driven refresh**: Replaced all manual `fetchRecordings()` polling calls with debounced WebSocket-driven refresh so the recordings list stays up to date without redundant API requests.
+  - **comskip exit code handling**: comskip treated exit code 1 ("no commercials found") as a fatal error, causing post-processing to fail on clean recordings. Exit code 1 is now recognised as a successful no-op.
+  - **Differentiated WebSocket notification events**: `recording_stopped`, `recording_cancelled` (in-progress cancel), and `recording_deleted` with a `was_in_progress` flag now allow the frontend to display distinct "Recording stopped", "Recording cancelled", and "Recording deleted" toasts.
 
 ## [0.20.2] - 2026-03-03
 

From f9d4f11cb20d723e6fd981df0680a873e7c6a4fb Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Thu, 5 Mar 2026 12:25:38 -0600
Subject: [PATCH 49/58] Bug Fix: update EPG data filtering to use end_time
 instead of start_time

---
 CHANGELOG.md         | 1 +
 apps/output/views.py | 2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 88f6797a..7691dafa 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -14,6 +14,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- EPG output was filtering programs using `start_time__gte=now` when the `days` parameter was specified, which caused currently-airing programs (started before the request time but not yet ended) to be omitted from the XML output. This produced a gap in clients' guides immediately after an EPG refresh, lasting until the next program started. Fixed by changing the filter to `end_time__gte=now` so any program that has not yet finished is included.
 - TS proxy connection slot leaks and TOCTOU races in stream initialization (Fixes #947) - Thanks [@CodeBormen](https://github.com/CodeBormen)
   - **TOCTOU race in slot reservation**: `get_stream()` previously used a `GET`→check→`INCR` sequence, allowing concurrent requests to both read the same count below the limit and both reserve a slot, silently exceeding `max_streams`. Replaced with an atomic `INCR`-first pattern: increment unconditionally, check the result, roll back with `DECR` if over capacity.
   - **Leak on URL generation failure**: `generate_stream_url()` called `get_stream()` (which `INCR`s the counter) but had no cleanup path if subsequent DB lookups or URL construction failed. The post-`get_stream()` block is now wrapped in a `try/except` that calls `release_stream()` on any error.
diff --git a/apps/output/views.py b/apps/output/views.py
index e99c424b..7cd5a24f 100644
--- a/apps/output/views.py
+++ b/apps/output/views.py
@@ -1643,7 +1643,7 @@ def generate_epg(request, profile_name=None, user=None):
                 # For real EPG data - filter only if days parameter was specified
                 if num_days > 0:
                     programs_qs = channel.epg_data.programs.filter(
-                        start_time__gte=now,
+                        end_time__gte=now,
                         start_time__lt=cutoff_date
                     ).order_by('id')  # Explicit ordering for consistent chunking
                 else:

From c02e060c28f501236045a9a7ca709be72e8e15d0 Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Thu, 5 Mar 2026 12:31:36 -0600
Subject: [PATCH 50/58] Enhancement: EPG output when no `days` parameter is
 specified now excludes already-ended programs instead of returning all
 historical data.

---
 CHANGELOG.md         | 4 ++++
 apps/output/views.py | 6 ++++--
 2 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 7691dafa..d542d965 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -12,6 +12,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - New Client Buffer proxy setting: new clients joining an active channel are now positioned a configurable number of seconds behind live rather than a fixed chunk count. The start position is determined by wall-clock chunk receive time (stored as a Redis sorted set alongside the buffer), so the buffer depth is consistent in seconds regardless of stream bitrate. Setting the value to `0` starts clients at live with no buffer. Defaults to 2 seconds. Existing chunk-count gating for the first client connecting to a channel is unchanged. The setting is exposed in Settings → Proxy as "New Client Buffer (seconds)".
 - Channel table filter for channels that have stale streams: A new "Has Stale Streams" filter option in the channel table header menu highlights and filters channels containing at least one stale stream. Channels with stale streams are visually distinguished with an orange tint. The filter is mutually exclusive with "Only Empty Channels". - Thanks [@JCBird1012](https://github.com/JCBird1012)
 
+### Changed
+
+- EPG output when no `days` parameter is specified now excludes already-ended programs instead of returning all historical data.
+
 ### Fixed
 
 - EPG output was filtering programs using `start_time__gte=now` when the `days` parameter was specified, which caused currently-airing programs (started before the request time but not yet ended) to be omitted from the XML output. This produced a gap in clients' guides immediately after an EPG refresh, lasting until the next program started. Fixed by changing the filter to `end_time__gte=now` so any program that has not yet finished is included.
diff --git a/apps/output/views.py b/apps/output/views.py
index 7cd5a24f..84c5f29e 100644
--- a/apps/output/views.py
+++ b/apps/output/views.py
@@ -1647,8 +1647,10 @@ def generate_epg(request, profile_name=None, user=None):
                         start_time__lt=cutoff_date
                     ).order_by('id')  # Explicit ordering for consistent chunking
                 else:
-                    # Return all programs if days=0 or not specified
-                    programs_qs = channel.epg_data.programs.all().order_by('id')
+                    # Return all non-expired programs if days=0 or not specified
+                    programs_qs = channel.epg_data.programs.filter(
+                        end_time__gte=now
+                    ).order_by('id')
 
                 # Process programs in chunks to avoid cursor timeout issues
                 program_batch = []

From ce9f7ccbf93fbc745ef8ac95b38d9f6e846ee405 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Thu, 5 Mar 2026 13:47:20 -0600
Subject: [PATCH 51/58] fix(m3u): fix Celery worker memory leak during M3U/XC
 refresh

Restructure cleanup in refresh_single_m3u_account so del statements
run before gc.collect(), enabling cyclic garbage collection. Add worker_max_memory_per_child as a safety net against memory fragmentation.

Closes #1012, Closes #1053
---
 apps/m3u/tasks.py                     |  84 +++++++++++---------
 apps/m3u/tests/__init__.py            |   0
 apps/m3u/tests/test_memory_cleanup.py | 106 ++++++++++++++++++++++++++
 dispatcharr/settings.py               |   4 +
 4 files changed, 156 insertions(+), 38 deletions(-)
 create mode 100644 apps/m3u/tests/__init__.py
 create mode 100644 apps/m3u/tests/test_memory_cleanup.py

diff --git a/apps/m3u/tasks.py b/apps/m3u/tasks.py
index a9413143..c0c90c57 100644
--- a/apps/m3u/tasks.py
+++ b/apps/m3u/tasks.py
@@ -1170,14 +1170,12 @@ def process_m3u_batch_direct(account_id, batch, groups, hash_keys):
 
     retval = f"M3U account: {account_id}, Batch processed: {len(streams_to_create)} created, {len(streams_to_update)} updated."
 
