logreader / kb /oom_killed.md
PatrickRedStar's picture
add
29fdac9
# OOM / Memory Pressure
## Symptoms
- Pod terminated with `OOMKilled` or JVM `OutOfMemoryError`
- Memory usage climbs until eviction
- Core dump or heap dump generated
## Checks
- Compare pod/container memory limits vs peak usage
- Inspect heap/thread dumps for leaks or unbounded caches
- Validate GC settings and memory flags
- Check for large payloads or unbounded batching
- Review sidecar/agent memory consumption
## Fix
- Right-size memory requests/limits
- Fix leaks or cap caches/buffers
- Split large batches/payloads
- Adjust GC or runtime memory options