This issue is addressed in a code fix in TCA 2.2
Workaround:
There are two workaround options for recovery after isolating the unresponsive ESXi servers in the vmconfig-operator log:1. Reconnect the unresponsive host, then delete the vmconfig-operator pod on tkg-mgmt. The vmconfig-operator will be rescheduled and started, then the nodepool deployment can proceed2. If the ESXi host connectivity issue cannot be resolved: The host validation can be prevented by replacing the host CR profile entry with an empty string.
a. Save the existing host CR to backup:
kubectl get esxinfo -n tca-system <hostname-fqdn> -o yaml > /tmp/<hostname-fqdn>.backup
b. Open the host CR with the following command, identify the hostprofile name line and change it to empty quotes ("")
kubectl edit esxinfo -n tca-system <hostname-fqdn> -o yaml
c. Restart the vmconfig-operator pod with the kubectl delete pod –n <namespace> <pod> command