Major severity Resolved

Elevated failure rates for GitHub-hosted CI runs

Affected: GitHub Cloud, Developer Portal
Started: May 14, 2026, 13:42 UTC
Resolved: May 14, 2026, 15:20 UTC
Duration: 1h 38m

Updates

Resolved May 14, 2026, 15:20 UTC

CI success rates have been stable for 30 minutes and the backlog of queued jobs has cleared. This incident is resolved. A brief retro will be linked below once available.
Monitoring May 14, 2026, 14:48 UTC

GitHub has applied a fix and success rates have returned to normal. We are monitoring queued and retried jobs to confirm full recovery before closing.
Identified May 14, 2026, 14:05 UTC
The issue has been identified as an upstream GitHub incident affecting runner image pulls in us-east. Impact is limited to:
- Workflows using ubuntu-latest hosted runners
- Backstage software templates that kick off CI on creation
Self-hosted runners are not affected. Teams needing an urgent build can re-run on a self-hosted runner label as a workaround.
Investigating May 14, 2026, 13:42 UTC

We are investigating elevated failure rates on GitHub-hosted Actions runners. Pipeline jobs that pull container images are timing out. Backstage scaffolder templates that trigger CI are also affected.

Summary

A roughly 90-minute degradation of GitHub-hosted CI runners caused intermittent job failures, primarily for workflows pulling container images. The root cause was an upstream GitHub incident. No data was lost and self-hosted runners were unaffected.

Follow-ups

Document the self-hosted runner fallback in the platform runbook.
Add a dashboard alert for CI success-rate drops below 95%.