PR metrics comment now non-blocking
Webhook failures no longer block PRs—the metrics comment script now gracefully handles 500s and network errors instead of failing the CI check.
The QA metrics webhook occasionally fails—returning 500 errors or becoming unreachable when the internal service is down. Previously, these infrastructure issues caused the CI step to fail, marking pull requests with red X checks unrelated to the actual code changes.
The PR metrics comment script has been updated to handle webhook failures gracefully. Network errors are now caught and logged as warnings, and non-ok HTTP responses (like 500s) also exit with warnings instead of crashing. In all cases, the script exits cleanly with code 0.
Developers can now merge code without waiting for the QA metrics service to recover, reducing unnecessary pipeline retries and noise in PR statuses.
View Original GitHub Description
Summary
- The QA metrics webhook can return 500 or be unreachable (e.g. internal n8n instance is down), which previously caused the CI step to
exit 1and show as a failed check on PRs - Wrapped the webhook fetch in try/catch to handle network errors (timeouts, DNS, connection refused)
- Changed non-ok HTTP responses from
exit 1toexit 0with a warning log
Test plan
- Verify the script exits 0 when the webhook returns 500
- Verify the script exits 0 when the webhook is unreachable (network error)
- Verify normal operation is unaffected when the webhook is healthy
🤖 Generated with Claude Code