# Monitoring Monitoring is built from provider snapshots stored in SQLite; normal pages do not call Plex or Jellyfin directly. ## Tabs - **Overview** — aggregate users, servers, watch time and top media. - **Location** — [user map](monitoring-map.md), live filters, heatmap and historical timeline. - **Now Playing** — live sessions, progress, playback decisions and artwork. - **Activity** — time-based playback trends. - **History** — completed sessions and session details. - **Libraries** — playback, [library growth](library-growth.md) and deferred media rankings. - **Users** — watch activity and usage per user. - **Servers** — activity and availability by server. - **Usage risk** — repeated IP/device patterns and suggested plan upgrades. - **Policies** — active rules and paginated enforcement history. - **User portal** — portal usage when the portal is enabled. Collector delays do not immediately delete Plex sessions. Missing sessions are confirmed across bounded refreshes, and a recent snapshot fallback is only used while the processing pipeline is genuinely busy. ## Policy scope Policies may target a user, server or global provider context. Available rules include stream/IP limits, server transcode limits and 4K-transcode blocking. Violations can warn or stop playback. Test new rules with conservative limits and review enforcement history before broad rollout. ## Loading and freshness Overview renders navigation, server state and live counts before its historical rankings load. The library table also appears before its most-watched-title section, which loads when visible. A loading placeholder is not a final zero. Historical aggregates reuse recent results; current collection delays can leave the last good snapshot visible. Automatic refresh pauses when the page is hidden and resumes on return. The cached artwork proxy can show a previous image while refreshing it in the background. Historical computation and remote image retrieval can still take time; see [Performance](performance.md) to distinguish them. ## Usage-risk recommendations Reports calculate the current filtered suggestions on read. A scheduled task refreshes recommendation history every 30 minutes, and the sender checks eligibility again before notification. Opening a report does not itself persist or send a recommendation. Location or shared IP observations alone are not proof that accounts are being misused.