Questions about redirect restore from advanced disk pool

[sonorous]

Dear all, My enviroment is 1 master, 10 media server, os is both aix, backup destination is disk pool. Since nbu7 dones not support shared disk, we maybe use advanced disk. Here is my question, 1. Any solution to use shared disk in nbu7 version, we have too many media server. 2. If we use advanced disk and assign Luns for each media server, question is if one media server fail, can I use another media server mount the lun and perform restore jobs.

[AAlmroth]

Hi,

The solution would be to use Advanced Disk and Storage Foundation Cluster File system (CFS). This way you can present all LUNs to all media server participating in the disk pool, and have a CFS mounted file system on all media servers. If one media server is down, NBU could try next media server in the disk pool. Can possibly be configured with storage unit group (failover mode), or by using failover read server, or even using the preferred/required read server option on disk pool level.

This approach minimize your manual work if the media server fail. If you present the LUN to another server, you would manually have to change the ownership of the images (bpmedia command), before you can restore. Using a disk pool on a CFS volume is absolutely the best approach. It does however come at a higher license price as you would need SF/CFS as well. Unless you can do some CFS alike in AIX.

/A

[sonorous]

thanx for you reply. Due to the reason of higher price, we will not to use CFS.

If we use advanced disk and assign serval LUNs for each media server, the question is: once a media server down, can I mount the LUN, import FS and perform restore by another, could you give some detailed process.

There also two detailed questions

1.Must the mount point be same on both old and new media server.

2.Before restore data from the LUN, any import process is required?

[AAlmroth · Accepted solution]

I haven't tested this, as we always implement using SF/CFS, but in theory;

1. Add the new server as storage server to the disk pool of the failed server.

2. Mount the relevant file systems.

3. Merge the disk pool from the old server, with the other disk pool on the new server.

4. Remove old server from disk pool.

Maybe step 3 is redundant, if you can set preferred server to the new server on disk pool level.

Thing here is that all images are referenced to the disk pool id (@aaae as an example). So therefore it is important that the new server use the same reference. You may also have to update the storage unit, so that the new server may use the disk pool (unless you use "use any server").

If possible, I will try to set up in the lab, but as I just have moved, no gear is running...

/A

[sonorous]

Your answer really helped, I appreciate your help.