Starting A Shell On A Kubernetes Pod

Overview#

I’ve been navigating through Kubernetes infrastructure so I figured I would post some content related to accessing interactive shells within Kubernetes pods. This is useful for troubleshooting. Whether that’d be for the networking for the container or the application.

Getting Pods#

The following command will list all pods in the lab namespace.

kubectl get pods -n lab

Output looks something like this. This output shows the name of the container, how many containers and replicas are ready, current status, restart count, and the age of the container.

NAME                                                 READY   STATUS    RESTARTS      AGE
grafana-fdc74df47-v78gv                              1/1     Running   0             12h

More information is provided by the describe command. But, that is out of scope for this note. I will write a note for that soon.

Starting A Shell On A Pod#

Note: Utilize the -n or --namespace options if you are utilizing name spaces. Personally, I would recommend it if you’re not.

Depending on if a pod has commands or a shell available to jump into. You can utilize the exec subcommand with -i and -t executing /bin/bash for use. Some grafana containers have curl(1) available to utilize.

This will start an interactive Bash shell on the grafana pod.

kubectl -n lab exec -it grafana-fdc74df44-v80gv -- /bin/bash

There is another way to run a temporary network troubleshooting container named netshoot¹. This command will run a temporary netshoot shell using the nicolaka/netshoot container image.

kubectl run tmp-netshoot --rm -it --image nicolaka/netshoot -n lab

Conclusion#

In this note I discuss how to list your pods and start a bash shell for troubleshooting. Which is useful if you ever need to troubleshoot network access or application issues within the container.

A note will be made for get and describe commands soon.