-    # Aggressive garbage collection
-    # del streams_to_create, streams_to_update, stream_hashes, existing_streams
-    # from core.utils import cleanup_memory
-    # cleanup_memory(log_usage=True, force_collection=True)
-
     # Clean up database connections for threading
     connections.close_all()
 
+    # Free batch data structures (reference-counted deallocation)
+    del streams_to_create, streams_to_update, stream_hashes, existing_streams
+
     return retval
 
 
@@ -2642,6 +2640,16 @@ def refresh_single_m3u_account(account_id):
     lock_renewer = TaskLockRenewer("refresh_single_m3u_account", account_id)
     lock_renewer.start()
 
+    try:
+        return _refresh_single_m3u_account_impl(account_id)
+    finally:
+        # Guaranteed cleanup on all exit paths (success, exception, early return)
+        lock_renewer.stop()
+        release_task_lock("refresh_single_m3u_account", account_id)
+
+
+def _refresh_single_m3u_account_impl(account_id):
+    """Implementation of M3U account refresh with guaranteed memory cleanup."""
     # Record start time
     refresh_start_timestamp = timezone.now()  # For the cleanup function
     start_time = time.time()  # For tracking elapsed time as float
@@ -2653,8 +2661,6 @@ def refresh_single_m3u_account(account_id):
         account = M3UAccount.objects.get(id=account_id, is_active=True)
         if not account.is_active:
             logger.debug(f"Account {account_id} is not active, skipping.")
-            lock_renewer.stop()
-            release_task_lock("refresh_single_m3u_account", account_id)
             return
 
         # Set status to fetching
@@ -2683,8 +2689,6 @@ def refresh_single_m3u_account(account_id):
         else:
             logger.debug(f"No orphaned task found for M3U account {account_id}")
 
-        lock_renewer.stop()
-        release_task_lock("refresh_single_m3u_account", account_id)
         return f"M3UAccount with ID={account_id} not found or inactive, task cleaned up"
 
     # Fetch M3U lines and handle potential issues
@@ -2699,6 +2703,7 @@ def refresh_single_m3u_account(account_id):
 
             extinf_data = data["extinf_data"]
             groups = data["groups"]
+            del data  # Free top-level dict; extinf_data/groups retain their references
         except json.JSONDecodeError as e:
             # Handle corrupted JSON file
             logger.error(
@@ -2734,8 +2739,6 @@ def refresh_single_m3u_account(account_id):
                 logger.error(
                     f"Failed to refresh M3U groups for account {account_id}: {result}"
                 )
-                lock_renewer.stop()
-                release_task_lock("refresh_single_m3u_account", account_id)
                 return "Failed to update m3u account - download failed or other error"
 
             extinf_data, groups = result
@@ -2768,8 +2771,6 @@ def refresh_single_m3u_account(account_id):
                 status="error",
                 error=f"Error refreshing M3U groups: {str(e)}",
             )
-            lock_renewer.stop()
-            release_task_lock("refresh_single_m3u_account", account_id)
             return "Failed to update m3u account"
 
     # Only proceed with parsing if we actually have data and no errors were encountered
@@ -2792,8 +2793,6 @@ def refresh_single_m3u_account(account_id):
             status="error",
             error="No data available for processing",
         )
-        lock_renewer.stop()
-        release_task_lock("refresh_single_m3u_account", account_id)
         return "Failed to update m3u account, no data available"
 
     hash_keys = CoreSettings.get_m3u_hash_key().split(",")
@@ -2939,6 +2938,9 @@ def refresh_single_m3u_account(account_id):
 
                 logger.info(f"Processing {len(all_xc_streams)} XC streams in {len(batches)} batches")
 
+                # Free the original list; batches hold independent sliced copies
+                del all_xc_streams
+
                 # Use threading for XC stream processing - now with consistent batch sizes
                 max_workers = min(4, len(batches))
                 logger.debug(f"Using {max_workers} threads for XC stream processing")
@@ -3099,32 +3101,38 @@ def refresh_single_m3u_account(account_id):
 
     except Exception as e:
         logger.error(f"Error processing M3U for account {account_id}: {str(e)}")
-        account.status = M3UAccount.Status.ERROR
-        account.last_message = f"Error processing M3U: {str(e)}"
-        account.save(update_fields=["status", "last_message"])
+        try:
+            account.status = M3UAccount.Status.ERROR
+            account.last_message = f"Error processing M3U: {str(e)}"
+            account.save(update_fields=["status", "last_message"])
+        except Exception:
+            logger.debug(f"Failed to update account {account_id} status during error handling")
         raise  # Re-raise the exception for Celery to handle
     finally:
-        lock_renewer.stop()
-        release_task_lock("refresh_single_m3u_account", account_id)
+        # Free large data structures regardless of success or failure
+        if 'existing_groups' in locals():
+            del existing_groups
+        if 'extinf_data' in locals():
+            del extinf_data
+        if 'groups' in locals():
+            del groups
+        if 'batches' in locals():
+            del batches
+        if 'all_xc_streams' in locals():
+            del all_xc_streams
+        if 'data' in locals():
+            del data
+        if 'filtered_groups' in locals():
+            del filtered_groups
+        if 'channel_group_relationships' in locals():
+            del channel_group_relationships
 
-    # Aggressive garbage collection
-    # Only delete variables if they exist
-    if 'existing_groups' in locals():
-        del existing_groups
-    if 'extinf_data' in locals():
-        del extinf_data
-    if 'groups' in locals():
-        del groups
-    if 'batches' in locals():
-        del batches
-
-    from core.utils import cleanup_memory
-
-    cleanup_memory(log_usage=True, force_collection=True)
-
-    # Clean up cache file since we've fully processed it
-    if os.path.exists(cache_path):
-        os.remove(cache_path)
+        # Remove cache file after processing (success or failure)
+        cache_path = os.path.join(m3u_dir, f"{account_id}.json")
+        try:
+            os.remove(cache_path)
+        except OSError:
+            pass
 
     return f"Dispatched jobs complete."
 
diff --git a/apps/m3u/tests/__init__.py b/apps/m3u/tests/__init__.py
new file mode 100644
index 00000000..e69de29b
diff --git a/apps/m3u/tests/test_memory_cleanup.py b/apps/m3u/tests/test_memory_cleanup.py
new file mode 100644
index 00000000..bd556e4c
--- /dev/null
+++ b/apps/m3u/tests/test_memory_cleanup.py
@@ -0,0 +1,106 @@
+"""
+Tests for memory cleanup behavior in M3U refresh tasks.
+
+Verifies that database connections are properly closed, task locks are
+released on all exit paths, and garbage collection runs where expected.
+
+"""
+from unittest.mock import patch, MagicMock
+
+from django.test import SimpleTestCase
+
+from apps.m3u.models import M3UAccount
+
+
+class ProcessM3UBatchCleanupTests(SimpleTestCase):
+    """Verify process_m3u_batch_direct cleans up after processing."""
+
+    @patch("apps.m3u.tasks.Stream")
+    @patch("apps.m3u.tasks.M3UAccount")
+    def test_connections_closed_after_batch(self, mock_account_cls, mock_stream_cls):
+        """Database connections must be closed after batch processing (thread safety)."""
+        from apps.m3u.tasks import process_m3u_batch_direct
+
+        mock_account = MagicMock()
+        mock_account.filters.order_by.return_value = []
+        mock_account_cls.objects.get.return_value = mock_account
+        mock_stream_cls.objects.filter.return_value.select_related.return_value.only.return_value = (
+            []
+        )
+        mock_stream_cls.generate_hash_key = MagicMock(return_value="hash123")
+
+        with patch("django.db.connections") as mock_connections:
+            process_m3u_batch_direct(1, [], {}, ["name", "url"])
+            mock_connections.close_all.assert_called()
+
+
+class LockReleaseTests(SimpleTestCase):
+    """Verify task lock is released on all exit paths."""
+
+    @patch("apps.m3u.tasks.delete_m3u_refresh_task_by_id", return_value=False)
+    def test_lock_released_on_account_not_found(self, mock_delete):
+        """release_task_lock must be called when account does not exist."""
+        with patch(
+            "apps.m3u.tasks.acquire_task_lock", return_value=True
+        ), patch("apps.m3u.tasks.release_task_lock") as mock_release, patch(
+            "apps.m3u.tasks.TaskLockRenewer"
+        ):
+            with patch(
+                "apps.m3u.tasks.M3UAccount.objects.get",
+                side_effect=M3UAccount.DoesNotExist,
+            ):
+                from apps.m3u.tasks import refresh_single_m3u_account
+
+                refresh_single_m3u_account(99999)
+
+            mock_release.assert_called_once_with(
+                "refresh_single_m3u_account", 99999
+            )
+
+    def test_lock_released_on_exception(self):
+        """release_task_lock must be called when an exception is raised."""
+        mock_account = MagicMock()
+        mock_account.is_active = True
+        mock_account.account_type = "STD"
+        mock_account.custom_properties = {}
+        mock_account.filters.all.return_value = []
+        mock_account.status = MagicMock()
+
+        with patch(
+            "apps.m3u.tasks.acquire_task_lock", return_value=True
+        ), patch("apps.m3u.tasks.release_task_lock") as mock_release, patch(
+            "apps.m3u.tasks.TaskLockRenewer"
+        ):
+            with patch(
+                "apps.m3u.tasks.M3UAccount.objects.get", return_value=mock_account
+            ):
+                with patch("os.path.exists", return_value=False):
+                    with patch(
+                        "apps.m3u.tasks.refresh_m3u_groups",
+                        side_effect=RuntimeError("test"),
+                    ):
+                        from apps.m3u.tasks import refresh_single_m3u_account
+
+                        try:
+                            refresh_single_m3u_account(1)
+                        except RuntimeError:
+                            pass
+
+            mock_release.assert_called_once_with("refresh_single_m3u_account", 1)
+
+
+class XCCategoryCleanupTests(SimpleTestCase):
+    """Regression guard: process_xc_category_direct must continue to clean up."""
+
+    @patch("apps.m3u.tasks.XCClient")
+    @patch("apps.m3u.tasks.M3UAccount")
+    def test_xc_category_calls_gc_collect(self, mock_account_cls, mock_xc_client):
+        """gc.collect() must be called after XC category processing."""
+        from apps.m3u.tasks import process_xc_category_direct
+
+        mock_account = MagicMock()
+        mock_account_cls.objects.get.return_value = mock_account
+
+        with patch("gc.collect") as mock_gc, patch("django.db.connections"):
+            process_xc_category_direct(1, {}, {}, ["name", "url"])
+            mock_gc.assert_called()
diff --git a/dispatcharr/settings.py b/dispatcharr/settings.py
index c8350895..50bac5d5 100644
--- a/dispatcharr/settings.py
+++ b/dispatcharr/settings.py
@@ -240,6 +240,10 @@ CELERY_BROKER_TRANSPORT_OPTIONS = {
 CELERY_ACCEPT_CONTENT = ["json"]
 CELERY_TASK_SERIALIZER = "json"
 
