While working on a TKG 1.2.0 on vSphere deployment, we faced this issue issue due to which the TKG management cluster deployment failed. In this blog post, I am discussing the details of the issue and how we resolved it
- Deploying a Tanzu Kubernetes Grid (TKG)1.2 management cluster on vSphere fails at stage 5 of 8
- The management cluster creation does not initiate and no virtual machines are created
- You see messages noting connectivity issue with vSphere in the capv-controller-manager pod logs, similar to the following:
E0107 09:29:44.221659 1 controller.go:248] controller-runtime/controller "msg"="Reconciler error" "error"="failed to create vSphere session: error setting up new vSphere SOAP client: ServerFaultCode: Cannot complete login due to an incorrect user name or password." "controller"="vspherevm" "name"="tkgmgmt-tanzu-control-plane-g5nj6" "namespace"="tkg-system"
- The password for the vCenter user is complex and contains special characters or the username is specifice in the “domain\username” format.
- You can log in to the vCenter Server using the same username and password supplied during the TKG installation.
The issue was caused due to the complexity of the vSphere user account password used and the format of the account.
This is a known issue and currently there is no resolution. To work around the issue,
- Use a simple password for the service account used to connect to vSphere during TKG deployment.
- Use the “username@domain” format for specifying the vCenter Server user, instead of “domain\username”