Two changes for issue #7 (HLS cleanup + orphan reaper) and the user's "SRT clips ingest but won't play" complaint: 1) New POST /assets/cleanup-live-orphans — lists every directory under /live/<uuid>/ and deletes the ones whose UUIDs don't match an asset row. These accumulate when a recorder crashes mid-capture: the live HLS dir is created but no asset is ever finalized in the DB, so the files just sit on disk forever. 2) POST /assets/:id/retry now also works for assets that are 'ready' or 'archived' but have no proxy_s3_key. The original behavior (only re-queue when status='error') made it impossible to re-generate a proxy for older recorder captures that landed without one — the user could see a thumbnail in the library but the player would just show "Preview not yet available" with no retry path. |
||
|---|---|---|
| .. | ||
| src | ||
| .env.example | ||
| .gitignore | ||
| Dockerfile | ||
| package.json | ||