Veeam and Storage Snapshot Only on NetApp Metro-Cluster

[LaurentS]

Hello,

Infrastructure :

Veeam Backup and Replication 9.5 u3
PROD : 2x NetApp FAS 8200 in Stretch MetroCluster Single Node.
Backup : 1x NetApp FAS 2620 - independant cluster
Data ONTAP 9.1p9

We want to implement Storage Snapshot Only (in FAS8200 metrocluster) and a Storage Snapshot Only to a Secondary Target (FAS2620) with SnapVault.

Everything works as expected, but when there is a NetApp switchover / switchback the relationship is broken (In Backup Job error message : cannot find any relevant NetApp SnapVault and Snapmirror relationShips).

Could you please confirm me that this configuration is supported ?
If yes, do you know how can i resolve this issue ?

Regards.
Laurent

[Andreas Neufert]

Hi, yes this should work.

Please open a support ticket and upload logs.

[foggy]

Hi Laurent, there should be a hotfix for this issue, please contact technical support to get it.

[LaurentS]

Thanks to you,

I just opened a case (02737992 french language) to ask if this infrastructure is really supported.
You tell me, yes that should work, thanks, but can you confirm if this solution is really supported by Veeam (cf : Storage Snapshot Only on Stretch Metro Cluster 2x FAS8200 + Storage Snapshot Only Snapvault to a secondary Independant Cluster FAS2620?

If so, i will create another case to ask for help and maybe a hotfix
If No, we will think for another way to Backup the Infrastructure.

Regards.
Laurent.

[foggy]

Generally such configuration is supported. Once you provide log files within the case, we will be able to review them and verify whether the available hotfix is applicable.

[LaurentS]

Thanks for the confirmation Foggy, i just got the same answer from your Support.
I will provide log files tomorrow.

[GoranK]

LaurentS

If you have this issue, you can manually update the SnapVault relationship from CLI on the NetApp and then Veeam will pick up from there.
Worked for us anyway.

[LaurentS]

Hello Goran,

Thanks for this information, after the NetApp switchover / switchback, we tried to use the mirror update CLI but without success.

Netapp > snapmirror update -destination-path vserver:volume

For information, we still have the veeam case open and i will update this post when i will get some more information.

Laurent.

[GoranK]

Hi,

You have to specify the option "-source-snapshot" as well and specify the latest snapshot on the source.
I also re-scanned all the volumes on both the primary and secondary NetApp in Veeam after the snapmirror update finished.

[LaurentS]

Hi,

We tried another switchover / switchback scenario today, and for the snapmirror update cli we specify the -source-snapshot as well with the latest snapshot on the source, and it works !
So thanks for this information!

This solution is a workaround and need some manual tasks from Storage Team, i hope Veeam will fix this issue in a futur release of veeam.

Laurent

[orb]

Hello,

Can we leverage a backup from snapshot aas snapmirror/snapvault mechanism with a NetApp metrocluster and Veeam? I didn’t find really any literature about this subject. I do know you can access a snapshot from the node who hosts the a replicated volume.

Oli

[DGrinev]

Hi Oli,

That should be possible, can you elaborate a bit on the infrastructure you have?
Please take a look on the existing discussion with the similar question. Thanks!

[orb]

Hi,

The customer has a metrocluster between 2 distant site. Each site runs a vSphere cluster. Both Cluster datastores are available on each sites but there is only 1 physical proxy.
We would like to avoid using the link between site to read the data if possible.

Oli

[foggy]

Hi Olivier, Veeem B&R reads data from the primary node (unless backup from secondary storage is configured), if you want to prevent it from reading over the remote link in case of switchover, you should disable its access to failover node.

[orb]

Foggy,

Is it correct to say, my DirectSAN on the node close to the proxy will allow Veeam to read all the metrocluster storage domain even if the primary source is actually located on the other side ?

Oli

[foggy]

Do you mean if you disable storage integration and use direct SAN access instead? In this case yes, you will be able to use the closest node.

[orb]

no I meant I enable the storage integration point but register all SVM on same node

[foggy]

So if you configure the job to use secondary storage (SM/SV) as backup source, it should read locally.