Understand the results

Troubleshoot a result

Start with the affected job and its most recent run. Read the status explanation, then check its output and timeline. The next step depends on whether work failed, stopped reporting, or never started.

A command failed

Read the first relevant error in the output tab. Check the command’s paths, working directory, environment variables, and permissions on its target machine. A command that works in your interactive shell may depend on a different user or environment.

After correcting the cause, use Run again and inspect the new result. Each attempt keeps its own history.

A machine is offline

On that machine, check the agent and its recent service logs:

omj-agent doctor
sudo journalctl -u omj-agent -n 100 --no-pager

Confirm the machine can reach the configured server URL. If a reverse proxy sits in between, check its timeout settings against the reverse proxy reference.

Then inspect the job’s missed-run policy. Skip records the occurrence as missed. Run late allows it within the configured grace period. Reconnecting does not replay every occurrence from an outage.

A run is lost

Lost means the server could no longer confirm the run’s state. It does not prove that the underlying command failed or stopped.

Check the machine, the actual operation, and the agent’s logs before starting another attempt. The agent troubleshooting reference explains restarts, lost contact, and late completion reports.

A monitored job has no reports

A new monitored job does not watch for missing work until its first report arrives. Run the reporting integration once and confirm that the result appears on the job page.

If no report appears, check the HTTP request’s response and make sure it uses the current ping URL from that job. An unknown token returns 404. Keep the token private.

If reports arrive but occurrences are marked Missed, compare the job’s expected schedule and timezone with the scheduler that actually runs the work. Allow enough time in Report within for completion and ordinary delays. See Monitor existing work for the complete integration.

The status is visible, but no alert arrived

The current development app has notification preferences, but email and webhook delivery are not yet implemented. Inspect failures and missing reports in the app; do not depend on a notification reaching you.

Once your jobs are reporting reliably, continue with Operating your server to protect their data and keep the service running.