Will there be a way to convert SMB backup respositories to NFS in v10?

[Novox]

In Gostev's weekend forum post he indicated that Veeam does "not recommend using non-continuously available (CA) SMB shares as backup targets for reliability reasons (no write-through support)." I was also surprised to hear this.

I have multiple Synology NAS's to which Veeam backs up via SMB. However, they also support NFS. Gostev seemed to indicate that the NFS protocol was much better due to "its write-through support for ANY file share" and some NAS manufacturers even indicate NFS performs better than SMB.

With v10's Native NFS support coming up, will there be a way to re-target existing SMB-destination jobs to NFS shares (backed by the same physical storage as the SMB shares) without Veeam having to perform full backups for everything?

Also, without NFS native support, I don't suppose there is a way to transition from SMB to NFS on a Synology NAS now, is there?

Finally, will native NFS support appear under "Storage Infrastructure" or "Backup Infrastructure" / "Backup Repositories"?

Thank you!

[veremin]

If you somehow make current backup files reachable via NFS (copy to separate location, re-configure share, copy files backups, for instance), then, it will be just a matter of mapping backup jobs to already present backup files. No new full backups will be required in this case.

Answering your second question, correct, you will be able to add NFS repository, using the given node.

Thanks!

[Novox]

Well, on my Synology NAS, I can simply add NFS shares to the current SMB shares, but since it's not DELL/EMC, HP, IBM... it sounds I will need to wait for v10 for the native NFS support?

At that time, should I research "mapping backup jobs" or are you more generally saying that if I edit each job's destination to be the new NFS shares, and if the Veeam backup files are already present, Veeam will be smart enough to continue the backup chain?

Thank you.

[veremin]

it sounds I will need to wait for v10 for the native NFS support?

Correct.

or are you more generally saying that if I edit each job's destination to be the new NFS shares, and if the Veeam backup files are already present, Veeam will be smart enough to continue the backup chain?

Correct.

[Novox]

Perfect, thank you!

[babzik25]

Same questions, but what if a CIFS repository is already an extent in scale-out repository?

[Gostev]

We will test and document the process.

In general, you will note we won't have a major push for NFS repositories in v10, because as with any new functionality, we expect it will need time to go through its teething issues and stabilize. However, for v11 we're already implementing an actual warning when you register an SMB-based backup repository that points to a non-CA share (we will query this share property). And we will also be implementing the corresponding alarm in Veeam Intelligent Diagnostic feature of Veeam ONE.

[babzik25]

I don't want to wait for v11 to start using NFS. We use Synology and QNAP for storage. So I can point both protocols to same location easy. It would be nice to have an easy process to make this change instead of creating new repository and moving files.

[Novox]

I believe I know how I would retarget existing jobs that are destined to SMB backup repositories to new, native NFS, repositories. Setup the NFS mount to point to the SAME folder/files at the SMB share, then modify the backup jobs to retarget to the NFS repository.

However, for jobs like those from Veeam Agents, since I cannot retarget them without visiting every computer/agent is there a way to replace the existing backup repositories?

i.e. can I delete an SMB repository and create an NFS repository named the exact same and expect it work? Or is there another way to replace the SMB repository vs. retarget it?

Thank you.

[Novox]

Also, I just retargeted other jobs to NFS instead of SMB (same backing mount point/physical storage/files), and now the disk copies are showing up under "Disk (Imported)."

How do I fix this so they only show up under "Disk (Copy)" like my original SMB files?

[Novox]

@gostev, I just tried creating a new NFS repository and then retargeting two copy jobs to it. Now, the files appear under "Disk (Imported)" and it appears to be backing everything up all over again.

I was under the impression from @veremin that Veeam would see the existing backup files after the re-targeting and would just keep building on the chain, but that doesn't seem to be the case? It's treating it like an all new repository and appears to be doing a brand-new active-full...

[Novox]

I also am not willing to wait until v.11 since we recently had a big discussion about how using non-CA NAS SMB shares was a major no-no, and that v.10 would include native NFC support to fix the issue...

[Novox]

More findings: after creating the NFS repository (at the same mount point as the SMB repository, basically SMB and NFS protocols both pointing to the same files), Veeam creates all new Job folders but appends "_1" and starts the backups all over

[Steve-nIP]

You need to rescan to import the backup, then map the backup. I haven't tried this scenario specifically, but what you're describing is very much what would happen if the job wasn't mapped properly.

[Dima P.]

Yes.

In order to move/map VM backups stored on a CIFS repository to a NFS repository (non Scale-Out repository scenario):

0. Disable backup jobs pointed to the CIFS repository.
1. Enable NFS support for the share published as CIFS (this should be possible in your NAS settings).
2. Remove the CIFS repository from configuration completely to avoid any issues with backup files being rescanned over NFS and CIFS. This will instruct Veeam B&R to "forget" existing backups (Remove from Configuration option does not delete actual backup files from disk). Keep in mind that you might need to adjusts other backup jobs with CIFS repository as a target, otherwise you wont be able to remove the repository from Veeam B&R console.
3. Register the NFS repository in Veeam B&R server.
4. Rescan the NFS repository. Backup files will appear under Home > Backup > Imported node.
5. On the Storage step of the Backup Job wizard change repository to the newly added NFS repository.
6. On the same step, Map backup files to the job by clicking Map backup link, and selecting the desired backups.
7. Click through the rest of the wizard to save new job settings.
8. Enable backup jobs.