+# Worker memory safety net: recycle prefork workers exceeding 512MB RSS.
+# Prevents unbounded growth from memory fragmentation or unexpected leaks.
+CELERY_WORKER_MAX_MEMORY_PER_CHILD = 524_288  # 512 MB in KB
+
 CELERY_BEAT_SCHEDULER = "django_celery_beat.schedulers.DatabaseScheduler"
 CELERY_BEAT_SCHEDULE = {
     # Explicitly disable the old fetch-channel-statuses task

From 109e52cf0fd4be4fdd1686dfe826445921329101 Mon Sep 17 00:00:00 2001
From: None <code@bormenusa.com>
Date: Thu, 5 Mar 2026 13:58:22 -0600
Subject: [PATCH 52/58] Update changelog.md

---
 CHANGELOG.md | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index c6fc1f9d..154e4820 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -19,6 +19,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   - **Multi-source poster resolution**: Added pipeline querying EPG, VOD, TMDB, OMDb, TVMaze, and iTunes for richer recording artwork.
   - **Series rules for currently-airing episodes**: Series rules now capture currently-airing episodes in addition to future scheduled ones. (Closes #473)
   - **Search and filter controls**: Added search and filter controls to the recordings list.
+  - **Stream generator throttling**: Cached the `ProxyServer` singleton reference per client and throttled Redis resource checks (1 s) and non-owner health checks (2 s), eliminating 3+ Redis round-trips per stream loop iteration.
+  - **Automatic crash recovery on worker restart**: A `worker_ready` Celery signal now fires `recover_recordings_on_startup` automatically when the worker starts, so recordings stuck in "recording" status are recovered without manual intervention.
 
 ### Fixed
 
@@ -40,8 +42,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - `get_instance()` deadlock: if `ProxyServer()` raised an exception during singleton construction, `_instance` was left permanently as the `_INITIALIZING` sentinel, causing all subsequent `get_instance()` callers to spin in an infinite `gevent.sleep()` loop. Construction is now wrapped in `try/except`; on failure `_instance` resets to `None` so the next call can retry.
 - Non-atomic ownership acquisition in `try_acquire_ownership()`: replaced the separate `setnx()` + `expire()` calls with a single atomic `SET NX EX`, eliminating the race window where a process crash between the two calls could leave an ownership key with no TTL (permanent ownership lock).
 - DVR bug fixes — Thanks [@CodeBormen](https://github.com/CodeBormen)
-  - **Duplicate recording execution**: Shared `ClockedSchedule` objects were firing `run_recording` multiple times for the same recording, producing corrupted output files. `run_recording` now exits immediately if the recording is already in progress, completed, or stopped (failing closed on DB errors to prevent duplicate tasks from overwriting valid recordings). `revoke_task()` also cleans up both the `PeriodicTask` and its `ClockedSchedule` on execution to eliminate orphaned schedule objects. (Fixes #641)
-  - **Long-range scheduling**: Recordings scheduled more than one hour out were handed off to multiple workers simultaneously because Redis' 3600 s timer limit triggered the job more than once, corrupting the output file. Migrated to `ClockedSchedule` for database-backed persistence that survives restarts and upgrades and avoids the multi-worker race entirely. (Fixes #940)
+  - **Duplicate recording execution**: `run_recording.apply_async(countdown=...)` exceeded Redis' default `visibility_timeout` (3600 s) for recordings scheduled more than one hour out, causing Redis to redeliver the task to multiple workers simultaneously and producing corrupted output files. Replaced `apply_async` with `ClockedSchedule` + `PeriodicTask` for database-backed one-shot scheduling that survives restarts and upgrades without the redelivery race. `run_recording` also now exits immediately if the recording is already in progress, completed, or stopped. `revoke_task()` cleans up both the `PeriodicTask` and its orphaned `ClockedSchedule` on execution. (Fixes #940, #641)
   - **Stream reconnection resilience**: Recordings now survive transient network drops with automatic reconnection retrying up to 5 times and appending to the existing file. DB operations use exponential-backoff retry for transient database errors throughout the recording lifecycle.
   - **Crash recovery pipeline**: On worker restart, recordings stuck in "recording" status have their segments concatenated and remuxed. Remux sanity checks reject MKV output that is less than 50% the size of a previous MKV (duplicate-task overwrite) or less than 10% of the source TS (corrupt first attempt); the source `.ts` is preserved for manual recovery on all failure paths. (Fixes #619, #624)
   - **Output file collision**: Fixed collision when multiple tasks targeted the same filename.
@@ -59,6 +60,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   - **WebSocket-driven refresh**: Replaced all manual `fetchRecordings()` polling calls with debounced WebSocket-driven refresh so the recordings list stays up to date without redundant API requests.
   - **comskip exit code handling**: comskip treated exit code 1 ("no commercials found") as a fatal error, causing post-processing to fail on clean recordings. Exit code 1 is now recognised as a successful no-op.
   - **Differentiated WebSocket notification events**: `recording_stopped`, `recording_cancelled` (in-progress cancel), and `recording_deleted` with a `was_in_progress` flag now allow the frontend to display distinct "Recording stopped", "Recording cancelled", and "Recording deleted" toasts.
+  - **Duplicate series rule evaluation race**: Creating a series rule fired `evaluate_series_rules.delay()` in the API view while the frontend immediately called the synchronous evaluate endpoint, racing to create duplicate recordings for the same program. Removed the redundant async call from the API; the frontend's explicit evaluate call is now the sole evaluation path.
+  - **Recording card S/E badge overlap**: Season/episode badges were overlapping and metadata was hidden on the recording card.
+  - **Orphaned recording fallback in series modal**: When a series rule no longer exists, the recurring rule modal now shows a "Delete Recording" button for the orphaned recording instead of failing silently.
 
