02-debugging.md

Debugging issues in Kubernetes

Using Liveness and Readiness Probes in Kubernetes

Simulation of readiness and liveness loss:

kubectl create ns debug-space
kubectl create -f src/basics/02-pod-probs.yaml -n debug-space
kubectl get pods -n debug-space -w

Creating a Termination Log

The application can write down the reason of termination to a pre-defined path, /dev/termination-log is the default, so that one can see the termination reason quickly by checking the Pod definition.

kubectl create -f src/basics/02-termination-pod.yaml -n debug-space
kubectl get pod termination-pod -o yaml

The output:

  containerStatuses:
  - containerID: docker://dc335ae8909454f2856bbe9abb8c6463316ff0455917dfe81fd6aba9d0e12d00
    image: busybox:latest
    imageID: docker-pullable://busybox@sha256:ba76950ac9eaa407512c9d859cea48114eeff8a6f12ebaa5d32ce79d4a017dd8
    lastState:
      terminated:
        containerID: docker://c8702390c534f64acbaf29d768e454cc31b98f01a014df2f664dd3001ee988cc
        exitCode: 0
        finishedAt: "2023-12-28T15:55:54Z"
        message: |
          Done sleeping
        reason: Completed
        startedAt: "2023-12-28T15:55:34Z"

Checking logs

To check the logs of the Pod using kubectl log:

kubectl logs <pod-name>

Get the logs of the previously crashed container (via specifying the --previous flag):

kubectl logs --previous <pod-name>

To attach shell to the running process inside the container:

kubectl attach <pod-name> -i -- sh (any command)

Clean up

kubectl delete pods --all -n debug-space