After a host-failure, the host appear as an orphaned registrations in the NSX-T Manager database. Because of this registration your unable to:
- repair the hosts via the NSX-T GUI.
- re-install the NSX-VIB’s via the NSX-T GUI.
- re-join the host to the NSX-T management plane.
You can basically do nothing as long as this registration exist or isn’t updated.
The quickest method to get a workable solution is to:
- Manually reinstall the NSX-T VIB on the host
- Download the VIB file from the VMware download site. (don’t forget to select the correct NSX-T version) and copy it to the /tmp folder on the ESXi host
- Execute the following command “esxcli software vib install -d <nsxvib.zip-file>”, for example: “esxcli software vib install -d /tmp/nsx-lcp-22.214.171.124.0.11294289-esx65.zip“
- Execute a NSXCLI reset command on the host.
- retrieve the NSX-T Manager API certificate thumbprint: in the NSX-T Manager System – Components – NSX Manager – Info
- After the NSX-T VIB(s) has been installed the NSXCLI is enabled on the host: start the NSXCLI.
- Execute the NSXCLI reset command, for example “reset management-plane 192.168.0.3 thumbprint a123457b1234567c1234567d1234567e1234567f1234567g1234567 username admin password NSXT4L!fe“. The allows the host registration in the NSX-T management plane to be overwritten by a new UUID, which is related to the newly installed host.
- Redeploy the corresponding Transport Node.
- Now that the host is reconnected to the NSX-T Fabric, the Transport Node configuration isn’t pushed to hosts again. You need to redeploy the corresponding Transport Node configuration to the host.
- First start with writing down the Transport Node configuration (and optionally unconfigure the auto-install configuration when it’s being used).
- Remove the corresponding Transport Node configuration (this can take a couple of seconds).
- The NSX-T management plane configuration and the actual configuration are in sync right now, we can now proceed with the Transport Node configuration.
- When the auto-install is used, configure Auto-Install configuration on the cluster -or- manually create a Transport Node.
- When the Transport Node is re-created, the NVDS is pushed towards the host.
At this time the host is completely re-installed and functioning again (with NSX-T).
Another option is to remove the host registration in the NSX-T Management Plane, by using the NSX-T API. This is describe in this document. After removing the host registration you can rejoin the host to NSX-T and re-deploy it as a Transport Node (basically the same steps as described above).