Management cluster deployment stuck during cluster creation

Products

• VMware Telco Cloud Automation
• VMware Telco Cloud Platform

Issue/Introduction

When attempting to deploy a new Management cluster from the CaaS Infrastructure page:

• The deployment remains in the Cluster Creation phase indefinitely.
• Control Plane (CP) nodes initialize, but worker NodePools fail to provision.
• Checking journalctl -u kubelet logs on the nodes prompts the below error message.
• Error Message: failed to get CsiNodeTopology for the node: "localhost". Error: failed to get API group resources: unable to retrieve the complete list of server APIs: cns.vmware.com/v1alpha1: the server could not find requested resource

Environment

TCA 3.4

TCP 5.0

Cause

The issue is caused by the use of a vSphere VM Template where cloud-init was executed prematurely.

If the base VM used to create the template was powered on at any point during its preparation, cloud-init runs and records its completion status and instance metadata to the local disk. When TCA clones this template to provision new nodes, cloud-init detects the existing metadata and skips execution.

Consequently, the unique hostname assigned by the Cluster API is never applied, and the VM retains the fallback hostname "localhost". The vSphere CSI/CPI drivers cannot correctly map a node named "localhost" to the vCenter topology, preventing the storage layer from initializing.

Resolution

To resolve this, you must ensure that cloud-init runs for the first time only when the cluster node is actually deployed.

1. Upload the base OVA/image directly into vCenter.
2. Convert to Template or mark as a VM Template.
3. Make sure that you Do not power on the VM.
4. Redeploy the cluster using this new template.