Reducing footprint of VCF management services cluster that is only hosting Day 0 component

Products

• VCF Operations

Issue/Introduction

After patching VCF Management Services Runtime from 9.1 to 9.1.1+ on a deployment that does not have Log Management and Real-time Metrics installed (Day-0-only deployment), the worker nodes are not automatically resized to the optimized 9.1.1 machine type. Any one of the following is sufficient to confirm you are hitting this issue:

• Following the 9.1 → 9.1.1+ patch, the worker node virtual machines remain at their pre-patch (larger) size; the expected reduction/rightsizing does not occur.
• VCF Management Services Runtime 9.1.1.0+ patched from 9.1.0.X.
• A Day-0-only deployment: the Day-N add-on components "Log Management" and "Real-time Metrics" are not installed.
• All the Components are Updated to 9.1.1+

Environment

• VCF management services Runtime 9.1.1

Cause

VCF Management Services Runtime 9.1.1+ introduces Day-0-optimized worker machine types and minimum worker replica counts.

For greenfield 9.1.1+ deployments these are applied automatically.

On a brownfield deployment that is patched from 9.1, the worker sizing is not reconciled onto the new values on its own, so the worker nodes stay at the larger size and replica count they were provisioned with before the patch. Realizing the optimization requires a one-time, manual re-application of the correct worker size, machine type, and minimum worker replica count for the deployment's size profile, high-availability (HA) configuration, and whether the deployment is a fleet (manages other clusters) or a standalone instance deployment — fleet deployments require additional worker headroom.

Resolution

Note: This procedure triggers a worker node rollout, which can disrupt ongoing operations on the deployed cluster. New worker nodes are created with the optimized machine type and the old nodes are removed. Perform this during a maintenance window.

The remediation script runs synchronously: it submits the resize and wait for the node rollout to finish.

This remediation requires direct access to a control-plane node of the VCF Services Runtime cluster. Follow the steps here to connect to the control-plane node: 

1. SSH to the control-plane node as vmware-system-user, elevate to root
   • Enable SSH access to the VCF Services Runtime cluster
2. Copy the remediation script to the control-plane node **
3. Run the script
   1. chmod +x <directory>/rightsize-day0-workers.sh
   2. <directory>/rightsize-day0-workers.sh
      
      • The script prints a highly visible warning and then prompts for confirmation. Type yes to proceed. It will:
        
        • Verify that no Day-N components ( "Log Management" and "Real-time Metrics" ) are installed — if either is found, it aborts without making any change.
        • Verify that this is a management cluster — if cluster.type is consumption, it aborts without making any change. This procedure does not apply to consumption clusters.
        • Auto-detect the platform size profile, HA setting, and fleet-vs-instance deployment type (a fleet deployment has a fleet ingress FQDN configured) from the vmsp-platform PackageDeployment.
        • Apply the correct worker size, machine type, and minimum worker replica count for that profile — the minimum replica count differs between fleet and instance deployments.
      • Review the script output: each step prints an OK: line on success or an ERROR: line on failure.
4. Verify the node rollout
   1. The resize is applied synchronously and Monitors progress by checking the PackageDeployment:
      • kubectl get pd vmsp-platform -n vmsp-platform
   2. Wait for its STATUS to reach Successful. Rolling out the new size to every worker node in the cluster takes time (the platform replaces nodes one at a time), so this can take a while to complete, especially on larger clusters. 
5. Remediation script (rightsize-day0-workers.sh)