Highlights
- Safer provider synchronization: authorization and restore decisions are rechecked inside the write transaction, and downloaded catalogs are rejected if provider fetch settings changed. Existing user assignments, customization and channel order are preserved.
- More reliable series synchronization: eligible accounts sharing a source can act as fallbacks, including identical series IDs. Account failures and local database errors are handled separately, and refused HTTP responses release their bodies and sockets.
- Lower EPG overhead: unchanged logo catalogs remain cached across freshness checks. Failed scheduled custom-source and provider EPG updates wait fifteen minutes after failure before retrying; manual updates remain available.
- SQLite and scheduling improvements: shorter/batched write operations, bounded catalog concurrency, configurable worker count, improved lock handling and cleanup, and clearer partial-sync reporting.
- Improved operational diagnostics and credential protection: upstream credentials are redacted from errors and logs. WAL diagnostics distinguish pending frames from allocated space, while checkpoints remain PASSIVE.
Container image
This release points to commit 14e7711, the same revision as the container published on September 27, 2026.
- Image at publication:
ghcr.io/bladestar2105/iptv-manager:latest - Immutable image reference:
ghcr.io/bladestar2105/iptv-manager@sha256:97c303764a0a5283adb90cc783b4ca0914ab7c52156c1e81bcbc1d850243b280
For Compose deployments, also review the current Compose configuration when upgrading: settings such as log rotation are not applied by pulling a new image alone. See the configuration documentation.
Validation
- Local full suite: 142 test files; 1,491 passed and one Linux-specific test skipped.
- Lint, build, public JavaScript/template syntax checks and Playwright smoke passed.
- GitHub validation and Docker pipeline passed for this commit, including Node 24 tests, dependency audits, native sandbox verification, Docker build and container smoke tests.
Live-provider load testing and signed-in external AI acceptance were not performed. Issue #630, concerning Strong 8K timeouts, has not been confirmed resolved by this release.