Hello Team,
I have 3 intel NUC servers managed by vcenter 6.7. Migrating to vcenter 6.7 now, i was able to remove all VMs from first server and just add to a new Vcenter 6.7 + install updates + move to distributed switch.
But now do have a problem: i have two servers with 1 uplink with many VMs running (including old+new vcenter) on the local storage. Those are in distributed switch on old Vcenter. Before i move it to new Vcenter i need to migrate it to a standard switch.
But it looks like there is no option to migrate to standard switch with a single uplink.
I could do it while migrating standard->distributed (migration of vmk0 + uplink at the same time), but opposite migration is not possible ?
What can i do now ?
Is the only option to shut down all VMs, boot ESX from console, reset network settings ? (this moves ESX into standard switch ?). Then boot both EXS, start both Vcenters and continue migration ?
(really painfull process for me)
Any ideas ?
Thanks,
Michal
Because a vDS is an object that requires vCenter to be available to manage port memberships and connectivity. A vSS is not. So going from a host-local object like vSS to vDS requires no interruptions to vCenter. When you're yanking away the NIC vCenter uses to communicate with its hosts you can't very well have that.
And now you've discovered one of the downfalls of the Intel NUC platform (and other small form-factor PCs). There's not a way to do an online migration from vDS to vSS with only a single uplink. What you may want to do is move your vCSA over to a vSS on a vacant host first, then use vCenter to shutdown and migrate them to hosts with a vSS already configured. Either way, you're not doing this online, but you don't need to reset ESXi's networking settings from scratch.
Thanks daphnissovfor the confirmation !
It's not a downfall of Intel NUC, we have 2018, i am installing different hypervisors on my phone or clock.
It's a downfall of vmware to not provide such basic functionality. Especially that opposite migration from standard to distributed with one uplink is working fine
Any reason for that inconsistency ?
Thanks,
Michal
ESXi is a server-class baremetal OS meant for server hardware. Server hardware means multiple NICs. When you choose to install ESXi on unsupported hardware like this, you forfeit certain abilities and uses this just being one you have to live with.
I agree with that statement, but it does not answer my question. Why we have inconsistency ?
(allowing seamless migration from standard->distributed while not the opposite) for 1 uplink/unsupported devices ?
Thanks,
Because a vDS is an object that requires vCenter to be available to manage port memberships and connectivity. A vSS is not. So going from a host-local object like vSS to vDS requires no interruptions to vCenter. When you're yanking away the NIC vCenter uses to communicate with its hosts you can't very well have that.