The downstream cluster was initially created using Terraform script, which clearly contained what we needed to make Vmware integration work. The TF includes the values for the CPI and CSI charts configuration.
The cluster is up and running and working as expected. The issue happens when selecting a newer kubernetes version to upgrade the cluster. In Rancher UI > Cluster Management > Select cluster > Edit Config > Cluster Configuration >Basics > Select a kubernetes version to upgrade cluster > Before saving to apply the upgrade, check the CSI and CPI in the Add-On Config tab.
The values will appear as empty for CPI when they should be accurately rendered.
Workaround:
A) Adding/Typing manually the CPI values in the Rancher UI
B) Modified the helm charts with additional manifests to indicated that some values are true
@rak-phillip was able to reproduce this in 2.9-head as of filing:
Reproduction Steps
Provision a new vSphere RKE2 cluster via Rancher Dashboard
Make the following changes under Cluster Configuration => Basics
Change Kubernetes Version to the lowest available version in the list
Change Cloud Provider to vSphere
Under Cluster Configuration => Add-On Config, make the following adjustments
Enter values for vSphere CPI Configuration => vCenter Host, Data Centers, Username, & Password
Enter values for vSphere CSI Configuration => vCenter Host, Data Centers, Username, & Password
Ensure that all other required for provisioning a vSphere cluster are populated and correct; click Create
Wait for the vSphere cluster to enter a ready state
Modify vSphere cluster created in step 1
Under Cluster Configuration => Basics, modify the Kubernetes version to the latest version in the list
Under Cluster Configuration => Add-On Config
All values for vSphere CPI Configuration are missing
All values for vSphere CSI Configuration are preserved
Optionally, if we change the selected Kubernetes version back to the value we selected in Step 1, we can observe that the vSphere CPI Configuration values are populated with their original values.
Internal reference: SURE-8310 Reported in 2.8.3
The downstream cluster was initially created using Terraform script, which clearly contained what we needed to make Vmware integration work. The TF includes the values for the CPI and CSI charts configuration.
The cluster is up and running and working as expected. The issue happens when selecting a newer kubernetes version to upgrade the cluster. In Rancher UI > Cluster Management > Select cluster > Edit Config > Cluster Configuration >Basics > Select a kubernetes version to upgrade cluster > Before saving to apply the upgrade, check the CSI and CPI in the Add-On Config tab.
The values will appear as empty for CPI when they should be accurately rendered.
Workaround: A) Adding/Typing manually the CPI values in the Rancher UI B) Modified the helm charts with additional manifests to indicated that some values are
true
@rak-phillip was able to reproduce this in 2.9-head as of filing:
Reproduction Steps
Optionally, if we change the selected Kubernetes version back to the value we selected in Step 1, we can observe that the vSphere CPI Configuration values are populated with their original values.