master hook error on coorinator-relation-changed

[Chris Sanders]

Juju debug-log: https://pastebin.ubuntu.com/p/xCz9zfwtmw/
Juju Status: https://pastebin.ubuntu.com/p/Fs72d8VYXZ/
Master config: https://pastebin.ubuntu.com/p/nbP8HNPxdR/

This is K8s in LXD, with workers deployed in KVM to enable the ceph integration. The error wasn't present on the initial deploy, but I'm not clear when it started. The cluster seems to be working it's a test cluster and I didn't notice the charm status until today.

There are 2 masters as shown on the status and only /1 produces this error, resolving it simply retries and triggers the same error.

[George Kraft (cynerva)]

I've seen similar symptoms on LXD before, when there are two different revisions of the same snap on two different units.

I suspect the kubernetes-master units are on different revisions of cdk-addons. Can you paste the output of the following command?

juju run --application kubernetes-master -- snap list

[Chris Sanders (chris.sanders)]

I've actually redeployed this cluster, I'll watch for this to happen again and check for this condition. Currently they are on the same versions and I'm not seeing the error, maybe on the next round of snap updates I'll see it again. I'll post back here when/if I trigger it again.