After SSPI upgrade, change data center name and re-connect vCenter, new nodes could not get any IPs

Products

• VMware vDefend Firewall
• VMware vDefend Firewall with Advanced Threat Prevention

Issue/Introduction

After SSPI (Security Service Platform Installer) upgrade, after changing the data center name and re-connecting vCenter, new nodes could not get any IPs. "antrea-agent" pods in the "kube-system" on workload cluster are crashing.

The configmap "cloud-config" in "kube-system" on the workload cluster is using datacenter name instead of datacenter moref id reference

Environment

Security Service Platform Installer 5.1

Cause

During the Security Service Platform Installer upgrade, the configmap "cloud-config" will be re-generated using cpi-manifest based on the clusterctl-init.yaml. We are using data center name in the clusterctl-init.yaml and overwrite the cloud-config with datacenter moref id reference during deployment. So re-generated configmap "cloud-config" will still use datacenter name instead of moref id reference. 

In this case, after we change the data center name, antrea-agent could not find the datacenter with the old name stored in the configmap "cloud-config". As a result, it could not assign IP to the new nodes. 

Resolution

Update configmap "cloud-config" in the namesapace "kube-system" on the workload cluster with datacenter moref id reference.

The datacenter moref id reference format looks like: Datacenter:<dc-moref-id>.

For example: