Maximum number of concurrent restores onto a single file space?

[atombaby]

Hi all

We're restoring a large (24+ TB) volume on a Windows 2008R2 server piecemeal, basically restoring each top-level directory within that filespace individually, vis:

dsmc restore s:\<dira>\<dirb>\ -subdir=yes -replace=no -tapeprompt=no
dsmc restore s:\<dira>\<dirc>\ -subdir=yes -replace=no -tapeprompt=no
dsmc restore s:\<dira>\<dird>\ -subdir=yes -replace=no -tapeprompt=no

.... and so on.... There's quite a few of those...

The client is 6.1, the server is AIX running TSM 5.5.

The problem (I think it's a problem) is that we can't get more than three of these going at the same time- one from the GUI, two from the command-line:

ANS5151S: This node currently has a pending restartable restore session. The requested operation cannot complete until this session either completes or is canceled.

My read of the Windows BA Client User's guide (see the "Restoring Large Amounts of Data" section) suggests that we should be able to do more.

However, if you enter multiple commands for the root directories in your c: file space, you can restore the files faster. For example, enter these commands:

dsmc restore c:\users\ -subdir=yes -replace=all -tapeprompt=no   
dsmc restore c:\data1\ -subdir=yes -replace=all -tapeprompt=no
dsmc restore c:\data2\ -subdir=yes -replace=all -tapeprompt=no

These specifications aren't overlapping on the file system, though there are some DFS junction point scattered about- we've rebuilt those and don't need to restore the file systems those junctions point to.

We've got "resourceutilization" and "maxnummp" set to 8, which I understand limits "producers" to two- could that be it? I can't find a specific limit on the number of restores- did I miss something in the docs?

Thanks all...

 

[Harry_Redl]

Hi,

I have never tested this so I am just guessing - is it possible you have tried the restore and then interrupted it? In that case restartable session ID is held in the TSM DB for (default 1day). Until it expires you cannot run new one (you can restart the original).
What "query restore" - on the server - says? I would stop all the restore sessions, cancel all restartable sessions ("cancel restore all") and try again ....
You can set the parameter "setopt restoreinterval 0" - so the restartable session IDs are expired immediately after the session disconnects (you may want to put it back later ....)

Harry

 

[ChrisRees]

Hi

Not really answering your question but thought you might benefit from reading this if you havn't done so already
https://www-304.ibm.com/support/docview.wss?uid=swg21142185

In your restore scenario at the moment, the last section on classic vs NQR restore might be of particular interest.

Cheers