Migrate VMs from Proxmox --> Hyper-V?

[NewDude]

So my business focus has changed a bit, and now I'm selling backups and DR. Which means I'll be running Veeam, which means I need to switch my cluster to Hyper-V or VMWare in order to support the most useful functions that Veeam offers.

Is there a solid tutorial on migrating to Hyper-V? Is it dead-simple, or is it a "now would be a great time to test Veeam's recovery options and see if that's a good way to migrate?"

I don't even know where to start here, and the new business direction has kept me working 80 hour weeks and I haven't put the research into this that I should have.

Sorry if this is a common question.

 

[spirit]

you can use "qemu-img convert" to convert your vm disks to hyperv vhd format.
(Don't known how to import them in hyperv, sorry)

 

[apoc]

I did it the other way around (from ESXi to Proxmox) but the great thing is: It will work both ways. And it should be completely independent from the Hypervisor.
It is some effort and I cant judge if a backup+restore with Veeam is a better solution.

Hyper-Visor#1 = HV1 = ESXi
Hyper-Visor#2 = HV2 = Proxmox

Here is how I did it.
- Installed a Linux-VM on HV2 and attached the target virtual disk (provided by HV2, becomes part of the new VM) to it.
- Then I have used open-iscsi-target to convert this disk into an iSCSI device.
- The iSCSI-device was presented to HV1
- HV1 discovered the new device and used the iSCSI-device as a passthrough-disk which got attached to the original/source VM
- The source VM now has twice the number of disks - the originals and the ones which come from the iSCSI-target of the new hyper-visor.
- Booted the source-vm with another linux live distro and simply used DD to transfer the whole disk from source to target. The reason why I was using dd: encrypted disks. Most of them were backup-disks (using rsync and link-dest) and everything else from my experience is quite a challenge to keep the link-dest (hardlinks in the FS) alive.
- After DD was completed, the original VM just got shut off on HV1.
- A new VM was created on HV2 using the disks which have been the target of the DD-command.

What originally was a "weird idea" became a proof of concept. It turned out to be the "migration procedure of choice".
The great thing here is: if something goes boom during the migration process, you just can fire up the original VM again.
And you can do VMs step by step, starting with the "less important ones" first.

Maybe that helps.
Cheers
Thomas

 

[NewDude]

Thanks, Thomas.

My first trial run with a Centos 7 machine worked once I selected the recovery kernel. I haven't tried the Ubuntu machine yet - I may need to try your solution there.

It's innovative though. I dig it.

 

[apoc]

Hyper-V is pretty ok for Linux-VMs as of today, as long as you run "recent" versions of the distribution.
The "enlightment tools" are part of the Linux kernel for quite some time now so I wouldnt expect too much problems in general.
Of course issues can happen, but overall it is quite solid.

>>It's innovative though.
Thanks. I found it pretty complex over all. Given the fact (which I haven't mentioned before) that my HV2 was a virtualized Proxmox on ESXi providing nested virtualization didn't make it less confusing .
But it was the best solution for me. And I am very happy with my choice as Proxmox just runs great - and it works with my "proven" (others would say old) hardware.

Good luck with your migration.
Thomas

 

[NewDude]

Thanks. I found it pretty complex over all.

Yeah, but I didn't want to say that because it's cool. Though I'm happy I didn't need to be that involved. Or at least, not yet.

And I am very happy with my choice as Proxmox just runs great - and it works with my "proven" (others would say old) hardware.

Good luck with your migration.
Thomas

Proxmox occupies a very good niche, with great utility and a very reasonable price. I'm glad I learned it. Business needs change (and businesses change direction) though, so it's time for me to move on.

So far things are going well in the sense that everything is working as expected. But the headaches to build that first Hyper-V cluster from scratch, and weird issues like "I bought both these servers identically specc'd at the same time - why am I getting errors about incompatible CPU features?"...

Almost done now.