Backup agent for Microsoft Windows servers and workstations (formerly Veeam Endpoint Backup FREE)
Post Reply
rstoner
Influencer
Posts: 12
Liked: 1 time
Joined: Aug 21, 2017 7:32 am
Full Name: Roderick Stoner
Contact:

Failover Cluster

Post by rstoner » Apr 13, 2018 5:36 am

Hi,

I have created two Protection groups, one configured as individual computers (Group A) and one that is Active Directory enumeration (Group B). I added the failover cluster nodes individually into Group A, whilst I added the failover cluster computer account from AD into Group B.

Then I set two jobs, one as server backup (Job 1) VBR managed the other as Failover backup (Job 2), I added Group A to Job 1 and Group B to Job 2 and launched a backup of each, one after the other. However, both backups are identical,there is nothing extra from the failover backup that isn't in the standalone backup. Which prompted me to ask the question "What is the difference between backing up individual nodes within a cluster and the cluster itself?"

The cluster currently has a file service role. However, we are planning to use Veeam agent within VMs that have RDMs and are running an MS SQL cluster. We don't need to backup the MS SQL as there is a SQL Job that does this and dumps the backup to a drive, we only need to backup that drive. So, is there any benefit by setting up Failover Cluster backup as opposed to individual nodes?

Thanks!

Dima P.
Veeam Software
Posts: 8426
Liked: 616 times
Joined: Feb 04, 2013 2:07 pm
Full Name: Dmitry Popov
Location: Prague
Contact:

Re: Failover Cluster

Post by Dima P. » Apr 13, 2018 6:03 pm

Hello Roderick,

Jobs with server type do not process cluster disks and cannot track failover. Due to this reason we cannot grantee the application constancy while you are protecting your SQL cluster in a regular server job.

rstoner
Influencer
Posts: 12
Liked: 1 time
Joined: Aug 21, 2017 7:32 am
Full Name: Roderick Stoner
Contact:

Re: Failover Cluster

Post by rstoner » Apr 16, 2018 4:52 am

Hi Dima,

From the backup I did with server type, I can still the cluster disks, so somehow it is processing them. What I am not sure about is how the following will work; if today the disks are on node A and we backed them up, then tomorrow they failover to Node B, and again backup. How will I know from which node I need to restore the data, as on day 1 the cluster drives were on Node A and on day 2 they're on node B.

Dima P.
Veeam Software
Posts: 8426
Liked: 616 times
Joined: Feb 04, 2013 2:07 pm
Full Name: Dmitry Popov
Location: Prague
Contact:

Re: Failover Cluster

Post by Dima P. » Apr 16, 2018 4:17 pm

rstoner wrote: From the backup I did with server type, I can still the cluster disks, so somehow it is processing them
Can you please provide a screenshot? Thank you.

rstoner
Influencer
Posts: 12
Liked: 1 time
Joined: Aug 21, 2017 7:32 am
Full Name: Roderick Stoner
Contact:

Re: Failover Cluster

Post by rstoner » Apr 18, 2018 7:17 am

Here you go.. these servers are backed up using standalone server VBR managed. Drive Q: is the disk witness and drive E: is a drive given to the Clustered File Service.

https://ibb.co/iyQRsS
https://ibb.co/eaJPe7

Dima P.
Veeam Software
Posts: 8426
Liked: 616 times
Joined: Feb 04, 2013 2:07 pm
Full Name: Dmitry Popov
Location: Prague
Contact:

Re: Failover Cluster

Post by Dima P. » Apr 20, 2018 10:51 am

Thanks for the screenshots. Yes in some cases volumes on the cluster disks might fit into backup. As I said before the problem is with the failover. With failover cluster job we do track the failover and continue the cluster disk backup incrementally. With regular server job (in case of standalone deployment and managed by Veeam B&R) backup will likely fail whenever failover is performed.

rstoner
Influencer
Posts: 12
Liked: 1 time
Joined: Aug 21, 2017 7:32 am
Full Name: Roderick Stoner
Contact:

Re: Failover Cluster

Post by rstoner » May 04, 2018 6:51 am 1 person likes this post

Hi,

Thank you for the reply. Yes we did configure them using the cluster in the end, makes more sense. However, it was interesting that we stumbled on this during our tests and wanted to share + get to know why :)

Post Reply

Who is online

Users browsing this forum: kevind01 and 13 guests