Fixed
- Retrying a failed combined-encode job now actually re-runs the convert. When a combined convert+cleanup job's video re-encode got discarded (e.g. NVENC crash, or the encoded output was larger than the original), the queue spawned an audio-only follow-up to apply the track cleanup without re-encoding. If that follow-up failed and the user clicked "retry", the old code re-ran the same audio-only job — so the source stayed h264 even though the user expected a fresh h265 convert. v0.7.21 escalates type='audio' retries to type='combined' when the source's
scan_results.needs_conversionflag is still set. Audio-only retries on already-h265 sources stay as-is (legitimate track-cleanup use case). The response now returns{"escalated": "audio→combined"}when the upgrade fires.
Docker images
Pinned to this release:
docker pull ghcr.io/i-ial9000/shrinkerr:0.7.21 # portable (amd64 + arm64, CPU)
docker pull ghcr.io/i-ial9000/shrinkerr:0.7.21-nvenc # x86_64 + NVIDIA GPU (NVENC)
docker pull ghcr.io/i-ial9000/shrinkerr:0.7.21-edge # as above but ffmpeg master
docker pull ghcr.io/i-ial9000/shrinkerr:0.7.21-edge-nvencFloating tags (:latest, :edge, :nvenc, :edge-nvenc) track the most recent release of each lineage.