 ## [0.20.2] - 2026-03-03
 

From 9bffd6347d46e9ae2d31e4eaefd281a96aed8b0a Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Thu, 5 Mar 2026 14:42:33 -0600
Subject: [PATCH 53/58] Enhancement: Floating video player now displays the
 channel, stream, or VOD title in the player header bar. Title is passed
 through from all preview entry points: channel table, stream table, stream
 connection card, guide, DVR, recording cards, recording details modal, VOD
 modal, and series modal.

---
 CHANGELOG.md                                  |  1 +
 frontend/src/components/FloatingVideo.jsx     | 29 ++++++-
 .../src/components/cards/RecordingCard.jsx    | 82 ++++++++++++++-----
 .../components/cards/StreamConnectionCard.jsx |  2 +-
 .../forms/RecordingDetailsModal.jsx           | 44 ++++++++--
 .../components/tables/ChannelTableStreams.jsx |  9 +-
 .../src/components/tables/ChannelsTable.jsx   |  2 +-
 .../src/components/tables/StreamsTable.jsx    |  6 +-
 frontend/src/pages/DVR.jsx                    | 32 ++++++--
 frontend/src/pages/Guide.jsx                  | 12 +--
 10 files changed, 169 insertions(+), 50 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index bf803bb2..15420a07 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -9,6 +9,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+- Floating video player now displays the channel, stream, or VOD title in the player header bar. Title is passed through from all preview entry points: channel table, stream table, stream connection card, guide, DVR, recording cards, recording details modal, VOD modal, and series modal.
 - New Client Buffer proxy setting: new clients joining an active channel are now positioned a configurable number of seconds behind live rather than a fixed chunk count. The start position is determined by wall-clock chunk receive time (stored as a Redis sorted set alongside the buffer), so the buffer depth is consistent in seconds regardless of stream bitrate. Setting the value to `0` starts clients at live with no buffer. Defaults to 2 seconds. Existing chunk-count gating for the first client connecting to a channel is unchanged. The setting is exposed in Settings → Proxy as "New Client Buffer (seconds)".
 - Channel table filter for channels that have stale streams: A new "Has Stale Streams" filter option in the channel table header menu highlights and filters channels containing at least one stale stream. Channels with stale streams are visually distinguished with an orange tint. The filter is mutually exclusive with "Only Empty Channels". - Thanks [@JCBird1012](https://github.com/JCBird1012)
 - DVR enhancements — Thanks [@CodeBormen](https://github.com/CodeBormen)
diff --git a/frontend/src/components/FloatingVideo.jsx b/frontend/src/components/FloatingVideo.jsx
index 557767ed..529a605c 100644
--- a/frontend/src/components/FloatingVideo.jsx
+++ b/frontend/src/components/FloatingVideo.jsx
@@ -718,13 +718,35 @@ export default function FloatingVideo() {
           boxShadow: '0 2px 10px rgba(0,0,0,0.7)',
         }}
       >
-        {/* Simple header row with a close button */}
+        {/* Header row with optional title and close button */}
         <Flex
-          justify="flex-end"
+          justify="space-between"
+          align="center"
           style={{
-            padding: 3,
+            padding: '3px 3px 3px 8px',
+            minHeight: '38px',
           }}
         >
+          {metadata?.name ? (
+            <Text
+              size="xs"
+              fw={600}
+              style={{
+                color: 'rgba(255,255,255,0.9)',
+                flex: 1,
+                overflow: 'hidden',
+                whiteSpace: 'nowrap',
+                textOverflow: 'ellipsis',
+                paddingRight: 4,
+                textShadow: '0px 1px 3px rgba(0,0,0,0.8)',
+                userSelect: 'none',
+              }}
+            >
+              {metadata.name}
+            </Text>
+          ) : (
+            <Box style={{ flex: 1 }} />
+          )}
           <CloseButton
             onClick={handleClose}
             onTouchEnd={handleClose}
@@ -735,6 +757,7 @@ export default function FloatingVideo() {
               minWidth: '32px',
               cursor: 'pointer',
               touchAction: 'manipulation',
+              flexShrink: 0,
             }}
           />
         </Flex>
diff --git a/frontend/src/components/cards/RecordingCard.jsx b/frontend/src/components/cards/RecordingCard.jsx
index 903bb7dd..05620053 100644
--- a/frontend/src/components/cards/RecordingCard.jsx
+++ b/frontend/src/components/cards/RecordingCard.jsx
@@ -99,7 +99,11 @@ const RecordingCard = ({
   const status = customProps.status;
   const isTimeActive = now.isAfter(start) && now.isBefore(end);
   const isInterrupted = status === 'interrupted';
-  const isInProgress = isTimeActive && !isInterrupted && status !== 'completed' && status !== 'stopped';
+  const isInProgress =
+    isTimeActive &&
+    !isInterrupted &&
+    status !== 'completed' &&
+    status !== 'stopped';
   const isUpcoming = now.isBefore(start);
   const isSeriesGroup = Boolean(
     recording._group_count && recording._group_count > 1
@@ -114,7 +118,9 @@ const RecordingCard = ({
 
   const handleWatchLive = () => {
     if (!channel) return;
-    showVideo(getShowVideoUrl(channel, env_mode), 'live');
+    showVideo(getShowVideoUrl(channel, env_mode), 'live', {
+      name: channel.name,
+    });
   };
 
   const handleWatchRecording = () => {
@@ -197,7 +203,9 @@ const RecordingCard = ({
     } finally {
       setBusy(false);
       setStopConfirmOpen(false);
-      try { await fetchRecordings(); } catch {}
+      try {
+        await fetchRecordings();
+      } catch {}
     }
   };
 
@@ -371,9 +379,15 @@ const RecordingCard = ({
                   </Menu.Target>
                   <Menu.Dropdown onClick={(e) => e.stopPropagation()}>
                     <Menu.Label>Extend recording by</Menu.Label>
-                    <Menu.Item onClick={(e) => handleExtend(15, e)}>+15 minutes</Menu.Item>
-                    <Menu.Item onClick={(e) => handleExtend(30, e)}>+30 minutes</Menu.Item>
-                    <Menu.Item onClick={(e) => handleExtend(60, e)}>+1 hour</Menu.Item>
+                    <Menu.Item onClick={(e) => handleExtend(15, e)}>
+                      +15 minutes
+                    </Menu.Item>
+                    <Menu.Item onClick={(e) => handleExtend(30, e)}>
+                      +30 minutes
+                    </Menu.Item>
+                    <Menu.Item onClick={(e) => handleExtend(60, e)}>
+                      +1 hour
+                    </Menu.Item>
                   </Menu.Dropdown>
                 </Menu>
               </Box>
@@ -391,7 +405,15 @@ const RecordingCard = ({
               </ActionIcon>
             </Tooltip>
           )}
-          <Tooltip label={isInProgress ? 'Cancel & delete' : isUpcoming ? 'Cancel' : 'Delete'}>
+          <Tooltip
+            label={
+              isInProgress
+                ? 'Cancel & delete'
+                : isUpcoming
+                  ? 'Cancel'
+                  : 'Delete'
+            }
+          >
             <ActionIcon
               variant="transparent"
               color="red.9"
@@ -467,12 +489,19 @@ const RecordingCard = ({
             </Text>
           )}
 
-          <Group justify="flex-end" gap="xs" pt={4} style={{ marginTop: 'auto' }}>
+          <Group
+            justify="flex-end"
+            gap="xs"
+            pt={4}
+            style={{ marginTop: 'auto' }}
+          >
             {isInProgress && <WatchLive />}
 
