Skip to main content
Solved

Move backup when using FastCLONE, REFS and reflink


Hello to all,
I have backup chains stored on XFS volumes with reflink that were created with FastCLONE.
How can I move these strings to another volume while preserving this? 
Thanks

Best answer by JMeixner

Ok, a repo on a VMDK can be copied on volume level. Then all ReFS links are preserved.

My repos are all on physical disks. I will try this with a test VBR server in the next time...

View original
Did this topic help you find an answer to your question?

18 comments

Mildur
Forum|alt.badge.img+12
  • Influencer
  • 1060 comments
  • September 20, 2021

You need a block cloning utility.

If you move the files to the new volumes, the storage savings will be lost. All files are dehydrated.


JMeixner
Forum|alt.badge.img+16
  • On the path to Greatness
  • 2686 comments
  • September 20, 2021

Hi a@Erwan ,

as far as I know you cannot preserve this when moving the backups.


haslund
Forum|alt.badge.img+14
  • Mr. VMCE
  • 396 comments
  • September 20, 2021

If you clone the volume on a block-level using for example “dd” in Linux in my previous tests this worked fine. Obviously this could take some time and no backups or restores should be happening during the process.


Ralf
Forum|alt.badge.img+1
  • Comes here often
  • 80 comments
  • September 20, 2021

If source and destination is identical in size you could try xfs_copy


vNote42
Forum|alt.badge.img+12
  • On the path to Greatness
  • 1246 comments
  • September 20, 2021
Ralf wrote:

If source and destination is identical in size you could try xfs_copy

I would be interested if this works! 


Chris.Childerhose
Forum|alt.badge.img+21
  • Veeam Legend, Veeam Vanguard
  • 9299 comments
  • September 20, 2021
vNote42 wrote:
Ralf wrote:

If source and destination is identical in size you could try xfs_copy

I would be interested if this works! 

Me too


JMeixner
Forum|alt.badge.img+16
  • On the path to Greatness
  • 2686 comments
  • September 20, 2021

Ok, this are all methods for XFS.

Is there a possible procedure for ReFS? I am not aware of any...


Mildur
Forum|alt.badge.img+12
  • Influencer
  • 1060 comments
  • September 20, 2021

@JMeixner 

 

The same is valid. Use a block cloning utility.

Or the simple way.

Create a backup with Veeam of the volume and restore it to another disk 😅


JMeixner
Forum|alt.badge.img+16
  • On the path to Greatness
  • 2686 comments
  • September 20, 2021

Yes, but which block cloning utility would you use for this?

And…. after a backup and restore with Veeam all files are inflated again, aren't they?


Mildur
Forum|alt.badge.img+12
  • Influencer
  • 1060 comments
  • September 20, 2021

You need todo a volume level restore. Not file.
 

You have to map backup jobs of course, after the cloning and do a compact operation.
Then, Fastclone will be used after that. Other option would be an active full. But this would need again all of the space of a entire full :)

 

There are multiple cloning utilitys out there. 
I have already used acronis to clone disks 5-10 years ago. Not for this specific use case, but for my home computers.


randyweis
Forum|alt.badge.img+4
  • Experienced User
  • 65 comments
  • September 20, 2021

The pointer based synthetic full will be reconstituted as a full backup unless you can clone the file system block for block.

Has anyone tried the BackupCopyMirrorAll registry key? This supposedly will work to copy old backup chains from one repo to another.

 


JMeixner
Forum|alt.badge.img+16
  • On the path to Greatness
  • 2686 comments
  • September 20, 2021
Mildur wrote:

You need todo a volume level restore. Not file.
 

You have to map backup jobs of course, after the cloning and do a compact operation.
Then, Fastclone will be used after that. Other option would be an active full. But this would need again all of the space of a entire full :)

 

There are multiple cloning utilitys out there. 
I have already used acronis to clone disks 5-10 years ago. Not for this specific use case, but for my home computers.


Thank you @Mildur 😎


Mildur
Forum|alt.badge.img+12
  • Influencer
  • 1060 comments
  • September 20, 2021
JMeixner wrote:
Mildur wrote:

You need todo a volume level restore. Not file.
 

You have to map backup jobs of course, after the cloning and do a compact operation.
Then, Fastclone will be used after that. Other option would be an active full. But this would need again all of the space of a entire full :)

 

There are multiple cloning utilitys out there. 
I have already used acronis to clone disks 5-10 years ago. Not for this specific use case, but for my home computers.


Thank you @Mildur 😎

I think, i will test this procedure with my test server this week.

Make a copy of the backup repo vmdk and try to reuse the old backup chain. I think it should work :)


JMeixner
Forum|alt.badge.img+16
  • On the path to Greatness
  • 2686 comments
  • Answer
  • September 20, 2021

Ok, a repo on a VMDK can be copied on volume level. Then all ReFS links are preserved.

My repos are all on physical disks. I will try this with a test VBR server in the next time...


  • Author
  • New Here
  • 3 comments
  • October 5, 2021

Thanks to all


vNote42
Forum|alt.badge.img+12
  • On the path to Greatness
  • 1246 comments
  • October 5, 2021
randyweis wrote:

The pointer based synthetic full will be reconstituted as a full backup unless you can clone the file system block for block.

Has anyone tried the BackupCopyMirrorAll registry key? This supposedly will work to copy old backup chains from one repo to another.

 

But I do not see how this could help here?


ger.itpro
Forum|alt.badge.img
  • Comes here often
  • 27 comments
  • October 5, 2021

not useful for XFS but if you use a windows server with local disks or sas/san disks (whatever) and ReFS as a repository and have a datacenter licence you can use storage replica with asynchronous replication.

https://docs.microsoft.com/en-us/windows-server/storage/storage-replica/storage-replica-overview

I hope I will find the time to test it over wan and also “locally” from one volume to another.. we ordered a HPE 4510 and why not testing it with a brand new empty server and “some” space..


Bo.
Forum|alt.badge.img
  • Comes here often
  • 14 comments
  • January 6, 2023

Solved in v12, as You probably know already ;-)


Comment