Day2 Kubectl Commands

Background

Here’s a note to self about the various oc or kubectl commands that I keep forgetting about. I call them my Day2 commands because these are more useful once you have actual workloads in your cluster, and you need finer grained navigation of the kubernetes objects.

Finding and Cleaning up Pods

Running an oc get pods command returns a list of all pods, in all various lifecycle phases. It’s sort of confusing to see a pod with status Completed, but kubernetes references it as being in a Succeeded state. You might need to read up the documentation if the command you’re running doesn’t exactly line up with the statuses being presented. https://kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle/

Nonetheless, the command to find these is straightforward.

# This will get you all the pods that are "done" and won't be restarted
oc get pods --field-selector=status.phase==Succeeded

# If you want to delete them, just run it with the delete command
oc delete pods --field-selector=status.phase==Succeeded

# Sometimes pods can be reported as `Completed` but its status is `Failed.
oc get pods --field-selector=status.phase==Failed

Finding and Cleaning up PipelineRuns

When using Tekton Pipelines, a lot of failed/completed PipelineRuns will be collected over the next few days, especially at the beginning when first developing and testing the pipeline itself. There might be a more elegant way to do this, but just in case there isn’t, there’s always the good ole brute force way using bash-fu.

oc get pipelinerun | grep "Failed" | awk '{print $1}' | xargs oc delete pipelinerun

This could get really fancy with timestamp checks, so pipelineruns older than a week (or whatever useful duration you decide) can be pruned, for example.

oc get pipelineruns -o go-template --template ' \n' | awk '$2 <= "'$(date -d '7 days ago' -Ins --utc | sed 's/+0000/Z/')'" { print $1 }' | xargs --no-run-if-empty oc delete pipelinerun

Create or Apply an Object

In my specific use-case, I have a file that I want to as data in a ConfigMap. Normally, you could just run oc create cm my-configmap --from-file=data.txt but if you tried running it twice, you’d get an error saying the object already exists.

Here’s a neat trick to create the file locally, and then use the native upsert ability of oc apply to determine if it needs to insert or update the object.

oc create cm my-configmap --from-file=data.txt --dry-run=client -o yaml | oc apply -f -