I am trying to migrate both the only vCSA and the VMKernel interface in my ESXi host from a standard vSwitch to a vDS, and that gave me all kinds of mess. How to do that properly? Bear in mind that the host in question also have the virtual machine that performs as the router.

My intended setup:

The host:

                     [      ]- vmnic0 -+- LACP --- Switch 1 (managed)
Internet --- vmnic3 -[ Host ]- vmnic1 -+
                     [      ]- vmnic2 --- Switch 2 (unmanaged)


vmnic3 --[ vSwitch0 ]-- vRouter


vmnic0 -+- LAG1 --[     ]-- vRouter
vmnic1 -+         [ vDS ]-- vCSA
         vmnic2 --[     ]-- vmk0 (management network)

vRouter (OS: Ubuntu Linux):

vSwitch0 --[ vRouter ]-- vDS

vCSA (and other VMs):

vDS --[ vCSA ]

Other physical computers are attached on Switch1 or Switch2. There is also another ESXi host.

  • Do you only have a single host? – ewwhite Jan 27 '16 at 6:03
  • @ewwhite There are two hosts but the second host is only on when there is the need for it. – Maxthon Chan Jan 27 '16 at 7:34
  • If you're just experimenting and want to use a vDS, that's fine, but can you explain why you need a distributed switch in this setup? – ewwhite Jan 27 '16 at 7:35
  • 1
    @ewwhite The only way ESXi can support LACP is vDS and that second host, when on, joins the vDS too – Maxthon Chan Jan 27 '16 at 7:37
  • Yes, I understand how distributed switches and LACP work. But I don't see the use case here. – ewwhite Jan 27 '16 at 12:14


I don't think a distributed switch or LACP belong here. And this definitely doesn't sound like an environment with VMware Enterprise licensing...

But use the network migration tool in VMware. This is fully-documented and a supported means of migrating virtual and physical adapters from Standard vSwitches to vDS.

  • That tool did not work for me as I am migrating either the sole VMKernel adapter or the vCSA itself. Such migration creates a network outage. Is there anything to do with the vDS uplink (that is, vDS uplink being incompatible with normal traffic) – Maxthon Chan Jan 27 '16 at 8:03
  • Also, this migration shouldn't involve an outage. But honestly, I'm not sure what you're doing. Can you take a screenshot of your network layout? – ewwhite Jan 27 '16 at 12:18
  • I think he is trying to get vcenter to migrate it's own network, this will trigger an outage and anuto roll back I think. He should probably just create a new network, and directly log into the host using the fat client to move the VM (VCSA) to the new network – Rqomey Jan 27 '16 at 15:31
  • No, it won't cause an outage if done to the documentation guidelines. But this setup seems a little too outside the norm and is not a good candidate for a distributed switch. – ewwhite Jan 27 '16 at 15:33

I don't follow your question very well, but I will make a suggestion all the same.

If VC looses network connection during a network reconfig, the change is automatically reverted (sometimes), even if it is only for a moment.

I would suggest putting your VCSA onto a temporary second network (so you have two routes), this way you can stop using the primary net while you do the moving around.


I think you should migrate in two steps, first to a vDS without LACP and then enable LACP. The first step should be straight-forward: Create the vDS and port groups you need and set vmnic0 as the first uplink and migrate your vmknics and vCenter to the vDS. Then set vmnic1 as the second uplink for your vDS. Then migrate to LACP.

I've stumbled on an interesting blog posting recently: Disabling vSphere vDS Network Rollback. Maybe this can help you, too.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.