OOMKilled
The container used more memory than its limit and the kernel killed it.
requests.memory is a scheduling hint. limits.memory is the hard ceiling. Exit 137 plus OOMKilled means the limit was crossed.
Commands to start with
- kubectl describe pod POD
- kubectl top pod POD
Explain a OOMKilled paste
Output
The result shows up here. Your input is not written to the log.
Questions
›Will raising the limit always fix it?
Only if the workload legitimately needs more memory. A leak will hit the new limit too. Compare usage over time with kubectl top.
›Why was the pod scheduled if there was not enough memory?
Scheduling uses requests, not limits. A pod can be scheduled and then killed when it exceeds its limit.