REFS volume after reinstallation?

[enoch]

Hi all,

Currently have a Veeam server which makes Backup Copy to another server. The server that is Backup Copy target is running Windows 2016 and all drives (except the system drive C:) is formatted with REFS 64K.

I'll like to reinstall that server and add the repositories (volumes with REFS) to a new Veeam server.

Question 1) Is everything about the REFS Block Cloning on the volume after the server is reinstalled?

Question 2) Can the 3 volumes be added again to new Veeam server (SOBR) and after rescan everything is like before?

[foggy]

Hi Peter, haven't done this personally, but apparently you should be ok with re-attaching the volumes to another server, provided it has the same or newer OS version. As for the second question, yes, you can add volumes to the new Veeam B&R installation, create corresponding repositories, and rescan them.

[enoch]

Thanks for reply, can anyone at Veeam comment on this? (better safe then sorry)

[sconley]

We may end up with some spare hardware and I am thinking of moving our virtual Veeam repository to a physical server to save resources in our virtual (VMware) infrastructure since it is a fairly resource heavy system. Our repository is currently on a separate array presented via iSCSI to the repository VM. If I simply present this to the new repository machine I shouldn't lose any of the fast clone functionality, correct? My understanding that everything required for this is contained in metadata within the FS itself. This is all assuming that the new physical server has the same visibility into the iSCSI storage hosting our datastores as our current VM since I do use direct storage transport mode.

[Rick.Vanover]

I believe that is the case, it is all self-contained as properties of the ReFS file system in particular in the "System Volume Information" part of the drive. The one thing that may be something to consider is the Windows Server that "owns" the ReFS storage. So that may be something to consider, especially if you want all of the backup jobs to be transparent to the move.

I do however recommend you do a quick test as follows:

-Make a test Windows Server 2016 VM
-Give it a VMDK
-Configure that VMDK (except size in GB/TB) the same as the one you are considering
-Add it to B&R
-Make it a repository
-Set a backup job or two
-Make a new test Windows Server 2016 VM
-"Steal" that VMDK to the new one (similar to what you plan to do)
-ReRun jobs

If it cannot be transparent... This may be a good time to incorporate the Scale-Out Backup Repository to your setup - so it will be transparent next time

[sconley]

I will give this test a try when I have some time and see if it works. I don't think a SOBR would necessarily help here as I am not adding additional storage, simply looking to migrate existing storage to a new server. Also, from my understanding, I could add my existing standard repository to a new SOBR to accomplish what you suggest. However, with a SOBR since the files were not created on the new extent wouldn't you lose all of the fast clone capabilities unless you run an active full backup or possibly a compact? I figured I would likely have to re-create backup jobs and map them to the migrated files. I didn't necessarily expect it to "just work" (although it would be nice if it did). My biggest concern is losing any fast clone functionality specific to REFS.

[andre.simard]

Hi,

We have a virtual proxy server (proxy1) with a REFS Disk on a vmdk. Can we disconnect the vmdk from proxy1 and attach it to Proxy2 and continue using the REFS fast clone or an active full will be required?

Thank you

[veremin]

Backup proxy server does not operate with REFS disks anyhow, are you probably talking about backup repository server and its advanced REFS support? Thanks!

[andre.simard]

Yes exactly backup repository.
Sorry for the confusion.

[andre.simard]

To make thing a little bit more clear here is more detail:

We have 2 virtual proxy (proxy1 and proxy2)
Proxy1 is also backup repository with 4 REFS repository each of them on a separate Disk(VMDK)

Proxy2 is a new proxy that we just deployer because we need more Concurent task and proxy1 is already at the 8vCPU (Maximum recommanded in the best practice for virtual proxy)

So the question is can we move 2 REFS repository from Proxy1(that is also backup repository server) to proxy2 by detaching the VMDK and attache it to the new server?

The reason why we have attached the repository direct to the proxy server is to avoid sending the data over the Network.

Thank you

[foggy]

Should be possible, but I recommend to run some tests first.