TKGm management cluster upgrade to v1.29.15+vmware.1 fails during postconfig

Products

• VMware Telco Cloud Automation

Issue/Introduction

Management cluster upgrade to version v1.29.15+vmware.1 are currently failing during the postconfig stage due to this issue.

Environment

3.4

Cause

When the management cluster is created on TCP 3.0, the clusterType is standard by default. However, with TCA 3.4 mc-spoke will try to apply the TKC dummy CR using the classystandard as default, which will be blocked by TKO. 

This is a known bug with TCA 3.4 and will be fixed in the upcoming TCA release. For existing TCA 3.4 version, please apply the workaround.

Note: This will not happen for clusterType classy.

Resolution

To work around the issue is to delete the dummy TKC CR, then retry the upgrade.

"kubectl delete tkc <mcname> -n <mcname>"

Additional Information

After upgrade succeeds, one might encounter two issues

1. The endpoint status is not shown in the TCA UI.

This is due to the missing endpoint data in the "K8sClusterDetails" table on TCA-Manager DB.

Please open an SR with the support team.

2. After fixing the connected endpoint missing issue, you would see false alarm.

The reported "unhealthy" status for "caip-in-cluster-controller-manager" is a false alarm with no functional impact; if resolution is required, deploy a new classystandard management cluster, rehome the workload clusters, and decommission the legacy management cluster.