             {!isUpcoming && <WatchRecording />}
             {!isUpcoming &&
-              (customProps?.status === 'completed' || customProps?.status === 'stopped' || customProps?.status === 'interrupted') &&
+              (customProps?.status === 'completed' ||
+                customProps?.status === 'stopped' ||
+                customProps?.status === 'interrupted') &&
               (!customProps?.comskip ||
                 customProps?.comskip?.status !== 'completed') && (
                 <Button
@@ -517,7 +546,11 @@ const RecordingCard = ({
             will be saved and available for playback.
           </Text>
           <Group justify="flex-end">
-            <Button variant="default" onClick={() => setStopConfirmOpen(false)} disabled={busy}>
+            <Button
+              variant="default"
+              onClick={() => setStopConfirmOpen(false)}
+              disabled={busy}
+            >
               Go Back
             </Button>
             <Button color="yellow" loading={busy} onClick={confirmStop}>
@@ -530,7 +563,9 @@ const RecordingCard = ({
       <Modal
         opened={deleteConfirmOpen}
         onClose={() => setDeleteConfirmOpen(false)}
-        title={isInProgress || isUpcoming ? 'Cancel Recording' : 'Delete Recording'}
+        title={
+          isInProgress || isUpcoming ? 'Cancel Recording' : 'Delete Recording'
+        }
         centered
         size="md"
         zIndex={9999}
@@ -544,11 +579,19 @@ const RecordingCard = ({
                 : 'This recording and all associated files will be permanently deleted.'}
           </Text>
           <Group justify="flex-end">
-            <Button variant="default" onClick={() => setDeleteConfirmOpen(false)} disabled={busy}>
+            <Button
+              variant="default"
+              onClick={() => setDeleteConfirmOpen(false)}
+              disabled={busy}
+            >
               Go Back
             </Button>
             <Button color="red" loading={busy} onClick={confirmDelete}>
-              {isInProgress ? 'Cancel & Delete' : isUpcoming ? 'Cancel' : 'Delete'}
+              {isInProgress
+                ? 'Cancel & Delete'
+                : isUpcoming
+                  ? 'Cancel'
+                  : 'Delete'}
             </Button>
           </Group>
         </Stack>
@@ -556,12 +599,13 @@ const RecordingCard = ({
     </>
   );
 
-  if (!isSeriesGroup) return (
-    <>
-      {ConfirmModals}
-      {MainCard}
-    </>
-  );
+  if (!isSeriesGroup)
+    return (
+      <>
+        {ConfirmModals}
+        {MainCard}
+      </>
+    );
 
   // Stacked look for series groups: render two shadow layers behind the main card
   return (
diff --git a/frontend/src/components/cards/StreamConnectionCard.jsx b/frontend/src/components/cards/StreamConnectionCard.jsx
index bf72b109..4952f6fe 100644
--- a/frontend/src/components/cards/StreamConnectionCard.jsx
+++ b/frontend/src/components/cards/StreamConnectionCard.jsx
@@ -431,7 +431,7 @@ const StreamConnectionCard = ({
       url = `${window.location.protocol}//${window.location.hostname}:5656${uri}`;
     }
 
-    showVideo(url);
+    showVideo(url, 'live', { name: actualChannel.name });
   };
 
   if (location.pathname !== '/stats') {
diff --git a/frontend/src/components/forms/RecordingDetailsModal.jsx b/frontend/src/components/forms/RecordingDetailsModal.jsx
index 4b2610c3..39643d27 100644
--- a/frontend/src/components/forms/RecordingDetailsModal.jsx
+++ b/frontend/src/components/forms/RecordingDetailsModal.jsx
@@ -82,7 +82,12 @@ const RecordingDetailsModal = ({
 
   // Derive poster URL from live store data instead of the stale prop snapshot.
   const livePosterUrl = React.useMemo(
-    () => getPosterUrl(customProps.poster_logo_id, customProps, getChannelLogoUrl(channel)),
+    () =>
+      getPosterUrl(
+        customProps.poster_logo_id,
+        customProps,
+        getChannelLogoUrl(channel)
+      ),
     [customProps.poster_logo_id, customProps, channel]
   );
 
@@ -91,7 +96,8 @@ const RecordingDetailsModal = ({
   const [savedTitle, setSavedTitle] = React.useState(null);
   const [savedDescription, setSavedDescription] = React.useState(null);
   const recordingName = savedTitle ?? (program.title || 'Custom Recording');
-  const description = savedDescription ?? (program.description || customProps.description || '');
+  const description =
+    savedDescription ?? (program.description || customProps.description || '');
 
   const [editTitle, setEditTitle] = React.useState('');
   const [editDescription, setEditDescription] = React.useState('');
@@ -186,7 +192,9 @@ const RecordingDetailsModal = ({
         channelsById[rec.channel] ||
         (rec.channel === recording?.channel ? channel : null);
       if (!ch) return;
-      useVideoStore.getState().showVideo(getShowVideoUrl(ch, env_mode), 'live');
+      useVideoStore
+        .getState()
+        .showVideo(getShowVideoUrl(ch, env_mode), 'live', { name: ch.name });
     }
   };
 
@@ -281,7 +289,9 @@ const RecordingDetailsModal = ({
     const se = getSeasonLabel(season, episode, onscreen);
     const posterLogoId = cp.poster_logo_id;
     const purl = getPosterUrl(posterLogoId, cp, livePosterUrl);
-    const epChannel = channelsById[rec.channel] || (rec.channel === recording?.channel ? channel : null);
+    const epChannel =
+      channelsById[rec.channel] ||
+      (rec.channel === recording?.channel ? channel : null);
 
     const onRemove = async (e) => {
       e?.stopPropagation?.();
@@ -533,12 +543,25 @@ const RecordingDetailsModal = ({
               size="sm"
               style={{ flex: 1 }}
               autoFocus
-              onKeyDown={(e) => { if (e.key === 'Enter') saveMetadata(); if (e.key === 'Escape') cancelEditing(); }}
+              onKeyDown={(e) => {
+                if (e.key === 'Enter') saveMetadata();
+                if (e.key === 'Escape') cancelEditing();
+              }}
             />
-            <ActionIcon size="sm" variant="subtle" color="green" onClick={saveMetadata}>
+            <ActionIcon
+              size="sm"
+              variant="subtle"
+              color="green"
+              onClick={saveMetadata}
+            >
               <Check size={14} />
             </ActionIcon>
-            <ActionIcon size="sm" variant="subtle" color="gray" onClick={cancelEditing}>
+            <ActionIcon
+              size="sm"
+              variant="subtle"
+              color="gray"
+              onClick={cancelEditing}
+            >
               <X size={14} />
             </ActionIcon>
           </Group>
@@ -552,7 +575,12 @@ const RecordingDetailsModal = ({
                   : `${recordingName}${program.sub_title ? ` - ${program.sub_title}` : ''}`}
             </span>
             {!isSeriesGroup && (
-              <ActionIcon size="sm" variant="subtle" color="dimmed" onClick={startEditing}>
+              <ActionIcon
+                size="sm"
+                variant="subtle"
+                color="dimmed"
+                onClick={startEditing}
+              >
                 <Pencil size={14} />
               </ActionIcon>
             )}
diff --git a/frontend/src/components/tables/ChannelTableStreams.jsx b/frontend/src/components/tables/ChannelTableStreams.jsx
index 40771673..4344fdb2 100644
--- a/frontend/src/components/tables/ChannelTableStreams.jsx
+++ b/frontend/src/components/tables/ChannelTableStreams.jsx
@@ -140,12 +140,12 @@ const ChannelStreams = ({ channel, isExpanded }) => {
   const authUser = useAuthStore((s) => s.user);
   const showVideo = useVideoStore((s) => s.showVideo);
   const env_mode = useSettingsStore((s) => s.environment.env_mode);
-  function handleWatchStream(streamHash) {
+  function handleWatchStream(streamHash, streamName) {
     let vidUrl = `/proxy/ts/stream/${streamHash}`;
     if (env_mode === 'dev') {
       vidUrl = `${window.location.protocol}//${window.location.hostname}:5656${vidUrl}`;
     }
-    showVideo(vidUrl);
+    showVideo(vidUrl, 'live', streamName ? { name: streamName } : null);
   }
 
   const [data, setData] = useState(channelStreams || []);
@@ -390,7 +390,10 @@ const ChannelStreams = ({ channel, isExpanded }) => {
                           color="blue"
                           variant="light"
                           onClick={() =>
-                            handleWatchStream(stream.stream_hash || stream.id)
+                            handleWatchStream(
+                              stream.stream_hash || stream.id,
+                              stream.name
+                            )
                           }
                           style={{ marginLeft: 2 }}
                         >
diff --git a/frontend/src/components/tables/ChannelsTable.jsx b/frontend/src/components/tables/ChannelsTable.jsx
index b6b93f43..7f92bfb4 100644
--- a/frontend/src/components/tables/ChannelsTable.jsx
+++ b/frontend/src/components/tables/ChannelsTable.jsx
@@ -693,7 +693,7 @@ const ChannelsTable = ({ onReady }) => {
     );
     const url = getChannelURL(channel);
     console.log(`Stream URL: ${url}`);
-    showVideo(url);
+    showVideo(url, 'live', { name: channel.name });
   };
 
   const onRowSelectionChange = (newSelection) => {
diff --git a/frontend/src/components/tables/StreamsTable.jsx b/frontend/src/components/tables/StreamsTable.jsx
index 65d2a418..7d4fd2c1 100644
--- a/frontend/src/components/tables/StreamsTable.jsx
+++ b/frontend/src/components/tables/StreamsTable.jsx
@@ -112,7 +112,7 @@ const StreamRowActions = ({
       'Hash:',
       row.original.stream_hash
     );
-    handleWatchStream(row.original.stream_hash);
+    handleWatchStream(row.original.stream_hash, row.original.name);
   }, [row.original, handleWatchStream]); // Add proper dependencies to ensure correct stream
 
   const iconSize =
@@ -1012,12 +1012,12 @@ const StreamsTable = ({ onReady }) => {
     });
   };
 
-  function handleWatchStream(streamHash) {
+  function handleWatchStream(streamHash, streamName) {
     let vidUrl = `/proxy/ts/stream/${streamHash}`;
     if (env_mode == 'dev') {
       vidUrl = `${window.location.protocol}//${window.location.hostname}:5656${vidUrl}`;
     }
-    showVideo(vidUrl);
+    showVideo(vidUrl, 'live', streamName ? { name: streamName } : null);
   }
 
   const onSortingChange = (column) => {
diff --git a/frontend/src/pages/DVR.jsx b/frontend/src/pages/DVR.jsx
index a8e0d6ed..31e6f1ba 100644
--- a/frontend/src/pages/DVR.jsx
+++ b/frontend/src/pages/DVR.jsx
@@ -26,7 +26,11 @@ const RecordingDetailsModal = lazy(
 );
 import RecurringRuleModal from '../components/forms/RecurringRuleModal.jsx';
 import RecordingCard from '../components/cards/RecordingCard.jsx';
-import { categorizeRecordings, filterRecordings, buildChannelOptions } from '../utils/pages/DVRUtils.js';
+import {
+  categorizeRecordings,
+  filterRecordings,
+  buildChannelOptions,
+} from '../utils/pages/DVRUtils.js';
 import {
   getChannelLogoUrl,
   getPosterUrl,
@@ -101,7 +105,12 @@ const DVRPage = () => {
     setDetailsOpen(false);
     setDetailsRecording(null);
     setEditRecording(null);
-    setRuleModal({ open: true, ruleId, recording, isDelete: isDelete || false });
+    setRuleModal({
+      open: true,
+      ruleId,
+      recording,
+      isDelete: isDelete || false,
+    });
   };
 
   const closeRuleModal = () => setRuleModal({ open: false, ruleId: null });
@@ -152,7 +161,8 @@ const DVRPage = () => {
   }, [channelsById, inProgress, upcoming, completed]);
 
   // Filtered buckets
-  const hasActiveFilters = searchQuery !== '' || selectedChannelId !== null || selectedStatus !== null;
+  const hasActiveFilters =
+    searchQuery !== '' || selectedChannelId !== null || selectedStatus !== null;
 
   const filteredInProgress = useMemo(() => {
     if (selectedStatus && selectedStatus !== 'recording') return [];
@@ -165,13 +175,21 @@ const DVRPage = () => {
   }, [upcoming, searchQuery, selectedChannelId, selectedStatus]);
 
   const filteredCompleted = useMemo(() => {
-    if (selectedStatus && !['completed', 'interrupted'].includes(selectedStatus)) return [];
+    if (
+      selectedStatus &&
+      !['completed', 'interrupted'].includes(selectedStatus)
+    )
+      return [];
     let filtered = filterRecordings(completed, searchQuery, selectedChannelId);
     if (selectedStatus === 'interrupted') {
-      filtered = filtered.filter((rec) => rec.custom_properties?.status === 'interrupted');
+      filtered = filtered.filter(
+        (rec) => rec.custom_properties?.status === 'interrupted'
+      );
     } else if (selectedStatus === 'completed') {
       // "Completed" includes both completed and stopped recordings
-      filtered = filtered.filter((rec) => rec.custom_properties?.status !== 'interrupted');
+      filtered = filtered.filter(
+        (rec) => rec.custom_properties?.status !== 'interrupted'
+      );
     }
     return filtered;
   }, [completed, searchQuery, selectedChannelId, selectedStatus]);
@@ -196,7 +214,7 @@ const DVRPage = () => {
         channel,
         useSettingsStore.getState().environment.env_mode
       );
-      useVideoStore.getState().showVideo(url, 'live');
+      useVideoStore.getState().showVideo(url, 'live', { name: channel.name });
     }
   };
 
diff --git a/frontend/src/pages/Guide.jsx b/frontend/src/pages/Guide.jsx
index 980f145e..51133323 100644
--- a/frontend/src/pages/Guide.jsx
+++ b/frontend/src/pages/Guide.jsx
@@ -724,7 +724,9 @@ export default function TVChannelGuide({ startDate, endDate }) {
         return;
       }
 
-      showVideo(getShowVideoUrl(matched, env_mode));
+      showVideo(getShowVideoUrl(matched, env_mode), 'live', {
+        name: matched.name,
+      });
     },
     [env_mode, findChannelByTvgId, showVideo]
   );
@@ -733,7 +735,9 @@ export default function TVChannelGuide({ startDate, endDate }) {
     (channel, event) => {
       event.stopPropagation();
 
-      showVideo(getShowVideoUrl(channel, env_mode));
+      showVideo(getShowVideoUrl(channel, env_mode), 'live', {
+        name: channel.name,
+      });
     },
     [env_mode, showVideo]
   );
@@ -1342,9 +1346,7 @@ export default function TVChannelGuide({ startDate, endDate }) {
           }}
           pos="relative"
         >
-          <LoadingOverlay
-            visible={isProgramsLoading || isChannelsLoading}
-          />
+          <LoadingOverlay visible={isProgramsLoading || isChannelsLoading} />
           {nowPosition >= 0 && (
             <Box
               ref={nowLineRef}

From badd5527472e0e9cac34b736c90ef209f4ac49dd Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Thu, 5 Mar 2026 14:52:10 -0600
Subject: [PATCH 54/58] Tests: Fix DVR test after video player change.

