Repository navigation
Conversation
smoke.yml and the post-deploy smoke in deploy.yml now alert once when the smoke test starts failing and once when it recovers, and stay silent while the status is unchanged. The previous result comes from the GitHub API (previous completed run of the same workflow and target), using the job's GITHUB_TOKEN with actions: read; no new secrets or state store. - telegram-alert: new status-change.mjs decides failed/recovered/none; an unknown previous result alerts on failure and stays silent on a pass. - smoke.yml: run-name carries the environment so runs can be matched. - deploy.yml: smoke compared per branch using the smoke step's conclusion; deploy failures still alert every time. A rollback to a commit without smoke.js now skips the step instead of reporting a pass. - e2e-nightly.yml is unchanged and keeps alerting on every failure.
The change check compared whole-run conclusions, so a lost Telegram send, an install flake, or a run that never finished the smoke test counted as "already alerted" and silenced the next real failure. Give the smoke step an id and compare its outcome and the previous run's step of the same name, as deploy.yml does. Failures before the smoke test finished alert every time in a separate step.
The checkout now runs after successful deploys too, so a GitHub fetch error would turn a good deploy red and then run an alert step with a missing action. Make it continue-on-error and run the alert steps only when it succeeded.
curl --retry 2 shrinks the chance that a transient Telegram error loses an alert. If status-change.mjs exits 0 without writing send= (for example when the entry-point guard is false), alert on failure like any other script error. A spawn test covers the script run as a program, and the test that asserted the impossible "current run is in the completed list" case is trimmed.
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
Warning Review limit reachedYou've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository. Next included review available in 32 minutes. View limit details
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. |
|
|
❌ The last analysis has failed. |
Problem
Scheduled smoke runs every 30 min and alerts on every failing run, so one broken check pings repeatedly.
Fix
Compare with the previous run of the same workflow and environment. Alert once on pass->fail and once on fail->pass; stay silent otherwise. Deploy failures still alert as before. CONTRIBUTING updated.