Sonarr Activity tab has 12 pages but we only fetched ~2 items.
Added pageSize=1000 to queue API and changed history default from 10 to 100.
This ensures all downloads are available for matching to SAB/qBittorrent.
Sonarr tracks the exact SAB download ID (nzo_id). Now tries to match
by downloadId first, then falls back to title matching. Also adds
debug to show if matches are via downloadId vs title, and logs
downloadIds in history to verify the link exists.
When a match is found, logs whether it came from queue or history.
When no match, shows history counts and sample titles to verify
history is being checked properly.
SAB items often persist after Sonarr has processed them.
Previously only checked the active queue, now also checks
history records so completed downloads still appear.
SAB filenames use dots (dora.the.explorer.s02e08) but Sonarr titles
use spaces (Dora the Explorer - S02E08). Now tries matching with
both formats to improve match rate.
Also logs actual Sonarr titles when no match found for debugging.
Shows exactly which SAB items match/don't match to Sonarr/Radarr:
- ✓ Sonarr match: SAB name → Sonarr name
- ✓ Radarr match: SAB name → Radarr name
- ✗ No match: SAB name (with Sonarr queue count)
This will help diagnose why Sonarr Activity Queue shows matches but Sofarr doesn't.
SABnzbd API returns labels as an array in newer versions,
but the code assumed it was a comma-separated string.
Now handles both cases to prevent 'slot.labels.split is not a function' error.
1. Fixed download client collision:
- SABnzbd client with id 'i3omb' was being overwritten by qBittorrent
- Now uses unique key ':' like the arr retrievers
2. Fixed webhook metrics showing 0:
- instanceName from webhooks is generic ('Sonarr', 'Radarr')
- Not the configured instance name ('i3omb')
- Now updates metrics for ALL instances of that type
When Sonarr and Radarr had the same instance ID (e.g., 'i3omb'),
the Radarr retriever would overwrite the Sonarr retriever in the Map.
This caused webhook refreshes to show '0 instance(s)' for Sonarr.
Now uses ':' as the unique key so both can coexist.
The status panel was showing webhooks as disabled (null) when no events
had been received yet. Now it checks Sonarr/Radarr API to see if the
Sofarr webhook notification is actually configured.
- Added checkWebhookConfigured() to verify webhook exists in Sonarr/Radarr
- Shows 'enabled: true' with 0 events when webhook is configured
- Only shows null when webhook is not configured at all
Add debug logging to trace:
- When downloads payload is built
- Data sizes from cache (SAB, qBit, Sonarr, Radarr)
- Number of downloads found and their titles
This will help diagnose why Dora downloads aren't appearing.
- Fixed webhooks section to load collapsed (content hidden, toggle arrow reset)
- Added webhook metrics card to status panel for admin users:
- Shows Sonarr/Radarr enabled/disabled status
- Shows events received and polls skipped counts
- Updated /api/dashboard/status endpoint to include webhook metrics
- Metrics are aggregated from all Sonarr/Radarr instances
The webhook notification payload was using string 'POST' for the method
field, but Sonarr/Radarr API expects numeric values:
- 1 = POST
- 2 = PUT
Also added onManualInteractionRequired: false to match the schema.
Fixes: Radarr/Sonarr rejecting webhook configuration with validation errors
The notification routes were using process.env.SONARR_URL directly,
which is undefined when using the newer SONARR_INSTANCES JSON format.
Changes:
- Added getFirstSonarrInstance() and getFirstRadarrInstance() helpers
- Updated /notifications, /notifications/test, and /notifications/sofarr-webhook
routes to use instance config from getSonarrInstances()/getRadarrInstances()
- Returns 503 error if no instances are configured
Fixes: 'Invalid URL' errors when calling Sonarr/Radarr notification APIs
Added detailed error logging to help diagnose 500 errors when calling
Sonarr/Radarr notification APIs. Logs include:
- Error message
- Response status (if available)
- Response data (if available)
This will help identify if the issue is:
- Missing SONARR_URL/RADARR_URL or API keys
- Network connectivity issues
- Sonarr/Radarr API version incompatibility
Webhook routes were only registered in app.js (the test factory) but
not in index.js (the production entry point). POST /api/webhook/*
was therefore falling through to the verifyCsrf middleware and being
rejected with 403 in production.
