rancher / fleet

Deploy workloads from Git to large fleets of Kubernetes clusters

Home Page:https://fleet.rancher.io/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Fleet 0.6.0 shipped with Rancher 2.7.3 still problems with kubeVersion

bennysp opened this issue · comments

Is there an existing issue for this?

  • I have searched the existing issues

Current Behavior

I installed Rancher 2.7.3 fresh and charts like Longhorn (1.4.1) (and others) are failing with:

Chart requires kubeVersion: >= 1.22.0-0 which is incompatible with Kubernetes v1.20.0

I did see issue #399 which was just closed out hours ago, but it seems that 2.7.3 still has version 0.6.0 of the fleet controller where this issue is identified.

(Installing the Longhorn with the same values via Marketplace works just fine)

Expected Behavior

It should read the real kubernetes versions so it doesn't break various charts (so far, Longhorn, Hashicorp Vault).

Steps To Reproduce

  1. Install Rancher 2.7.3 on RKE2 1.25.7
  2. Create a fleet.yaml with the below values and add to Continuous Delivery
defaultNamespace: longhorn-system
helm:
  releaseName: longhorn
  chart: longhorn
  repo: https://charts.longhorn.io
  version: 1.4.1

Environment

- Architecture: amd64
- Fleet Version: 0.6.0
- Cluster:
  - Provider: RKE2
  - Options:
  - Kubernetes Version: 1.25.7

Logs

No response

Anything else?

No response

Yes, this should be fixed in 0.7.0

@manno Sorry for the bump, but when will it land into a stable version of Rancher 2.7 ?

This is real trouble since I can't upgrade to v2.7.5-rc1 ... I'm running 2.7.3 right now - too 😞 And I don't feel like it's a good idea to update fleet myself ...

i'm running 2.7.5-rc1 but it still doesnt work. I am installing hashicorp/vault which has kubeVersion: ">= 1.22.0-0" listed in the chart.
Getting Chart requires kubeVersion: >= 1.22.0-0 which is incompatible with Kubernetes v1.20.0
Cluster runs on v1.25.9

@manno To add to what @wagnst said: We had this issue before, but it (apparently temporarily) disappeared after the upgrade from rancher 2.6.x to 2.7.3, but it is back now. I think it started happening after upgrading my cluster to either 1.24.x or 1.25.x.

nvm, we figured it out: the rancher k8s had disk pressure, which prevented the deployment of the new fleet.

@wagnst
I tried the following on a parallel rancher test cluster and it seems to work...

helm -n cattle-fleet-system upgrade --create-namespace --wait    fleet-crd https://github.com/rancher/fleet/releases/download/v0.7.0-rc.2/fleet-crd-0.7.0-rc.2.tgz
helm -n cattle-fleet-system upgrade --create-namespace --wait    fleet https://github.com/rancher/fleet/releases/download/v0.7.0-rc.2/fleet-0.7.0-rc.2.tgz

That's just a temporary patch though ... It was enough for me to prove that an upgrade will help.
That is why I asked when will the patch lands in a stable version of Rancher... This issue right now is leaving us with no choice but to manually patch or deploy some helm chart to manage our clusters ... That's suboptimal ...

Thanks @funilrys , that trick worked for me too.