Cleanup routine after failed `bunk up` operation

[pogzie]

Dustin, I noticed that when trying to run bunk and forgetting to have Docker running beforehand, bunk fails with an exit status 1.

:
Writing 60 serviceaccounts resources from file: serviceaccounts.yaml
Writing 28 services/specs resources from file: services.yaml
Writing 1 signingkeies resources from file: signingkeies.dex.coreos.com.yaml
Writing 1 statefulsets resources from file: statefulsets.apps.yaml
Writing 1 storageclasses resources from file: storageclasses.storage.k8s.io.yaml
Writing 4 validatingwebhookconfigurations resources from file: validatingwebhookconfigurations.admissionregistration.k8s.io.yaml
Skipping empty volumeattachments resource file: volumeattachments.storage.k8s.io.yaml
Writing 1 volumesnapshotlocations resources from file: volumesnapshotlocations.velero.io.yaml
2021/07/14 13:41:28 Creating k3d cluster
2021/07/14 13:41:28 exit status 1

After I've launched Docker and re-run the command, it complains about a .kbk already existing. I assume this is some kind of "session" or "state" from the previous launch?

In anycase, I just rm -rf .kbk on the bundle directory and re-run bunk up and everything is ok.

Im thinking that a cleanup routine can be added to automatically purge this folder if a launch fails? Or is it worth keeping for bunk run logs?

[some-things]

@pogzie Thanks for the report 👍 . Most likely the issue was that you had a leftover k3d cluster as it fails when trying to create the k3d cluster, but it is interesting it worked for you just removing the .kbk dir. Another simple option is bunk down && bunk up, which will delete the .kbk dir if it exists, and also remove any existing k3d cluster, then create the fresh cluster.

One idea was to create a user .kbk dir which would contain a file for tracking bunk clusters. That way if you 'forget' one, the tool will be a bit more intelligent. Will have to think about some improvements with DKP2 bundle.