Kubernetes & deploys · Production incident
A JVM pod that restarts, vanishes or crawls
A JVM service's containers keep restarting at peak. What do you change so that they stop?
Friday 13:30 · pricing-service, 6 podsIllustrative scenario, not a real incident
pricing-service: 14 container restarts in 24 hours
The pricing service is a Java 21 application on 6 pods. Each container has a memory limit of 2Gi and a CPU limit of 1.
Since Tuesday's release the containers have restarted 14 times between them, mostly around the midday peak. Each restart fails the requests that the pod was handling.
What the dashboards and logs show
- Restarts
- 14 in 24 hours across 6 pods, most between 12:00 and 13:30
- Application log
- stops in the middle of a request, with no stack trace
- Failed requests
- about 40 per restart
- p99 latency at peak
- 900 ms, against 120 ms off peak
Clarify · optional · scored under "Hypothesis before change"
Ask the interviewer
0 of 3 asked
Each question reveals a fact. What you ask before you act is part of your score.
First move · scored under "Mitigate before root cause"
Each restart fails about 40 requests, and the next peak is on Monday. What do you do first?
Your rubric score, the model answer, the follow-up questions and the related lessons appear once you have answered the last step.