Manual workflows fail with startup_failure before any jobs are created

[adricanfarell-blip]

🏷️ Discussion Type

Question

💬 Feature/Topic Area

ARC (Actions Runner Controller)

Discussion Details

Two workflow_dispatch runs on November 11, 2026 remained queued and then ended with startup_failure.

For both runs, the REST API returned no jobs and no artifacts, so there are no job logs available. The web interface displayed “An unexpected error has occurred.”

The workflow file passes actionlint, and PR checks previously passed for the same code tree. One new manual attempt reproduced the failure.

What should I check next, and how can I obtain a diagnostic when no jobs are created?

[aesharhel-bot]

Thanks for sharing this detailed issue. I'm researching GitHub Actions reliability and workflow monitoring.

Do you currently use any external monitoring to detect runner startup failures, or do you rely on GitHub's dashboard and Kubernetes logs?

Also, how much time did diagnosing this issue take?