Daily operation and maintenance windows
Daily checks
Record program hash, IJOB_HOME, service account, ingress and database. Inspect node online state, freshness, roles, scanning and renewal in the dashboard; inspect queues, long runs, failures and executing nodes in the run center. UNKNOWN/stale/unavailable does not mean zero load. A working web page does not prove Worker/SCHEDULER health.
Application logs explain startup, connectivity and coordination; instance logs explain individual executions. Correlate time, instance/trigger IDs and node names; sanitize exports. Native logs are under home/logs; use Compose logs for Docker. Monitor disk, database connections, file/plugin mounts and clock synchronization too.
Maintenance window
- Record schedules/ingress to pause and their original enabled state; pause new business triggers.
- Review active instances; wait for completion or request stop and verify the result.
- Shut down gracefully using platform commands, verify process/container state and take a consistent backup.
- Verify readiness, login, node responsibilities and a separate sample before restoring schedules.
Stopping an instance, disabling a definition, disabling a schedule and stopping the platform have different effects. None reverses committed external side effects. Check the target system before retrying an uncertain result.
Recommended command entry
Use this instance’s ijob.cmd or ijob.sh show/start/stop/restart/status/logs; see Unified entry. Avoid mixing another instance’s legacy environment variables; legacy examples require an explicitly correct home.