Preparing a standlone host cluster for NSX-T 3.2.2 results in failure.

Problem

In this scenario, there are two NSX-T deployments connected to same vCenter Server in NSX-T 3.1.x. The vCenter Server is registered with NSX-T-1 as compute manager and NSX-T-2 is using standalone cluster registered to the same vCenter Server.

You upgrade NSX-T-1 to v3.2.2, then enable multi-NSX flag on the vCenter Server. However, you do not upgrade the NSX-T-2 which remains on 3.1.x. Since NSX-T-1 is enabled with multi-NSX, the NSX Manager UI lists the standalone cluster of NSX-T-2 as an unprepared cluster.

If you try to prepare the standalone NSX-T-2 cluster from NSX-T-1, the standalone host goes into failure state.

Cause

NSX-T-2 hosts go into failure state because it is running NSX-T v3.1.x.

Solution

  1. Log in to https://<vCenter-Server-IP>.
  2. Migrate all VMs from that host to the other hosts using vCenter Server.
  3. Remove NSX from both managers using below API:
    DELETE https://<NSX Manager-IP>/api/v1/transport-nodes/<transportnode-id>?force=true&unprepare_host=false
  4. Configure NSX-T on the desired NSX Manager.