v0.9.59
Fixed
- A conversion that fails on a transient "database is locked" now requeues instead of dying. Under heavy concurrency (a rescan holding the write lock past the 60s timeout while conversions run), jobs were permanently failed; they now go back to pending and retry (bounded), so a lock storm no longer discards a running encode.
- A single-folder rescan no longer runs full-library write sweeps. The "restore converted flags" and duplicate-detection passes reset/reloaded the entire scan_results table under the write lock on every scan; they're now scoped to the walked paths (unchanged for a full scan), which stops a folder rescan from locking out running conversions.
Docker images
Pinned to this release:
docker pull ghcr.io/i-ial9000/shrinkerr:v0.9.59 # portable (amd64 + arm64, CPU)
docker pull ghcr.io/i-ial9000/shrinkerr:v0.9.59-nvenc # x86_64 + NVIDIA GPU (NVENC)
docker pull ghcr.io/i-ial9000/shrinkerr:v0.9.59-edge # as above but ffmpeg master
docker pull ghcr.io/i-ial9000/shrinkerr:v0.9.59-edge-nvencFloating tags (:latest, :edge, :nvenc, :edge-nvenc) track the most recent release of each lineage.