- Added POST /api/webhook/sonarr and POST /api/webhook/radarr endpoints
- Implemented webhook secret validation via SOFARR_WEBHOOK_SECRET environment variable
- Added logging for all incoming webhook events using existing logToFile utility
- Returns HTTP 200 immediately to prevent webhook retries
- Mounted webhook routes before CSRF middleware (called by external services)
- Non-breaking: no changes to polling, caching, SSE, or any existing behavior
- Lays groundwork for Phase 2 (cache + SSE integration) without implementing it yet
- Remove instanceConfig parameter from all retriever methods (getTags, getQueue, getHistory)
- Retriever instances now use this.url, this.apiKey, this.id instead of passed parameter
- Convert ArrRetrieverRegistry from class with convenience functions to pure singleton object
- Export singleton instance directly instead of class + convenience functions
- Update poller.js and historyFetcher.js to call methods on singleton directly
- All 261 tests pass with zero behavior changes
- Create ArrRetriever abstract base class defining pluggable interface
- Implement PollingSonarrRetriever and PollingRadarrRetriever with HTTP polling
- Add ArrRetrieverRegistry for managing retriever instances
- Refactor poller.js to use retriever registry instead of direct Axios calls
- Update historyFetcher.js to use retriever registry
- Preserve all cache keys, TTLs, timing logs, SSE broadcasts, error handling
- Enable future webhook listeners without touching poller logic
- Remove auto-appending of /RPC2 from RTorrentClient constructor
- Use exact URL from config (supports custom paths like whatbox.ca/xmlrpc)
- Update .env.sample with clear URL path documentation and examples
- Update README.md with comprehensive PDCA section and all download clients
- Add URL path verification tests (whatbox.ca, custom paths, no auth)
- Update architecture diagram to include Transmission and rTorrent
- Update Docker Compose example to include all download clients
- Update prerequisites to mention all supported download clients
- Update "What It Does" and "The Matching Process" sections
- Implement RTorrentClient extending DownloadClient abstract class
- Use xmlrpc package (v1.3.2) for XML-RPC communication
- Support HTTP Basic Auth when credentials are configured
- Map rTorrent states (d.state, d.is_active, d.is_hash_checking) to normalized statuses
- Calculate ETA from download speed and remaining bytes
- Add getRtorrentInstances() to config.js
- Register RTorrentClient in downloadClients.js registry
- Add 8 comprehensive unit tests covering all functionality
- Update .env.sample with rtorrent configuration examples
- Update ARCHITECTURE.md with rtorrent client details
- Update ADDING-A-DOWNLOAD-CLIENT.md with rtorrent-specific notes
Remove undefined QBittorrentClient export that was causing
container startup failures. The actual implementation is now
in server/clients/QBittorrentClient.js
- Add abstract DownloadClient base class with standardized interface
- Refactor QBittorrentClient to extend DownloadClient with Sync API support
- Create SABnzbdClient implementing DownloadClient interface
- Add TransmissionClient as proof-of-concept implementation
- Implement DownloadClientRegistry for factory pattern and client management
- Refactor poller.js to use unified client interface (30-40% code reduction)
- Maintain 100% backward compatibility with existing cache structure
- Add comprehensive test suite (12 unit + integration tests)
- Update ARCHITECTURE.md with detailed PDCA documentation
- Create ADDING-A-DOWNLOAD-CLIENT.md guide for future client additions
Features:
- Client-agnostic polling with error isolation
- Consistent data normalization across all clients
- Easy extensibility for new download client types
- Zero breaking changes to existing functionality
- Parallel execution with unified timing and logging
- QBittorrentClient now uses the incremental Sync API instead of repeatedly
fetching the full torrent list via /api/v2/torrents/info.
- Per-client state: lastRid, torrentMap, fallbackThisCycle.
- Handles full_update, delta updates, and torrents_removed.
- Falls back to legacy torrents/info at most once per poll cycle.
- getAllTorrents() resets fallback flags before each cycle.
- Added 9 new unit tests covering: first sync, delta merge, full_update,
torrents_removed, fallback path, direct-legacy-after-fallback, 403 re-auth,
completed-field computation, and fallback reset.
- Added addedOn timestamp to qBittorrent torrent mapping
- Added canBlocklist helper function: true for admins, true for non-admins when (importIssues OR (torrent >1h old AND availability<100%))
- Added canBlocklist field to all download objects in /user-downloads and SSE /stream routes (8 blocks total)
- Frontend button now shows when (isAdmin OR download.canBlocklist) && download.arrQueueId
- Poller now stores _instanceKey alongside _instanceUrl on Sonarr/Radarr queue records
- dashboard route threads arrQueueId/arrType/arrInstanceUrl/arrInstanceKey/arrContentId/arrContentType as admin-only fields on downloads with importIssues
- POST /api/dashboard/blocklist-search: admin-only, removes queue item with blocklist=true then triggers EpisodeSearch/MoviesSearch
- Button renders in download card header (admin + importIssues + arrQueueId only)
- Confirm dialog, loading/success/error states on the button
- Kicks a background poll on success so SSE reflects removed item promptly
- /health endpoint now includes version field
- Footer displays 'sofarr vX.Y.Z' fetched on page load
- Subtle .app-version styling (smaller, dimmed)
- Bump version to 1.2.1, update CHANGELOG
Sonarr queue and history records do not expose episodeNumber at the
top level — it is only present inside the nested episode object
(record.episode.episodeNumber). Same for seasonNumber. The original
extractEpisode() read record.episodeNumber which was always undefined,
so gatherEpisodes() always returned an empty array.
Fix: prefer the nested episode object fields, falling back to the
top-level fields for forward-compatibility.