---
 frontend/src/pages/__tests__/DVR.test.jsx | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/frontend/src/pages/__tests__/DVR.test.jsx b/frontend/src/pages/__tests__/DVR.test.jsx
index 2f5e5021..ae56d219 100644
--- a/frontend/src/pages/__tests__/DVR.test.jsx
+++ b/frontend/src/pages/__tests__/DVR.test.jsx
@@ -34,7 +34,9 @@ vi.mock('../../api');
 // Mock Mantine components
 vi.mock('@mantine/core', () => ({
   ActionIcon: ({ children, onClick }) => (
-    <button data-testid="action-icon" onClick={onClick}>{children}</button>
+    <button data-testid="action-icon" onClick={onClick}>
+      {children}
+    </button>
   ),
   Box: ({ children }) => <div data-testid="box">{children}</div>,
   Container: ({ children }) => <div data-testid="container">{children}</div>,
@@ -58,7 +60,9 @@ vi.mock('@mantine/core', () => ({
     >
       <option value="">{placeholder}</option>
       {(data || []).map((opt) => (
-        <option key={opt.value} value={opt.value}>{opt.label}</option>
+        <option key={opt.value} value={opt.value}>
+          {opt.label}
+        </option>
       ))}
     </select>
   ),
@@ -542,7 +546,8 @@ describe('DVRPage', () => {
 
       expect(mockShowVideo).toHaveBeenCalledWith(
         expect.stringContaining('stream.url'),
-        'live'
+        'live',
+        { name: 'Channel 1' }
       );
     });
 

From 78aa8cf35376131ed7243247040e4ed417b48cc5 Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Thu, 5 Mar 2026 15:36:12 -0600
Subject: [PATCH 55/58] Enhancement: Change default new client behind seconds
 from 2 to 5 for stability.

---
 CHANGELOG.md                                                  | 2 +-
 apps/proxy/config.py                                          | 4 ++--
 apps/proxy/ts_proxy/config_helper.py                          | 2 +-
 core/api_views.py                                             | 2 +-
 core/models.py                                                | 2 +-
 core/serializers.py                                           | 2 +-
 frontend/src/utils/forms/settings/ProxySettingsFormUtils.js   | 2 +-
 .../forms/settings/__tests__/ProxySettingsFormUtils.test.js   | 2 +-
 8 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 15420a07..18195305 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -10,7 +10,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Added
 
 - Floating video player now displays the channel, stream, or VOD title in the player header bar. Title is passed through from all preview entry points: channel table, stream table, stream connection card, guide, DVR, recording cards, recording details modal, VOD modal, and series modal.
-- New Client Buffer proxy setting: new clients joining an active channel are now positioned a configurable number of seconds behind live rather than a fixed chunk count. The start position is determined by wall-clock chunk receive time (stored as a Redis sorted set alongside the buffer), so the buffer depth is consistent in seconds regardless of stream bitrate. Setting the value to `0` starts clients at live with no buffer. Defaults to 2 seconds. Existing chunk-count gating for the first client connecting to a channel is unchanged. The setting is exposed in Settings → Proxy as "New Client Buffer (seconds)".
+- New Client Buffer proxy setting: new clients joining an active channel are now positioned a configurable number of seconds behind live rather than a fixed chunk count. The start position is determined by wall-clock chunk receive time (stored as a Redis sorted set alongside the buffer), so the buffer depth is consistent in seconds regardless of stream bitrate. Setting the value to `0` starts clients at live with no buffer. Defaults to 5 seconds. Existing chunk-count gating for the first client connecting to a channel is unchanged. The setting is exposed in Settings → Proxy as "New Client Buffer (seconds)".
 - Channel table filter for channels that have stale streams: A new "Has Stale Streams" filter option in the channel table header menu highlights and filters channels containing at least one stale stream. Channels with stale streams are visually distinguished with an orange tint. The filter is mutually exclusive with "Only Empty Channels". - Thanks [@JCBird1012](https://github.com/JCBird1012)
 - DVR enhancements — Thanks [@CodeBormen](https://github.com/CodeBormen)
   - **Stop Recording**: A new Stop button (distinct from Cancel) cleanly ends an in-progress recording early and keeps the partial file available for playback. The API returns immediately; stream teardown and task revocation happen in a background thread to prevent 504 timeouts. When multiple recordings run simultaneously, stopping one only terminates that recording's proxy client by ID, leaving all others unaffected. (Closes #454)
diff --git a/apps/proxy/config.py b/apps/proxy/config.py
index dbdd8278..68e27a6f 100644
--- a/apps/proxy/config.py
+++ b/apps/proxy/config.py
@@ -43,7 +43,7 @@ class BaseConfig:
                 "redis_chunk_ttl": 60,
                 "channel_shutdown_delay": 0,
                 "channel_init_grace_period": 5,
-                "new_client_behind_seconds": 2,
+                "new_client_behind_seconds": 5,
             }
 
         finally:
@@ -82,7 +82,7 @@ class TSConfig(BaseConfig):
     # Buffer settings
     INITIAL_BEHIND_CHUNKS = 4  # How many chunks behind to start a client (4 chunks = ~1MB)
     CHUNK_BATCH_SIZE = 5       # How many chunks to fetch in one batch
-    NEW_CLIENT_BEHIND_SECONDS = 2  # Start new clients this many seconds behind live (0 = start at live)
+    NEW_CLIENT_BEHIND_SECONDS = 5  # Start new clients this many seconds behind live (0 = start at live)
     KEEPALIVE_INTERVAL = 0.5   # Seconds between keepalive packets when at buffer head
     # Chunk read timeout
     CHUNK_TIMEOUT = 5        # Seconds to wait for each chunk read
diff --git a/apps/proxy/ts_proxy/config_helper.py b/apps/proxy/ts_proxy/config_helper.py
index f0f7a5b7..f9a0418e 100644
--- a/apps/proxy/ts_proxy/config_helper.py
+++ b/apps/proxy/ts_proxy/config_helper.py
@@ -48,7 +48,7 @@ class ConfigHelper:
         Loaded from DB proxy_settings so users can change it at runtime."""
         from apps.proxy.config import TSConfig
         settings = TSConfig.get_proxy_settings()
-        return settings.get('new_client_behind_seconds', 2)
+        return settings.get('new_client_behind_seconds', 5)
 
     @staticmethod
     def keepalive_interval():
diff --git a/core/api_views.py b/core/api_views.py
index 43e55d02..dd15886c 100644
--- a/core/api_views.py
+++ b/core/api_views.py
@@ -183,7 +183,7 @@ class ProxySettingsViewSet(viewsets.ViewSet):
                 "redis_chunk_ttl": 60,
                 "channel_shutdown_delay": 0,
                 "channel_init_grace_period": 5,
-                "new_client_behind_seconds": 2,
+                "new_client_behind_seconds": 5,
             }
             settings_obj, created = CoreSettings.objects.get_or_create(
                 key=PROXY_SETTINGS_KEY,
diff --git a/core/models.py b/core/models.py
index 7ce7713d..4e6322e7 100644
--- a/core/models.py
+++ b/core/models.py
@@ -329,7 +329,7 @@ class CoreSettings(models.Model):
             "redis_chunk_ttl": 60,
             "channel_shutdown_delay": 0,
             "channel_init_grace_period": 5,
-            "new_client_behind_seconds": 2,
+            "new_client_behind_seconds": 5,
         })
 
     # System Settings
diff --git a/core/serializers.py b/core/serializers.py
index b0b730ee..bff5f701 100644
--- a/core/serializers.py
+++ b/core/serializers.py
@@ -78,7 +78,7 @@ class ProxySettingsSerializer(serializers.Serializer):
     redis_chunk_ttl = serializers.IntegerField(min_value=10, max_value=3600)
     channel_shutdown_delay = serializers.IntegerField(min_value=0, max_value=300)
     channel_init_grace_period = serializers.IntegerField(min_value=0, max_value=60)
-    new_client_behind_seconds = serializers.IntegerField(min_value=0, max_value=120, required=False, default=2)
+    new_client_behind_seconds = serializers.IntegerField(min_value=0, max_value=120, required=False, default=5)
 
     def validate_buffering_timeout(self, value):
         if value < 0 or value > 300:
diff --git a/frontend/src/utils/forms/settings/ProxySettingsFormUtils.js b/frontend/src/utils/forms/settings/ProxySettingsFormUtils.js
index e6e0464b..eddbba91 100644
--- a/frontend/src/utils/forms/settings/ProxySettingsFormUtils.js
+++ b/frontend/src/utils/forms/settings/ProxySettingsFormUtils.js
@@ -14,6 +14,6 @@ export const getProxySettingDefaults = () => {
     redis_chunk_ttl: 60,
     channel_shutdown_delay: 0,
     channel_init_grace_period: 5,
-    new_client_behind_seconds: 2,
+    new_client_behind_seconds: 5,
   };
 };
diff --git a/frontend/src/utils/forms/settings/__tests__/ProxySettingsFormUtils.test.js b/frontend/src/utils/forms/settings/__tests__/ProxySettingsFormUtils.test.js
index a48e5a37..21c48269 100644
--- a/frontend/src/utils/forms/settings/__tests__/ProxySettingsFormUtils.test.js
+++ b/frontend/src/utils/forms/settings/__tests__/ProxySettingsFormUtils.test.js
@@ -61,7 +61,7 @@ describe('ProxySettingsFormUtils', () => {
         redis_chunk_ttl: 60,
         channel_shutdown_delay: 0,
         channel_init_grace_period: 5,
-        new_client_behind_seconds: 2,
+        new_client_behind_seconds: 5,
       });
     });
 

