Cluster support for agents

[pj888888888888]

So our primary use case for this agent is mainly for our Oracle RAC's, and unfortunately we are seeing this isn't supported, which is problematic as we are trying to simplify our data protection environment by using Veeam, but now the primary reason for even buying these agents isn't really going to work due to not supporting clustered nodes.

Right now we are using it and its actually working, we just have omitted the volumes used by ASM and are only doing local mountpoints (which are all EXT4, not ocfs2)

What is the roadmap for getting this feature added?
Right now we might have to consider going back to our legacy product due to this limitation....which is very irritating as this is going to severely complicate our plans.

I'd like it if this could be added for us.
Preferably in two phases:
1) Ability to backup the local (non-clustered) filesystems on cluster nodes (Which we have been doing without issues....so I almost think this could be verified tested relatively quickly)
2) Ability to backup the clustered filesystems on cluster nodes (IE a "catch-all" for anyone using ocfs2, GPFS, etc..)

[jmmarton]

If you can export the clustered filesystems via NFS (or CIFS), then our upcoming NAS support in Veeam Backup & Replication v10 might be an option to back these up.

Joe

[pj888888888888]

The clustered FS is from an Oracle Exadata presented as OCFS2, so that wouldn't really work.
I mean honestly......we really only care about the local mountpoints...RMAN protects the data within ASM, so that would be redundant trying to back those up.

I have been reading more about clusters and Veeam, and from my understanding....its not supported on the level that it wouldn't be aware of a failover event (which our use case...that really isn't a huge concern)

Support more or less said its not supported and closed the ticket on me without clarifying my use case....which is simply for local protection of the single node, not the clustered portions.
I'd really just like to get this clarified....that way I can continue down the merry path I am on and once you guys support clusters....we can say we have more options if our DBAs grow weary of RMAN (That is unlikely)

[PTide]

Hi,

That is, it would be sufficient for your case if VAL could just automatically skip all OCFS/ASM devices and backup only local mountpoints, is that correct?

Thanks

[pj888888888888]

I mean that would be cool too, that would make setting the jobs up a tad easier.
Honestly, I have it configured to omit the OCFS/cluster volumes and only do local mountpoints, and its been working fine.
I have even conducted a BMR several times on a cluster node without issues.

But.....I'm being told its officially "not supported" so I had to roll most of the clusters in Linux back to Netbackup (which is irritating as the recovery process with Veeam agent is 100 times easier for a BMR)

I had even spoke with an escalation manager about getting this supported (just for image backup), but QA stated that was a no-go as it opens up too many potential problems due to the nature of the product.

Im still a little irritated as this was going to provide massive cost savings (and data protection consolidation)

[Scott_K]

What is the current status for supporting VAL on Oracle Exadata (physical servers) - and 'only' backing up the OS partitions? Note - These are not clustered file systems, and i'm not asking about ASM disks...

[root@exatadata1-mgmt ~]$ uname -a
Linux exatadata1.f.q.d.n 4.1.12-94.7.8.el6uek.x86_64 #2 SMP Thu Jan 11 20:41:01 PST 2018 x86_64 x86_64 x86_64 GNU/Linux

[root@exadata1-mgmt ~]$ df -h
Filesystem Type Size Used Avail Use% Mounted on
/dev/mapper/VGExaDb-LVDbSys1 ext4 30G 15G 14G 51% /
tmpfs tmpfs 252G 16K 252G 1% /dev/shm
/dev/sda1 ext4 488M 29M 424M 7% /boot
/dev/mapper/VGExaDb-LVDbOra1 ext4 99G 61G 34G 65% /u01

[root@exadata1-mgmt ~]# pvs
PV VG Fmt Attr PSize PFree
/dev/sda2 VGExaDb lvm2 a--u 114.00g 29.00g
/dev/sda3 VGExaDb lvm2 a--u 1.52t 1.42t
[root@exadata1-mgmt ~]# lvs
LV VG Attr LSize Pool Origin Data% Meta% Move Log Cpy%Sync Convert
LVDbOra1 VGExaDb -wi-ao---- 100.00g
LVDbSwap1 VGExaDb -wi-ao---- 24.00g
LVDbSys1 VGExaDb -wi-ao---- 30.00g
LVDbSys2 VGExaDb -wi-a----- 30.00g
LVDoNotRemoveOrUse VGExaDb -wi-a----- 1.00g



As you can see, all the ASM disks do not show up here. The DBAs manage Oracle backups with RMAN - as the Storage/Backup Engineer - we only care about backing up the OS.

Is this supported?

[PTide]

Hi,

As long as the machine comes with a supported operating system. It seems that in your case you have Oracle Enterprise Linux 6 (UEK Release 4) which is supported, but VAL has never been tested on actual Exagrid systems. I'd recommend you to make sure that the OS on the server is indeed Oracle Linux without Exadata-specific modifications. Also you should keep in mind hardware drivers - should you decide to try VAL, don't forget to create custom recovery media.

Thanks

[pj888888888888]

So.....what I'm seeing is if as log as we aren't attempting to backup clustered filesystems, then the system would be supported?
Because I had a similar situation where the agent caused a system to suddenly reboot when the backup started and support had told me that due to their being clustered storage on the node, that it was "not supported" and the ticket was closed.

Has the verbiage changed since then to loosely support our scenario?

[PTide]

Unfortunately it hasn't changed yet. While technically VAL will backup cluster nodes' root partitions, we haven't conducted extensive testing on such setups, and therefore we don't know what problems might emerge during/after restore. We haven't forgotten about the existence of clustered file systems, and fully understand the need to backup them too.

Thanks