This message was deleted.

For my own CSI, it doesn't have "Generate CSI vSphere Config Secret", instead pointing to the vsphere-config-secret file. Was that not in place in yourse? It sounded like you were upgrading, rather than building a new cluster, so I'd have expected the secret to be in place already. Is that incorrect?

yeah, its on upgrading and the secret already exists indeed

Probably better to refer to it rather than creating another.

true, when selecting the existing secret it takes vcenter server user password from that. Two things though. 1. All the options like "Enable CSI Full sync" are unchecked 2. On the CPI page no secret can be set and password vcenter and user are also empty there. oh wait, its still early. There is also a vsphere-cpi-creds secret

If I look into the cluster config Add-ons, all the extra options do seem to still be in place. Without attempting to upgrade, do you get the same? The Add-on: vSphere CPI section of the configuration seems to pull the vCenterHost/Data Centers values from the Credential's Secret Name (rancher-vsphere-cpi-credentials). Do you not see that?

Every time I go into that section of any cluster all the checkboxes are unchecked while they were checked when creating the cluster. (except windows compatibility). Dont know if its a bug or not

Do you have a Credential's Secret Name for CPI?

yes

In your vsphere-config-secret, are there entries for VirtualCenter, datacenters and suchlike? csi-vsphere.conf [Global] cluster-id = "c-m-*" user = "admin@vsphere*" password = "NotHere" port = "443" insecure-flag = "1" [VirtualCenter "FQDN"] datacenters = "Whereabouts" Is csi-vsphere.conf present on your Control Plane servers (/var/lib/kubelet/pods/*/volumes/kubernetes.io~secret/vsphere-config-volume/csi-vsphere.conf)? In secrets, I have entries for chart-values-rancher-vsphere-cpi and chart-values-rancher-vsphere-csi (no, I didn't put them there deliberately, they were a by-product of the clusters being built). Are they in place for you? As you might be able to tell, I'm good at random statement and straw-clutching. Actually knowing what I should be talking about is a little beyond me. Sorry.

did you update the kubernetes version? we used to always reset the add-on values after changing the kubernetes version, but it got smarter in 2.13 to preserve those values if possible. Here's the related github issue: https://github.com/rancher/dashboard/issues/5837

the add-on values are constantly resetting now. Also wehn creating a new cluster. After entering the addon values and add a machine pool to the cluster and the addon values are empty again. Im scared to touch anything now, a test cluster is already broken