Deploying Kubecost helm receipt failed

Hello,

I have a created a new k8s cluster with Karpenter and on this cluster I can’t deploy kubecost I have this error : ready.go:284: [debug] PersistentVolumeClaim is not bound: z6138a1e6-tools-development/helm-zeb8ba73f-kubecost-cost-analyzer

I don’t understand why it’s not working because I just clone the Kubecost helm application from my other (production and non karpenter cluster) that works.

Any idea of what happened ?

Thanks

Hello @Orkin ,

I’ll check with the team and get back to you.

Regards,
Charles-Edouard

Hello @Orkin,

Looking deeper into this error, it seems that there is no storage class set for the chart.
Looking at the chart doc, it seems you can set both storage classes:

  • prometheus.server.persistentVolume.storageClass
  • persistentVolume.storageClass

Those have to be set on the new version.

Looking at your setup, kubecost on dev cluster is set to version cost-analyzer:2.4.3 whereas the production one (working) is set to cost-analyzer:2.1.0.

Can you try to either deploy the same version as your production cluster cost-analyzer:2.1.0 OR update chart values to set those missing values?

Cheers

CleanShot 2024-11-27 at 09.52.22
CleanShot 2024-11-27 at 09.52.13

Hello @bchastanier,

I have the same problem with the version 2.1.0. That’s why it’s strange. Do you have some advices about storageClass to provide ? I never had to do this :sweat_smile:.

What I can understand it’s there is no default storage class setted on my cluster and can see some of them


I can set one of them as default, but don’t know if there is an incidence with my qovery managed cluster ?

Thanks

Actually, there is no default storage class declared, it used to be GP2, but it’s not anymore.

The issue is on our side and boils down to a regression most likely.
You should be able not to set any storage class and default one will be picked-up.

I am investigating and will work on a fix. I will let you know once fixed.

Ok thanks for your feedback ! To unlock me have you the ability to put the gp2 as default storage class manually or I should have to wait a fix ?

Thanks

I just did set it as default, let me know how it goes for the time being.

Yes it workds, thanks a lot @bchastanier

Hello @Orkin !

The fix has just landed and is available for NON production clusters and will be available for production clusters on Monday.

In the meantime, if you need this patch on production cluster, let me know I can patch it.

Cheers

Thanks it’s fine for the production cluster, there is already a default StorageClass :slight_smile:.

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.