Log Analytics
Local-First
← Back to analyzer

Troubleshooting

Kubernetes Pod OOMKilled

Container terminated because memory usage exceeded the cgroup limit.

SQL Lens

Run this query locally to surface offenders.

SELECT message, time FROM logs WHERE message ILIKE '%OOMKilled%' ORDER BY time DESC
Investigate in HUD

Playbook

  1. Compare pod memory requests/limits with actual peak usage in metrics.
  2. Capture heap/pprof snapshot from workload before eviction.
  3. Add vertical pod autoscaler or tune GC flags for the runtime.