The task view shows current or latest execution information and the next scheduled time. It is not an unlimited execution-history archive.
The API field last_run is the latest observed task creation time, not its end. Missing task records do not prove success. Timeout stops are recorded as XJobFailed events; if the execution record has already disappeared, consult events and logs.
Inspect a result
Check the task name, scheduled or manual trigger, state, timestamps, and result. Open application logs with the XJob source and the matching task name to inspect output.
A waiting task has not yet completed. A running task has not yet reported success. A successful exit means the command returned zero, which must represent the intended business outcome.
Diagnose failures
| Symptom | Check |
|---|---|
| Task does not start | Eligible servers, resources, image pulling |
| Command not found | Image contents, path, explicit shell |
| Cannot reach service | Group connectivity, host and port, credentials |
| Runtime exceeded | Command timeout, blocked dependency, maximum runtime |
| No expected output | Correct task log source, range, driver and retention |
| Side effect repeated | Command’s duplicate-handling behavior |
The job runs in a separate container, so it cannot rely on transient files inside an application container. For persistence or external systems, verify the actual accessible data source.
Retry deliberately
Inspect whether the previous command partially completed before retrying. A nonzero exit does not automatically undo database writes or external requests.
After changing the command or schedule, verify a fresh execution and the next-run time. If you require a complete business history, record it in the business system rather than assuming the task’s latest status provides it.