DevOps Troubleshooting
The errors you actually hit, with the causes ranked by how often they are responsible and the exact command that confirms each one. No product pitch, no filler — paste the error, find the cause.
Docker errors
- Cannot connect to the Docker daemonThe Docker CLI cannot reach the daemon. Six causes — daemon not running, group membership, wrong context, Desktop not started, WSL integration, CI without a daemon.
- Docker: no space left on deviceDocker filled the disk. Find what actually consumed it — images, build cache, volumes or the container logs that prune never touches — and stop it recurring.
- exec format errorYour container image was built for a different CPU architecture — usually arm64 on Apple Silicon running on amd64 servers. How to confirm it and build multi-arch.
Helm errors
Kubernetes errors
- CrashLoopBackOffYour container starts, exits, and Kubernetes restarts it with growing delay. Here is how to find which of the six causes is responsible, in the order they actually occur.
- ImagePullBackOff and ErrImagePullKubernetes cannot pull your container image. The events name the exact reason — here is how to read them and fix each of the five causes.
- OOMKilled (Exit Code 137)The kernel killed your container for exceeding its memory limit. Why the application logged nothing, how to tell a leak from a bad limit, and the JVM trap.
- Pod stuck in Pending — 0/N nodes are availableThe scheduler could not place your Pod. The FailedScheduling event names every reason per node — here is how to read it and fix each one.
- Service has no endpointsYour Kubernetes Service accepts connections and forwards them nowhere. There are exactly two causes — a selector that matches nothing, or Pods that are not Ready.