Forced to Delete Plugins in Web GUI before updating Botkube Version

[evwitmer]

Description

When I go to update versions (last time it was 1.6->1.8. This time it was 1.8->1.10) it forces me to delete all the plugins that I have installed in the web GUI before it lets me update.
Maybe it is just my setup (Docker>Minikube on a Mac M1 Air) but it has happened twice now. I recorded it happening here. The volume is low, but I have voice on the videos, I promise!

Failing to Upgrade Botkube Version Video: https://drive.google.com/file/d/1l1D7rDrpZbxRsTgOSYL5g90c-kykQ5bT/view?usp=drive_link

Botkube Upgrade working after deleting all plugins: https://drive.google.com/file/d/1jHafkUVPlExU-Taz3hCHaOuHx4CG4GU7/view?usp=sharing

Expected behavior

Running the update command will upgrade Botkube from 1.8 to 1.10 while keeping my plugins.

Actual behavior

Botkube fails before the update can finish unless I go into app.botkube.io and delete all plugins first and then run the command

Steps to reproduce

1. Run Upgrade command with Kubernetes, Prometheus, or Doctor plugin installed on an outdated version of Botkube (1.9, 1.8....)
2. Wait for update to fail in the terminal.

[pkosiec]

Reposting original answer from https://kubeshop.slack.com/archives/C03MRCX7UE9/p1714978750726999?thread_ts=1714662799.959579&cid=C03MRCX7UE9:

Thanks Evan for the report!
Based on the first recording it looks like Botkube restarted itself because of timeout on kubectl download. It might be related to internet connectivity, or dl.k8s.io had some issues 🤔 At later stage we should mirror all of the dependencies to fetch them from GCS.
So if CLI says "upgrade failed", the actual upgrade was executed successfully but Botkube restarted with an error. However, with such networking issue, probably a single restart or two should resolve this.

You could try to adjust the time when Kubernetes restarts Botkube pod with the livenessProbe configuration (https://github.com/kubeshop/botkube/blob/main/helm/botkube/values.yaml#L742C5-L749C24):

e.g.

--set deployment.livenessProbe.initialDelaySeconds=30

Next time if you'll have such issue please try that and if that doesn't work, please keep the failed Botkube instance as it is for further debugging, ok? 👍 Thank you!