This might read like a VCP exam question!
We have a test environment or 3 hosts on which I had created separate standard switches with respective port groups for production, offnettest, and vmkernel port groups. I.E.:
vSwitch0 = Production port group (port 1) & VMKernel Management
vSwitch1 = OffnetTest port group (port 2)
vswitch2 = vmkernel port groups
I used Veeam 7 to replicate some VMs from production to the test environment hosts, and copied some more VMs from backup to test hosts. In both cases I changed the network port groups to OffnetTest on each VM, and they could all see each other.
A big issue came up today where we determined that VMs in the production port group were getting time from a replica DC on our test host. Very bad! I verified that the replica DC was on the OffnetTest port group, but I did see that when looking at host networking, the DC appeared connected to vSwitch0 as well as vSwitch1. I did a little toobin to find that this was a holdover from a snapshot taken during replication, so removing the snapshot resulted in the VM only being a member of vSwitch0. Red herring. What else could explain this?
We fixed the problem by putting a router between test and production hosts, which our network team is satisfied with (because time is correct across the board now), but I'm not convinced we know how we got there. It seems to me that a reasonable explanation might be that a production hardware switch might have been using an incorrect default port of 2 when trunking over to our test hardware switch, thus forwarding traffic to port 2 on our test environment, BUT that would also mean that we should have been seeing IP conflicts and other problems too, which we did not see. Is there any other explanation you can think of?
Thanks for your thoughts!