Service health, logs and capacity

Check the deployed services

On the packaged deployment host, run epmctl status with the active release controller. Compare its service list with the selected profile. Read a bounded log window for the affected service, for example epmctl logs --tail 100 worker, and check schema readiness, signing-key availability and worker health. For a source build, use the Compose profile’s startup and status commands.

Inspect task state and capacity separately

1. In the selected application, open Monitor. Workload and bottlenecks groups the tasks visible to your account by work family and state. Choose a nonzero cell to filter the task table. Use Clear matrix filter before looking for a task outside that group. Task investigation and recovery
2. As application Administrator, call read_v2_capacity_metrics on the application router with the exact workspace_id and application_id. Compare the returned scope and metricScopes with the incident scope. The queue depths, queue ages and claim counts cover the selected application’s fixed metric families. Connection and database temporary-file counters cover the PostgreSQL database. The memory and temporary-disk readings describe the serving container’s configured resources. Exact tool signature
3. For a disposable check, observe queueDepths.business after admitting one typed job, then again after its publication. The isolated sample showed 1 and then 0. Compare that transition with the exact job receipt and bounded queue wait samples. Measure production capacity separately for the exact deployment and workload.

Record the release and image identities, time window, selected scope and correlation IDs. During recovery, check queue age, expired claims, lock waits, connections and disk pressure. Verify both the affected workload and an unrelated application after recovery.