There are few products in our environment, including vSphere 6.7, vCloud Director 9.7 and NSX-V 6.4.x. We are planning to upgrade Virtual Distributed Switches to ver. 6.6.0, and this is not a big deal at the first sight. However, there are thousands of "up" ports and a lot of portgroups as long as logical switches (vxlan-backed PGs) on our Switches, so I want to be sure to do it the right way.
Actually, Im concerned about a restore process. Well, if something will go wrong, I'll try to use "Rectify" opertaion. If it won't help, I'm going to restore vDS. It says in the documentation that in case of a failure during the vDS upgrade, I should import the previously exported vDS config with preserving the morefs (sounds logical). However, this procedure will recreate the vDS, so I'm not sure if all the VMs (thousands of them) will be connected to the correct PGs. Should I really use the "import" instead of "resotre" in case of a bad scenario?
Ofcourse, I will test it in a lab first (not only the upgrade, but also a vDS import/restore), but I don't want to ruin a lab without asking a question here.
If the switch is already there, then you can restore its configuration. Restore will overwrite all of the settings and port groups. Other port groups are not part of the backup will be retained.
If the switch is not exists yet, you can create it by choosing import configuration. If a switch will be re-created, and preserve original identifiers was enabled during the import, vms will be connected.