I m mid update of Rancher to 2 11 from 2 10 and notice a bun Rancher Users #kubernetes

I'm mid update of Rancher to 2.11 from 2.10 and no...

worried-state-78253

06/04/2025, 9:51 AM

I'm mid update of Rancher to 2.11 from 2.10 and notice a bunch of pods with the following in the logs -

Copy code

Error: chart requires kubeVersion: >= 1.25.0-0 which is incompatible with Kubernetes v1.24.10+k3s1

Will the Rancher update (via Helm) update this underlying cluster? The upgrade looked to go well but I've an increasing pod count on its cluster with many helm operations with the same error....

worried-state-78253

06/04/2025, 9:53 AM

Think i figured it out - as Rancher manages the cluster it can be upgraded from the control panel, just did that in cluster management - so hopefully this will sort itself out soon!

worried-state-78253

06/04/2025, 11:53 AM

Ok - lesson learnt, make sure your K8s version is correct before upgrading, doing it after you could end up with a bit of a mess... What I've found is that the rollback works fine - rancher is now running on the previous version, however the cluster upgrade no appears stuck on "removing" I went to the Upgrade controller, plans and deleted the two upgrade plans since the version I chose for k8s wasn't compatible with the rolled-back version of rancher - since it hadn't appeared to start I assumed it would be safe to abort... Now it might be ok - however its been about an hour and still says "removing" ... 1.32 not 2.10 compatible and was the version I'd ran to update to during the attempt to update to 2.11 - if I can get past this my plan is to update to a version both support before updating! Question is - do I wait this out? Or does something need kicking?

hundreds-evening-84071

06/04/2025, 12:29 PM

always a good idea to review support matrix before upgrades OS/K8s/Rancher https://www.suse.com/suse-rancher/support-matrix/all-supported-versions/rancher-v2-11-1/

worried-state-78253

06/04/2025, 12:37 PM

True - however, since rancher is managing the cluster I had anticipated that it would manage this itself, I did check the matrix to make sure the host was supported - which it is, but it didn’t seem obvious that this was something the user needed to check. Given the update is actually just helm rather than rancher itself it seems obvious with hindsight. I was also following the upgrade instructions which also make no mention of checking k8s version.

worried-state-78253

06/04/2025, 12:43 PM

OK - spotted finalisers were present - so manually removed those from these jobs -

Copy code

finalizers:
    - systemcharts.cattle.io/rancher-managed-plan

Then they reappeared as active plans again! The concern is that this k8s version isn't compatible with 2.10, hence trying to stop this progressing... the nodes in this cluster don't appear to be updating yet - no sign of draining etc, so should be possible to stop this...

worried-state-78253

06/04/2025, 1:07 PM

Deleting the plan it just gets recreated... So currently Rancher is trying to update the local cluster to 1.32, that version isn't on the support Matrix, I can't stop the upgrade - I can now delete thrm from the plans (Cluster->local->More resources->Upgrade Controller->Plans) however the upgrade immediately appears in the list again (i think because I removed the finalizer, though begs the question - if it was deleted how did the changes persist?

worried-state-78253

06/04/2025, 1:08 PM

I did backup first anyway so will look at running the restore next!

worried-state-78253

06/04/2025, 2:12 PM

Right - restore done, now begins the upgrade in a better order!

🤞 2

12 Views

Open in Slack

Previous Next