Customer firstly upgrade VC from 7.x to 8.0u3 P05 which has been successful. Post that Supervisor upgrade was attempted (from 1.27.5 -> 1.28.3), which failed with the following errors :
ImageRegistryUpgrade
error message: "Component ImageRegistryUpgrade did not finish within 360 seconds timeout deadline
EnvPropsUpgrade status
Failed to run command kubectl get vspherezones -o Jason ret=1 out = err=error: the server doesn't have a resource type "vspherezones"
Referring below KB:
https://knowledge.broadcom.com/external/article/393911/supervisor-upgrade-failed-at-envprops-co.html
After applying the KB fix, Supervisor upgrade was re-attempted, but this failed in pre-checks stage due to VKS clusters not running.
After re-triggering upgrade from dcli component ImageRegistryUpgrade completed with upgrade.
Below two components in failed status:
CsiControllerUpgrade
VMOperatorUpgrade
Error in var/log/vmware/upgrade-ctl.log
VMOperator configuration error: Component VMOperatorUpgrade failed VMOperatorUpgrade did not finish within 360 seconds timeout deadline
vSphere Kubernetes Service
This issue can't be prevented before the upgrade has taken place, but only the workaround can be applied upon the issue during the upgrade.There is no safe way to prepare the Supervisors ahead of time for this.
IMPORTANT NOTE: Engineering does NOT recommend doing this step before committing the upgrade. This can be done in-flight of an upgrade, or safely after upgrade has failed. There is no safe way to prepare the Supervisors ahead of time for this.