When attempting to use the CLI to migrate a cluster from NVDS switches to VDS switches, the process keeps moving forward with putting hosts in Maintenance Mode without completing the previous host.
Checking the upgrade settings verifies that the upgrade mode is Serial, not parallel.
NSX 3.2
Vsphere 7.0U3q
The migrations can be completed one host at a time using API calls:
POST https://<mgr-ip>/api/v1/nvds-urt/precheck
(Click Send)GET https://<mgr-ip>/api/v1/nvds-urt/status-summary/<precheck-id-from-previous-step>
(Click Send)GET https:/<mgr-ip>/api/v1/nvds-urt/topology/<precheck-id>
(Click Send)POST https://<mgr-ip>/api/v1/nvds-urt/topology?action=apply
GET https://<mgr-ip>/api/v1/nvds-urt/status-summary/<precheck-id>
(Click Send)POST https://<mgr-ip>/api/v1/transport-nodes/<tn-id>?action=migrate_to_vds
(Click Send)This information is based on the Migration instructions.
The documentation has some "Alternate" calls. Pick ONE and stick with it.
EVERY Time you run the POST from Step 1 or its alternative PUT it generates a NEW Pre-check ID invalidating the previous Pre-Check ID.
For consecutive hosts, you could possibly keep the initial Pre-Check id, do steps 1-3, then skip to 4g, but this has not been tested.