Failover Cluster backup

[deubelm]

Hello,
we want to use Veeam B&R to backup our Windows fileserver. The fileserver are 3 node Windows Failover Cluster.
Once in a while we are faced with full backups because of the message "Full backup is required for cluster disk <Clusterdisk>: cluster membership was changed". We could find any explanation what it means and what might be the cause.
We checked the cluster has not failed over.

Regards
Marcus

[deubelm]

I meant: We could not find any explanation what it means and what might be the cause.

Nobody has ever encountered this message "Full backup is required for cluster disk <Clusterdisk>: cluster membership was changed" while backing up Windows cluster?

[Dima P.]

Hi,

This message should be displayed when failover was performed. If you confirm that failover was not perfromed, kindly, open a support case to investigate the root case. Please do not forget to share the case ID. Thank you in advance.

[deubelm]

Hi,

This message should be displayed when failover was performed. If you confirm that failover was not perfromed, kindly, open a support case to investigate the root case. Please do not forget to share the case ID. Thank you in advance.

Thanks Dima for the reply. So can you confirm that a full backup is triggered every time a fail over is performed?

[deubelm]

Hi Dima,
is it planed to correct this behavior performing a full backup if a failover has occured? It prevents us to use the cluster function.

[chrislott]

Dima, did you find a solution for this? We're having this problem with our failover cluster every time it backs up, but we aren't changing memberships or having a failover situation or anything...

[Dima P.]

Hello and welcome to the community Chris,

This behavior is unexpected, so please submit a support case and share your case ID in this thread. Thank you in advance!

[chrislott]

My case # is 03476838.. we can't find the cause for this full backup. Our failover cluster has no events or failovers or anything, yet every time we do a backup it gives this message.

[Dima P.]

Chris,

I've discussed this issue with QA team. Full backup might be required only when one of the nodes went down during previous job run, so please keep working with our support team to understand why full backup is performed in your case. Thanks!

[egonthier]

Hello,

Same problem for me, did you find a solution?

Regards

[Dima P.]

Hello Emilien,

Since this behavior might be related to multiple issues pleas raise a support case and share the case id for future reference. Thank you in advance!

[lang.matthias]

Hello,

I´ve the same "problem". As Dima already said it was a limitation in Agent GA 2.2 (Important section - https://helpcenter.veeam.com/archive/ba ... pport.html)
"In case a backup task within a Veeam Agent backup job that processes a cluster completes unsuccessfully, Veeam Agent for Microsoft Windows will create full backup of all shared disks of the cluster."

My problem is, that if i have an planned downtime for my fileservers for updates,... i switch my cluster packages to other nodes so that they have an max. availability.
At this point i have an problem when the backup needs an full because the filecluster has about 60 TB

I´m now going to update my agents on 4 filecluster nodes to the actual version (3.0.1.1039) - so is this behavior also in the new Agent?

Best Regards
Matthias

[Dima P.]

Hello Matthias,

My problem is, that if i have an planned downtime for my fileservers for updates,... i switch my cluster packages to other nodes so that they have an max. availability.
At this point i have an problem when the backup needs an full because the filecluster has about 60 TB

Can you disabled the backup jobs for the time of planned downtime? If yes, next run wont be a full backup but a full read (job might need to read the shared storage due to failover but that wont result into full backup). Cheers!

[Tobias K.]

Hello,

same problem here.
Is there any solution in sight?

Some details about our case: The CBT-driver is installed on the active node in our cluster. The volumes are single drives active only on the current failover role master.
I've created the last full backup after the CBT driver was installed. Still, the incremental backup states that a full backup is needed (same as in first post in this thread).
The incremental was startet just after the full backup finished, so I'm pretty sure the cluster configuration hasn't been changed at all. Never the less, the goal should be for incremental backups to work even if the master role changes from one node to another.

For details about the hardware and OS please extract the information from:
https://social.technet.microsoft.com/Fo ... Clustering
(we haven't found any solution to this problem either...)

Current veeam Support case: #03641047 (I just introduced the problem and added a link to this forum thread).


Best regards
Tobias

[Dima P.]

Hello Tobias,

This issue is being investigated by our support team, so kindly keep working with your engineer and let us know how it goes. Cheers!

[Tobias K.]

Hello everyone,

just to inform anyone else regarding this problem: the case has to be closed unsuccessfully.
We're running a file cluster with "storage spaces direct". Although the disks are mounted as local drives with drive letters(instead of cluster-shared-volumes) this is not supported by veeam. We hope for the R&D-Team to implement this feature in future, but sadly we weren't given any ETA for this feature.

Best regards
Tobias

[s.middlebrook]

We have a new Veeam Availability suite server and backup storage which I have setup. And was wondering if Storage Spaces Direct supported?

The 'normal' Cluster File Servers (connect to Equallogic iSCSI SAN) are working well, as well as the Virtual (VMWare Environment). But the S2D cluster every day comes up with the following error message: "Full backup is required for cluster disk {GUID}: cluster membership was changed".
I have setup forever incremental backups (currently for 7 days), but due to it saying cluster membership was changed it completes a full backup every time. Does CBT not work with S2D?
The File Server Role has not changed/moved/failed-over from one server to the other from the previous days backup.

I have read here: veeam-agent-for-windows-f33/failover-cl ... 51763.html this is a similar issue but related to Hyper-V.

Any one else had any success stories with File based Server Roles shares with S2D and Veeam?

Thanks in advance!

[wishr]

Hi Stefan,

Welcome to Veeam Community Forums and thanks for posting!

S2D is not officially supported yet, however, this is somewhat we do plan to implement in later product versions (no ETA yet). As a workaround, you may back up each node separately.

I'm merging your question with the existing thread.

Thanks

[sraj]

Hi Dima, Is the Original issue is addressed in v10 Agent version ?

[Dima P.]

Rajesh,

Several issues were, addressed with v10 but full read is still required if backup job failed and at the same time failover was performed. Can you please clarify your issue? Thanks!

[sraj]

"Full backup is required for cluster disk <Clusterdisk>: cluster membership was changed". - This message even though no failover is occured for incremental backup

[Dima P.]

Can you please confirm that cluster configuration was not changed (i.e. new node was not added to the cluster)? Thanks!

[ludo.augeard]

same for me.. no change but full again with membership changed

[Dima P.]

Thanks! Please raise a support case, share the case id with me and I'll ask QA team to review the debug logs.

[paul.timson]

Hi, we are having this issue on our 4 node windows cluster and wondered if any solutions have been found. I have raised a support ticket in the meantime but hoping there's an answer here from someone?

[Dima P.]

Paul,

Can you please share the case ID? Thank you!

[paul.timson]

Case #05254989 — Windows cluster with storage replica

[paul.timson]

sorry got 2 cases open on the same clsuters
Case #05249401 — Hyper-V Cluster disks membership changed - without failover

[Dima P.]

Thank you! Forwarded both to QA folks for a review.

[JEiben76]

I know it's been a while, but in my newly-gained experience, you can get back to incrementals by:

1. Stop the running job
2. Failover to the node with the previous successful incremental backup
3. Start the job

It's not great, but it beats waiting for TB after TB of full backups.