Issues with changing dvUplink names on distributed vSwitch...

[jsutherland1979]

We had our four ESX hosts running on a single distributed switch with vCenter itself running as a VM on one of these hosts. The distributed switch had connected four dvUplink ports to about 10 vlans. We thought it would be a good idea to change the dvUplink port names to something more descriptive and at the same time add two extra dvUplink ports to our existing four. Upon clicking OK, vCenter lost connectivity to all hosts in the cluster and we realised all VMs on all 10 vlans were offline. When we logged into each host via the command line we could see the changes had taken effect i.e. the names of the four dvUplink ports had changed. Basically we need to know two things:

- What caused this loss of connectivity and how could it have been prevented?

- Is there any way to remove a host from a distributed vSwitch from the command line when all physical ports are bound to the distributed switch?

We are very tempted to abandon the idea of running a distributed switch altogether as it seems to represent a large single point of failure. Does anyone have any thoughts on this?

Any help would be much appreciated as it caused quite a headache!

Kindest regards,

Jarrod

[Flammi]

I had the exact same problem.

Although I wasn't sure whether it was because I added two dvUplink ports or because I changed the names of the dvUplink ports. (I did both at the same time)

Now that I read your post, I tent to believe that this is a general issue with the renaming of dvUplink ports. It was impossible to connect to either one of the ESXi hosts. The entire vNetwork was down.

I ended up reinstalling the ESXi host because the environment had to be up and running again as quickly as possible. I'm not familiar with the console and didn't know what else to do. (One of the hosts is still down)

This also made me think about the potential "single point of failure" of the distributed switch. I'm not sure whether I will use a distributed switch again, once I got my vCenter up and running again.

Did you open a call at VMWare? Maybe this is a known bug?

[bartojc]

I know this is an old thread, but we had the exact issue yesterday with 5.1 and brought down over 500 VMs.  VMware support centered around recovering vCenter which we finally did, but overall we wasted some 7 hours.  Support said a simple rename "should have worked" not much comfort at 3:00 am.

-Jeff