[UPDATED] Made changes based on feedback from QA.

[Gostev]

Yup, that's the theory, and how it should normally be done when migrating backups between repositories of the same type.
Per my earlier comment, our QC has the task in the pipeline to verify there's no NFS-specific quirks.

[Dima P.]

In order to move/map VM backups stored on a CIFS extent to a NFS extent (Scale-Out repository scenario):

0. Disable backup jobs pointed to a Scale-Out repository
1. Enable NFS support for the share published as CIFS (this should be possible in your NAS settings).
2. Remove the CIFS extent from the Scale-Out repository. Ignore the evacuate notification – click Yes when prompted ‘Remove the extent without evacuating its backups’.
3. Register the NFS repository in Veeam B&R server.
4. Add the NFS repository as the Scale-Out repository extent.
5. Rescan the Scale-Out backup repository.
6. Remove the CIFS repository from configuration completely to avoid any issues with backup files being rescanned over NFS and CIFS. This will instruct Veeam B&R to "forget" existing backups registered under the CIFS repository.
7. Enable backup jobs.

[Novox]

This is great! The step I was missing was:

"Map backup files to the job by clicking Map backup link, and selecting the desired backups."

Thank you!

[Novox]

I almost forgot, the second part of my question.

For jobs like those from Veeam Agents, since I cannot retarget them without visiting every computer/agent is there a way to replace the existing backup repositories?

[babzik25]

One note to steps above, you can't remove extent if it is the only extent you have. So you need to add NFS repository first, then remove CIFS and add NFS in same step.

[Dima P.]

For jobs like those from Veeam Agents, since I cannot retarget them without visiting every computer/agent is there a way to replace the existing backup repositories?

For standalone agents you can move the backups to the NFS repository (the entire folder with the backup from the repository root folder) and then change the repository to NFS. Automapping should work (Veeam agent automatically imports backup chain based on the so called job id during backup job start), if it does not - you can manually map agent jobs to the backup files. Please keep in mind that only latest agent version v4 supports NFS repository.

One note to steps above, you can't remove extent if it is the only extent you have. So you need to add NFS repository first, then remove CIFS and add NFS in same step.

Make sense. Cheers!

[Mildur]

2. Remove the CIFS repository from configuration completely to avoid any issues with backup files being rescanned over NFS and CIFS. This will instruct Veeam B&R to "forget" existing backups (Remove from Configuration option does not delete actual backup files from disk).

This step cannot be done, because we can't delete a backup repo, if there is an existing job

[GianlucaCroci]

But it can be done if you use a Scale-out. right? Because you don't modify the Backup Job but you add/remove an Extents of the Scale-Out.

[Dima P.]

because we can't delete a backup repo, if there is an existing job

Correct, all other jobs should be pointed to a new simple (non-ScaleOut repository) as well. Added note to my previous post, thanks!

you don't modify the Backup Job but you add/remove an Extents of the Scale-Out.

Correct, it's not required to modify jobs pointed to the Scale-Out repository.

[Mildur]

Correct, all other jobs should be pointed to a new simple (non-ScaleOut repository) as well. Added note to my previous post, thanks!

Hi Dima

If I want to to modify a job to a new CIFS Repo, I need first to move the backup files.
Modifying the job will not work without moving the backup files first.
https://www.veeam.com/kb1729

I have converted 2 customers to NFS until today.
On Both customer, I have added the new NFS backup Repo without removing first the CIFS Backup Repo.
I haven't faced any problems for now. Incremental Jobs are continuing to running successfully.

[veremin]

If I want to to modify a job to a new CIFS Repo, I need first to move the backup files.

Have you tried to clone the given job and change its repository afterwards? Thanks!

[Mildur]

No, I can give it a try next time I have to convert a cifs Repo.

[veremin]

Ah, I'm so good in giving timely suggestions

Anyway, at least next time the process will be easier for you.

Thanks!

[Mildur]

Thanks veremin
i hope i can test it soon

[mcz]

Hi everybody,

I'm honest: I'm one of those "mad quys" who still use SMB on repository side
Now since I'm causing a lot of headache for you veeam-guys, I'm ready to switch to NFS which I already did for a new repository and it's working well. So now I'd like to switch from SMB to NFS for one of the primary repositories and I'm not sure which way I should go to make the switch "safe". So what do I mean?

On the target (repository) side, I can easily enable NFS for the same share but I guess in veeam I cannot just edit the repository and switch to NFS. So I'd have to create a new repository, rescan and map the existing backup jobs to this one. But when I add a new NFS repository and do a rescan, it basically contains the same information as the existing SMB-repository. Now I don't wanna mess up the whole database or any other structures. I could now create a new NFS repository, copy the contents there, but then I can't remove the old repository because many jobs like SureBackup and so on point to this repository.

What is the recommended approach and how can I achive a good result in a short time?
Thanks!