From 2b44bf1f8eff61be8d2bf7c0961b8363fe11f4f5 Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Thu, 5 Mar 2026 16:55:24 -0600
Subject: [PATCH 56/58] changelog: Update changelog for M3U refresh memory
 improvements.

---
 CHANGELOG.md | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 18195305..a8bd0551 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -29,6 +29,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- Celery worker memory leak during M3U/XC refresh causing 20–80 MB growth per cycle with no reclamation (Fixes #1012, #1053) - Thanks [@CodeBormen](https://github.com/CodeBormen)
+  - Restructured `refresh_single_m3u_account()` with a `try/finally` that guarantees `del` of large data structures runs before Celery's `gc.collect()`, and lock release on all exit paths (success, exception, early return)
+  - Re-enabled batch data cleanup in `process_m3u_batch_direct()` (was commented out)
+  - Added `CELERY_WORKER_MAX_MEMORY_PER_CHILD = 512 MB` as a safety net against pymalloc arena fragmentation
 - EPG output was filtering programs using `start_time__gte=now` when the `days` parameter was specified, which caused currently-airing programs (started before the request time but not yet ended) to be omitted from the XML output. This produced a gap in clients' guides immediately after an EPG refresh, lasting until the next program started. Fixed by changing the filter to `end_time__gte=now` so any program that has not yet finished is included.
 - TS proxy connection slot leaks and TOCTOU races in stream initialization (Fixes #947) - Thanks [@CodeBormen](https://github.com/CodeBormen)
   - **TOCTOU race in slot reservation**: `get_stream()` previously used a `GET`→check→`INCR` sequence, allowing concurrent requests to both read the same count below the limit and both reserve a slot, silently exceeding `max_streams`. Replaced with an atomic `INCR`-first pattern: increment unconditionally, check the result, roll back with `DECR` if over capacity.

From a51eb37075cb80163e08c8d795fd2a5b1c343ca8 Mon Sep 17 00:00:00 2001
From: Jonathan Caicedo <jonathan@jcaicedo.com>
Date: Fri, 6 Mar 2026 08:10:30 -0500
Subject: [PATCH 57/58] fix: M3U EXTINF parsing for attributes ending in double
 equals

fixes: #1055
---
 apps/m3u/tasks.py                     |  6 ++--
 apps/m3u/tests/test_extinf_parsing.py | 41 +++++++++++++++++++++++++++
 2 files changed, 45 insertions(+), 2 deletions(-)
 create mode 100644 apps/m3u/tests/test_extinf_parsing.py

diff --git a/apps/m3u/tasks.py b/apps/m3u/tasks.py
index c0c90c57..162359e9 100644
--- a/apps/m3u/tasks.py
+++ b/apps/m3u/tasks.py
@@ -482,8 +482,10 @@ def parse_extinf_line(line: str) -> dict:
     attrs = {}
     last_attr_end = 0
 
-    # Use a single regex that handles both quote types
-    for match in re.finditer(r'([^\s]+)=(["\'])([^\2]*?)\2', content):
+    # Use a single regex that handles both quote types.
+    # Keys must stop at '=' so values like base64-padded URLs ending with '=='
+    # don't get folded into the preceding attribute name.
+    for match in re.finditer(r'([^\s=]+)\s*=\s*(["\'])(.*?)\2', content):
         key = match.group(1)
         value = match.group(3)
         attrs[key] = value
diff --git a/apps/m3u/tests/test_extinf_parsing.py b/apps/m3u/tests/test_extinf_parsing.py
new file mode 100644
index 00000000..367f569b
--- /dev/null
+++ b/apps/m3u/tests/test_extinf_parsing.py
@@ -0,0 +1,41 @@
+from django.test import SimpleTestCase
+
+from apps.m3u.tasks import parse_extinf_line
+
+
+class ParseExtinfLineTests(SimpleTestCase):
+    def test_preserves_equals_padding_in_tvg_logo(self):
+        line = (
+            '#EXTINF:-1 tvg-id="cp_891ee08a2cdfde210ec2c9137127103b" '
+            'tvg-chno="1001" '
+            'tvg-name="UK Sky Sports Premier League" '
+            'tvg-logo="https://e3.365dm.com/tvlogos/channels/1303-Logo.png?'
+            'U2t5IFNwb3J0cyBQcmVtaWVyIExlYWd1ZQ==" '
+            'group-title="Team Games",UK Sky Sports Premier League'
+        )
+
+        parsed = parse_extinf_line(line)
+
+        self.assertIsNotNone(parsed)
+        self.assertEqual(
+            parsed["attributes"]["tvg-logo"],
+            "https://e3.365dm.com/tvlogos/channels/1303-Logo.png?U2t5IFNwb3J0cyBQcmVtaWVyIExlYWd1ZQ==",
+        )
+        self.assertEqual(parsed["attributes"]["group-title"], "Team Games")
+        self.assertEqual(parsed["name"], "UK Sky Sports Premier League")
+
+    def test_supports_single_quoted_attributes(self):
+        line = (
+            "#EXTINF:-1 tvg-name='Channel One' tvg-logo='https://example.com/logo==.png' "
+            "group-title='Sports',Channel One"
+        )
+
+        parsed = parse_extinf_line(line)
+
+        self.assertIsNotNone(parsed)
+        self.assertEqual(
+            parsed["attributes"]["tvg-logo"],
+            "https://example.com/logo==.png",
+        )
+        self.assertEqual(parsed["attributes"]["group-title"], "Sports")
+        self.assertEqual(parsed["display_name"], "Channel One")

From 4209794cfafa7fee7e53a6be79ab051488c6e105 Mon Sep 17 00:00:00 2001
From: SergeantPanda <chester.cullen@gmail.com>
Date: Fri, 6 Mar 2026 09:05:30 -0600
Subject: [PATCH 58/58] changelog: Update changelog for extinf attribute
 parsing.

---
 CHANGELOG.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index a8bd0551..7826d7e9 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -29,6 +29,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- M3U EXTINF attribute parsing for values containing `=` or `==` (e.g. base64-padded `tvg-logo` URLs, catchup tokens with query strings). The previous regex used `[^\s]+` for the key pattern, allowing `=` signs inside a quoted value to be greedily absorbed into the next attribute's key name, causing that attribute and all subsequent ones on the line to be silently dropped. Changed to `[^\s=]+` so the key match always stops at the first `=`. (Fixes #1055) - Thanks [@JCBird1012](https://github.com/JCBird1012)
 - Celery worker memory leak during M3U/XC refresh causing 20–80 MB growth per cycle with no reclamation (Fixes #1012, #1053) - Thanks [@CodeBormen](https://github.com/CodeBormen)
   - Restructured `refresh_single_m3u_account()` with a `try/finally` that guarantees `del` of large data structures runs before Celery's `gc.collect()`, and lock release on all exit paths (success, exception, early return)
   - Re-enabled batch data cleanup in `process_m3u_batch_direct()` (was commented out)