Supervisor service upgrade does not report an error and bypasses image availability check under certain Async Supervisor upgrade conditions

Products

• VMware vSphere Kubernetes Service

Issue/Introduction

• When performing a Supervisor Service upgrade in an environment with multiple subscribed content libraries from different Async Supervisor builds, the vSphere Client may allow upgrading to a newer embedded service version without upgrading the Supervisor Cluster first. 
• The UI does not display a compatibility error for some versions, even if the Supervisor is running an older, incompatible version.
• Supervisor Service upgrade may fail as the image required by the new service version is not available.

Environment

VMware Cloud Foundation 9.0

Cause

• The core service catalog in wcpsvc holds the list of Supervisor services and versions that are registered by default.  
• This in-memory catalog is reset whenever the list of Supervisors changes, which can occur when switching or adding subscribed content libraries for Async Supervisor builds.
• When the catalog resets, previously loaded service versions may be lost, resulting in them being treated as manually registered versions, bypassing the image source check.  

Resolution

If the Supervisor Service upgrade failed, follow the below steps to recover,

1. Identify the Supervisor Service version bundled with the running Supervisor (the target version).
2. If the target Supervisor version is higher than the current Supervisor Service version, upgrade the Supervisor Service to the target version.
3. If the target Supervisor version is lower than the current Supervisor Service version, upgrade the Supervisor to the version that bundles the target service version (since that Supervisor includes the required service image).

Note: 

• Avoid adding multiple subscribed content libraries for Async Supervisor builds without upgrading the Supervisor Cluster.
• If multiple builds must be referenced, ensure Supervisor is upgraded after switching content libraries to keep the core service catalog consistent.

Additional Information

• The issue described in this KB article does not affect the first Async Supervisor release.  
• The issue may occur during consecutive Async Supervisor upgrades, where versions from intermediate builds can be lost from the service catalog.