Bug type
Regression (worked before, now fails)
Summary
Cron jobs working as expected until upgrading to 2026.3.8 after which they will no longer run.
Steps to reproduce
- Upgrade a working macOS OpenClaw install to 2026.3.8.
- Do not change OpenClaw config after the upgrade.
- Run
openclaw cron list and confirm an existing isolated cron job is present.
- Run
openclaw cron run --id <job_id>.
- Run
openclaw cron runs --id <job_id> --limit 20.
- Run
openclaw gateway status.
- Repeat the manual run and observe job state, run history, and gateway behavior.
Observed after upgrade: cron execution became inconsistent. Sometimes no run entry was created, sometimes gateway restart state was broken, and at least one run later timed out.
Expected behavior
After upgrading, existing isolated cron jobs should continue working as they did before. Running openclaw cron run --id <job_id> should reliably launch the job, create a visible run entry in openclaw cron runs, and then finish with either success or a clear surfaced error. Gateway lifecycle should also remain stable, without stale listeners, port conflicts, or requiring manual LaunchAgent cleanup/reinstall.
Actual behavior
After upgrading to 2026.3.8, isolated cron behavior became inconsistent. Manual openclaw cron run --id <job_id> was accepted/enqueued, but often did not immediately create a visible run entry in openclaw cron runs. In some cases the job stayed idle with no run history; in another case it moved to running and later failed with Error: cron: job execution timed out. At the same time, gateway lifecycle became unstable: openclaw gateway stop could unload the LaunchAgent while leaving a gateway process still listening on port 18789, causing gateway already running / port already in use errors until manual cleanup/reinstall.
OpenClaw version
2026.3.8
Operating system
macOS 26.3
Install method
npm global
Model
codex-5.4
Provider / routing chain
openclaw > codex 5.4
Config file / key location
No response
Additional provider/model setup details
No response
Logs, screenshots, and evidence
Impact and severity
No response
Additional information
No response
Bug type
Regression (worked before, now fails)
Summary
Cron jobs working as expected until upgrading to 2026.3.8 after which they will no longer run.
Steps to reproduce
openclaw cron listand confirm an existing isolated cron job is present.openclaw cron run --id <job_id>.openclaw cron runs --id <job_id> --limit 20.openclaw gateway status.Observed after upgrade: cron execution became inconsistent. Sometimes no run entry was created, sometimes gateway restart state was broken, and at least one run later timed out.
Expected behavior
After upgrading, existing isolated cron jobs should continue working as they did before. Running openclaw cron run --id <job_id> should reliably launch the job, create a visible run entry in openclaw cron runs, and then finish with either success or a clear surfaced error. Gateway lifecycle should also remain stable, without stale listeners, port conflicts, or requiring manual LaunchAgent cleanup/reinstall.
Actual behavior
After upgrading to 2026.3.8, isolated cron behavior became inconsistent. Manual openclaw cron run --id <job_id> was accepted/enqueued, but often did not immediately create a visible run entry in openclaw cron runs. In some cases the job stayed idle with no run history; in another case it moved to running and later failed with Error: cron: job execution timed out. At the same time, gateway lifecycle became unstable: openclaw gateway stop could unload the LaunchAgent while leaving a gateway process still listening on port 18789, causing gateway already running / port already in use errors until manual cleanup/reinstall.
OpenClaw version
2026.3.8
Operating system
macOS 26.3
Install method
npm global
Model
codex-5.4
Provider / routing chain
openclaw > codex 5.4
Config file / key location
No response
Additional provider/model setup details
No response
Logs, screenshots, and evidence
Impact and severity
No response
Additional information
No response