Improve cron workflow robustness and diagnostics

[tbrandenburg]

Motivation

• Cron-triggered workflows were occasionally missing runs and offered limited runtime visibility, making failures hard to debug.
• The scheduler should tolerate brief pauses and avoid overlapping runs while capturing richer per-job metadata to surface issues in the API and logs.

Description

• Hardened scheduler registration by setting max_instances=1 and adding misfire_grace_time=300 to reduce missed or overlapping runs and make behavior predictable.
• Added explicit warning logs when workflows are skipped for missing schedule, missing shellScriptPath, or missing script files to aid immediate diagnosis.
• Captured per-job diagnostics in runtime state: lastStartedAt, lastFinishedAt, lastDurationMs, lastExitCode, lastError, nextRunAt, and running by introducing get_cron_job_diagnostics and backing fields (_last_finished_by_job, _last_duration_ms_by_job, _last_exit_code_by_job, _last_error_by_job).
• Exposed diagnostics in the workspace workflows API by wiring get_cron_job_diagnostics() into list_workspace_workflows so UI/clients can display runtime status alongside lastRun.
• Added a failedJobsSinceStartup counter to get_cron_clock_status and ensured runtime diagnostics maps are cleared on clock start.
• Small behavior adjustments: new logs and ensured single-instance per job, and updated/added unit tests to cover diagnostics and API wiring.

Testing

• Ran unit tests with uv run --project packages/pybackend python -m pytest packages/pybackend/tests/unit/test_cron_service.py packages/pybackend/tests/unit/test_workflow_service.py packages/pybackend/tests/unit/test_api.py which executed the test suite and all tests passed (109 passed, 1 warning).

---

Codex Task

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
made	Ready	Preview, Comment	Mar 14, 2026 6:21am