Docker

Problem

A container stops unexpectedly and `docker inspect` or `docker ps -a` shows exit code 137, meaning the Linux kernel's OOM killer terminated the process for using more memory than allowed.

Symptoms

• Container exits with code 137
• `docker logs` shows no application error before the exit
• `dmesg` or `journalctl -k` shows an oom-kill event for the container's process
• Container restarts repeatedly if using --restart unless-stopped

Root Cause

The container's process exceeded its memory limit (either a Docker --memory limit or the host's available memory), and the kernel's OOM killer sent SIGKILL to reclaim memory.

Solution

docker inspect <container> --format='{{.State.OOMKilled}}'

docker inspect <container> --format='{{.HostConfig.Memory}}'

docker stats <container> --no-stream

docker run -d --memory=1g --memory-swap=1g <image>

services:
  app:
    deploy:
      resources:
        limits:
          memory: 1g

docker exec <container> ps aux --sort=-%mem | head -10

Prevention

• Set explicit memory limits on every container so a leak fails predictably instead of taking down the host
• Add a health check that restarts before memory grows unbounded
• Monitor container memory with cAdvisor, Prometheus node-exporter, or `docker stats`
• Profile memory usage in staging under realistic load before shipping to production

References

• Docker runtime memory constraints
• Understanding OOM in Linux