OpsCortex
Back to archiveIncident Response

When Checkout Broke, I Did Not Guess

The first minute of an incident should be boring: describe, events, previous logs, then mitigate.

Production idea

CrashLoopBackOff, OOMKilled, event timelines, previous logs, and rollback.

Command set
kubectl describe pod <pod>kubectl get events --sort-by=.lastTimestamp -n <ns>kubectl logs <pod> --previous
Monday morning
  • Read events before guessing.
  • Use logs --previous for crashed containers.
  • Capture events before they expire.
  • End with owned, dated postmortem actions.