Latest Blog

Preserve Jellyfin config first, then repair or roll back the container layer; a failed image update should not become a database or library reset.

Treat post-restart Jellyfin failures as dependency changes first: verify mounts, app-data writes, GPU access, and network paths.

Choose Plex household accounts by the identity and access boundaries you need, then validate managed-user limits and recovery behavior before scaling sharing.

Distinguish Plex process growth from Linux cache, then investigate only when memory rises across repeated cycles or creates real pressure.

Audit Plex storage roles before failure: separate live state, rebuildable data, media, and backups, then prove the layout with a timed restore.

Measure Plex CPU headroom from the busiest normal workload and a clear failure threshold instead of reserving an arbitrary percentage.

Test Plex offline behavior by separating local reachability, cached sessions, user switching, and remote access instead of assuming every client works the same

A backup is not proven until a clean Plex instance can restore state, permissions, representative users, playback, and an older recovery point.