Second Sentry sweep (2026-07-08) surfaced four new noise-class issues on 2026.7.0 — all expected operator-input or transient conditions logged/captured as errors. None are bugs; each already has a proper user-facing outcome. This drops them so real issues stay visible.
- ANTHIAS-3W —
POST /api/v2/recover with a non-gzip / non-backup file logs logger.exception('Backup recovery failed') (→ Sentry error) even though the endpoint already returns a 400 with "Invalid backup archive." Operator input, not a bug. → log at warning.
- ANTHIAS-3V — the viewer's
Scheduler.get_next_asset logs error('Asset not found or processed') when an operator jumps to an asset that was since deleted or is still processing (a benign race). → log at warning, name the asset, fall back to the playlist.
- ANTHIAS-3T — a yt-dlp
DownloadError (network timeout, geo-block, private/deleted video, bad URL) is captured by the Celery Sentry integration. The download task's on_failure already records metadata.error_message, so the operator sees a "Failed" pill. → drop yt_dlp DownloadError in before_send (matched by name+module, no yt_dlp import).
- ANTHIAS-3X —
celery.backends.asynchronous logs "Retry limit exceeded while trying to reconnect to the Celery result store backend" at a fatal level during a sustained redis outage — the same transient-redis noise as the loggers already ignored (celery.backends.redis, etc.). → ignore_logger('celery.backends.asynchronous').
Also resolved ANTHIAS-3S (No module named 'channels', from a non-device session-* context on 2026.06.2) as environmental.
Fixes are one-liners + log-level downgrades, each with a regression test.
Second Sentry sweep (2026-07-08) surfaced four new noise-class issues on
2026.7.0— all expected operator-input or transient conditions logged/captured as errors. None are bugs; each already has a proper user-facing outcome. This drops them so real issues stay visible.POST /api/v2/recoverwith a non-gzip / non-backup file logslogger.exception('Backup recovery failed')(→ Sentry error) even though the endpoint already returns a 400 with "Invalid backup archive." Operator input, not a bug. → log at warning.Scheduler.get_next_assetlogserror('Asset not found or processed')when an operator jumps to an asset that was since deleted or is still processing (a benign race). → log at warning, name the asset, fall back to the playlist.DownloadError(network timeout, geo-block, private/deleted video, bad URL) is captured by the Celery Sentry integration. The download task'son_failurealready recordsmetadata.error_message, so the operator sees a "Failed" pill. → drop yt_dlpDownloadErrorinbefore_send(matched by name+module, no yt_dlp import).celery.backends.asynchronouslogs "Retry limit exceeded while trying to reconnect to the Celery result store backend" at a fatal level during a sustained redis outage — the same transient-redis noise as the loggers already ignored (celery.backends.redis, etc.). →ignore_logger('celery.backends.asynchronous').Also resolved ANTHIAS-3S (
No module named 'channels', from a non-devicesession-*context on2026.06.2) as environmental.Fixes are one-liners + log-level downgrades, each with a regression test.