Updating HA clusters fails #857
Assignees
Labels
bug
Something isn't working
Comments
|
I checked behavior with the newest k0smotron version |
|
Yeah, thanks for reporting the issue! I prepared a fix, we'll try to merge and release new version asap |
|
I checked the latest image with the fix present. The etcd pods indeed are not crashing anymore, the other ones do though 🤔 It does happen only when upgrading the cluster. May I ask you to reopen the issue if it turns out to be still k0smotron-related, @makhov ? |
|
Yeah, I'll check. Could you also send a full log of crashing pod? |
|
Sure, the logs look like this. Then it just repeats itself Toggle, not to clutter the issue |
|
@franpog859 sorry, I did a lot of tests in different configurations and was not able to reproduce the issue. Also, nothing in the logs above shows the problem. Looks like the liveness probe is failing, but I can't see why. Are you still experiencing the issue? Could you show me the full Cluster object? |
|
Sorry for the delay, I'm still having this issue. I managed to make it work by removing the apiVersion: k0smotron.io/v1beta1
kind: Cluster
metadata:
name: basic-k0s-test
namespace: infra
spec:
replicas: 3
k0sImage: k0sproject/k0s
version: v1.30.4-k0s.0
# persistence:
# autoDeletePVCs: false
# persistentVolumeClaim:
# apiVersion: v1
# kind: PersistentVolumeClaim
# spec:
# accessModes:
# - ReadWriteOnce
# resources:
# requests:
# storage: 10Gi
# volumeMode: Filesystem
# type: pvc
monitoring:
enabled: false
etcd:
autoDeletePVCs: false
persistence:
size: 10Gi
service:
type: LoadBalancerWhen the I'll try to test if my use case could work without the persistence config. Either way, the persistence described in HA docs is needed, I think it's a bug in the persistence configuration or the documentation then |
|
Thanks for the clarification! I'll check it with persistence enabled |
|
Yeah, I was able to reproduce the issue with persistence enabled. I've checked updates from and to different k0s versions, and looks like the issue was introduced in k0s v1.31.1+k0s.0, earlier releases work fine. @franpog859 persistence is not required anymore for HA clusters, I'll fix it the docs, thanks for pointing it out! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
What happened?
I created a HA cluster, the
k0smotron.io/v1beta1Clusterspec.replicasequals3. I tried to follow the docs to update the k0s, I changed thespec.versionfromv1.30.4-k0s.0tov1.31.2-k0s.0. The impact is as follows:Involved Areas
Standalone
What did you expect to happen?
The cluster is upgraded and works fine. Maybe I use wrong documentation, could you point me to some working instructions?
Steps to reproduce
Create a cluster with
spec.replicas=3, wait for it to start working, edit the cluster and bump the version, watch the cluster burnk0smotron version
v1.1.2
k0s version
v1.30.4-k0s.0
Anything else we need to know?
No response
The text was updated successfully, but these errors were encountered: