This post is only available in its original language.

A Minimal kubectl Debugging Toolbox

Sample post — seeded as placeholder content for this site’s test suite, not a claim about any specific team’s process. Replace or remove before this post is treated as real.

Most incidents don’t need a service mesh dashboard or a tracing UI. They need kubectl describe, kubectl logs --previous, and someone who reads the output slowly.

The core loop

kubectl get pods -o wide
kubectl describe pod <name>
kubectl logs <name> --previous --tail=200

describe tells you what the scheduler and kubelet think happened — Events at the bottom is almost always where the actual story is. logs --previous matters the moment a container has already restarted; by the time you look, the current log stream may only have a few seconds of output.

Watching state change

kubectl get pods -w
kubectl get events --sort-by=.lastTimestamp

Neither of these replaces metrics or tracing. They’re just cheaper to reach for, and cheap tools you actually reach for beat expensive ones you have to go set up mid-incident.