Fix: Kubernetes Pod OOMKilled

!Symptoms

• Pod status shows OOMKilled
• Container exits with code 137
• Application suddenly terminates without graceful shutdown logs
• Node memory pressure events in kubectl describe node

?Root Causes

• Memory limits set too low for the workload
• Memory leak in the application
• JVM heap size exceeding container memory limit
• Large in-memory caches or data structures growing unbounded
• Multiple processes inside the container consuming more than allocated

#Diagnosis Steps

1. 1Run `kubectl describe pod <pod-name>` and check the Last State reason for OOMKilled
2. 2Check `kubectl top pod <pod-name>` to see current memory usage
3. 3Review container memory limits in the deployment spec
4. 4Check node-level memory with `kubectl top nodes`
5. 5Use `kubectl logs --previous` to find any memory-related messages before the kill

>Fix

1. 1Increase memory limits in the pod spec if the workload legitimately needs more
2. 2For JVM apps, set -Xmx to ~75% of the container memory limit
3. 3Fix memory leaks in the application code
4. 4Add memory-aware garbage collection tuning
5. 5Split large workloads across multiple pods with horizontal scaling

*Prevention

• Right-size memory limits based on profiling under realistic load
• Set up memory usage alerts before hitting limits (e.g., at 80%)
• Use Vertical Pod Autoscaler (VPA) to automatically adjust resource limits
• Implement proper connection pooling and cache eviction policies
• Load test with realistic data volumes before deploying

Related Error Messages