Backup of Cluster

[Hinux]

Hi

We are implementing Oracle RAC running on 4 Linux Servers.
I am a bit baffled on how to approach the backup on the cluster. The setup doesn't use 'n virtual node and I have never backed up a cluster without the virtual node.

At the moment everything is backed up twice , because I dont know how to tell Legato that the Servers are Running in a cluster.

Any comments/ideas regarding Oracle RAC backups would be great.

Regards
Hinux

 

[NWNinja]

Hi Hinux!
Here´s some information i found regarding this.

---
Follow docs in the NMO Admin Guide which reference OPS and RAC environments, but you don't need to setup a virtual client due to the intrinsic behaviour of the way these Oracle environments behave.

The NMO Admin Guide references virtual nodes in the chapter describing the installation in an Oracle RAC environment. For RAC or OPS environments it is not necessary to create a virtual NMO client for the backup
---

Installation guide:

http://web1.legato.com/infodev/publications/BSM_Oracle/4.2/nmoig_4.2.pdf

Admin guide:

http://web1.legato.com/infodev/publications/BSM_Oracle/4.2/nmoag_4.2.pdf

Hope it helps!

Cheers!
Maverick

 

[Hinux]

This seems to be if you backup the nodes as individual nodes - by their own names, or I am missing something.

This doesn't seem to allow for any Failover backup and
doesn't have any failover intelligence built in. It is seems to be treated purely as a node ,however, we have
duplication of backup data - as both nodes "share" the same disk.

What we are trying to accomplish is to backup the data from a particular node (the active node), and in case node A Fail restart the backup on node 2. Similarly for restores.
This seems fairly easy when you make use of virtual nodes, but in the RAC setup, there is no virtual node. The RAC manages the nodes.
Which makes it confusing.. What do I put in The "Save set" field of NODE A and B if I put : RMAN:/backup/database/scripts/TSTF_FULL.cmd
in both nodes "Save set" fields, they are going to backup the same thing? because it is shared storage.

How will Networker know that NODE A has failed and should try NODE B, without talking to a virtual host.

Thanks for any feedback.
Hinux