Two days ago, we upgraded NSX-T from 3.1.3 to 184.108.40.206.
We completed the upgrade successfully. No errors.
Yet, a short time afterward, Edge nodes began throwing errors and BGP began failing. Upon investigation, the uplinks assigned to the Edge nodes were down. We attempted to reconfigure them and watched as NSX-T either ignored our configuration or altered it.
On occasion the actual VMs network mapping will be messed up (disable interfaces, move to NIC 4, etc.)
We are working with VMware support, but they are stumped.
Anyone seen this behavior where the NSX-T managers keep modifying and screwing up the configuration?
What do you mean by everything dies; please clarify further
What was your previous version; was it 3.1.2.x ?
Check if upgrade bundle is present on the node. ‘ls /image’ will show upgrade bundle if it is present. Check "get version" CLI to see if MP is in from-version or to-version.
I had a similar issue in the lab where I had host and edge TEPs on the same VLAN (VLAN 0) - after the upgrade tunnels were down, BGP down (idle, administrative down). Had to move those to separate VLANs to get the routing working again.