Would it be correct to follow steps 2-4 in reverse as per the following blog;
Step 6 - Now, we just simply perform the reverse set of steps from 2-4, by going from VSS to VDS and re-creating our LACP bundle on the new vCenter Server."
https://www.virtuallyghetto.com/2017/11/moving-esxi-hosts-with-lacplag-between-vcenter-servers.html
Yes. You break the LACP team in order to remove one vmnic from the vSS and to add it to the vDS. When you migrated one vmnic to the vDS you have connectivity and you can migrate your VMs to the vDS. When the vSS is empty (no VMs and no vmkernel adapters) you can remove the second vmnic from the vSS and add it to the vDS. Then you can reconfigure the LACP team.
The response I got from the Networks guys was "port-channel with mode on for core switch".
I think those terms all mean the same. Port-Channel, LACP are all teaming related.
Agreed. With LACP enabled, and after I migrate one of the vmnic's, I done a test migrate of vmk0 (management) and it disconnected. I also tested some VM networking, also disconnecting.
Is there any official VMware documentation on this?
Found a nice article on this;
Migrating to a VDS switch when using etherchannels | Yellow Bricks
I'm not sure if there is official documentation. But I did a similar migration (from vDS to vSS) and I can confirm that you will lose connectivity when you move one vmnic from one switch to another without preparation. I did a similar migration (from vDS to vSS) and followed these steps:
I think these steps can also be used when you want to migrate a vSS to vDS.
Oh, we had to change the vSwitch load balancing policy from originating virtual port to IP hash. But that is related to the physical switch configuration.
jburen
Duncan's post states to remove the etherchannel configuration completely in a single step
I have seen it done your way. Is there a definitive answer on which way is correct?
We had to preserve the port-channel because of a specific config in the physical switch environment. Otherwise, you probably could delete the port-channel. I don't think there is only one correct answer. It depends on your environment.
For some reason, I don't see your comment here but I will try to answer your questions...
In my steps, we moved vmnic0 first. This would mean performing a shut on the vmnic0 port forcing all traffic to vmnic1. Because of the physical switch requirements, we had to move vmnic0 to a temporary port-channel. While it was shut down we removed vmnic0 from the old switch and added it to the new switch. In our case that was from vDS to vSS. After it was added to the new switch we did a no-shut on vmnic0.
At this point, we had a vDS with vmnic1 in the original port-channel and a vSS with vmnic0 in a temporary port-channel. So at this point, we migrated the vmkernel port(s) and VMs.
After migrating all vmkernel ports and VMs the old switch (vDS) is now empty. So the next step is to remove vmnic1 from the old switch and add it to the new switch. Now you have the new switch with vmnic0 and vmnic1 but they are both in separate port-channels... So now we moved vmnic0 from the temporary port-channel back to the original port-channel. In our situation, we didn't have to shut vmnic1 because there was no traffic flowing through it.
I hope it makes sense now.
jburen
Originally I didn't understand step 8 - "Move vmnic0 to the previous port-channel (on the physical switch)"
But after reading it a few times, it made sense.
One other question, if we decided to remove EtherChannel (port-channel) altogether, is the process just to simply remove the EtherChannel (port-channel)?
In place, trunk ports on the physical switch ports and using a better LB policy like 'Route based on physical NIC load' with NIOC on the vDS.
jburen
Oh, we had to change the vSwitch load balancing policy from originating virtual port to IP hash. But that is related to the physical switch configuration.
At what stage did you switch the Load balancing policy on the vSS? I am assuming before you added vmnic0 to it?
That's right. We did that before we added the first vmnic.