This message was deleted.
# rke2
a
This message was deleted.
c
set the cloud provider to external
s
Can we do that with the existing cluster then? Just update the yaml file to have that value?
c
I don’t think so. Cloud provider selection should not be changed once the cluster is up.
the nodes will have been initialized and have the providerid set from the built-in rke2 default cloud provider (which is what if you pick none)
s
So do we need to delete and rebuild the clusters?
c
you can try switching it but I don’t think the cloud provider will be able to take over the existing nodes. They will all have the providerid set from the default one.
You might be able to switch it and then delete and re-add the nodes one by one if you want them owned by the new cloud provider
s
When building a cluster selecting cloud provider as either external or vsphere, there’s a bunch of nodes in pending status, that have reason FailedScheduling, 0/1 nodes are unavailable: 1 node had taint {node.cloudprovider.kubernetes.up/uninitialized:true} that the pod didn’t tolerate
Just to add we are deploying this via the Rancher Management UI
c
Yes, that means that the cloud provider has not come up yet. If you pick external you are on your own to deploy and configure a cloud provider. If you pick vsphere then there should be a spot in the UI to place your vsphere cloud provider configuration, which would include the address and creds for your vsphere server.
s
Cool. Looks like there is a firewall block on our end connecting to vsphere. Will retest once we get that opened
t
Good morning. I ran into this issue aswell with Rancher 2.7.0/1. RKE2 cluster deploys happily without cloud provider, but when selecting vSphere and filling in the blanks, the Rancher agent wont' startup on the downstream cluster nodes because of the wrong taint, like describe here. Values from the vSphere CSI fields in Rancher don't seem to get passed to the helm chart that executes them, so it's a +1 for us.
@creamy-pencil-82913: I saw you also participated in this thread. Also setting the values in the additional config under the CSI config didn't do the trick for me. I'm pretty new to this, and adressing issues on Github. Could you tell me what I could do to contribute to a solution and to get this issue on a roadmap to be fixed?
We've chosen RKE1 because this issue was a showstopper for us.