Netbackup 6.5.6 wont use all tape drives

[Gosiame]

Hi,

I've got 10 X LTO3 tape drives, 5 shared between the master and media and the other 5 are allocated to the master server. Everytime when the backups are running, you'll find 4 occupied on the shared drives and 4 on the master server's ones, meaning that 1 of each are not occupied but they won't be used even if there's queued backups that can be allocated to either of the available drives.

Once more, i've noticed that the master server won't use the shared drives if all of the drives allocated to it are used.

Please help!!

Regards

Gosiame

[RiaanBadenhorst]

What is the storage unit configuration?

[Gosiame]

The storage configuration is as follows:

For the master server: Maximum concurrent write drives was set to 4 instead of 5.

For the media server: Maximum concurrent write drives was set to 5 which is correct according to the drives configured for it.

Can you please let me know if i need to put a tick on "On Demand Only"

[Zahid_Haseeb]

See this for reference:

https://www-secure.symantec.com/connect/ko/forums/more-two-tape-drives-are-not-working

[Andy_Welburn]

put a tick on "On Demand Only" "

Probably not:

"The On demand only setting specifies whether the storage unit is available exclusively on demand. That is, only when a policy or schedule is explicitly configured to use this storage unit. Clear the On demand only check box to make the storage unit available to any policy or schedule.

For SnapVault and NearStore storage units, On demand only is selected by default and cannot be changed.

Caution: If On demand only is selected for all storage units, be sure to designate a specific storage unit for each policy or schedule. Otherwise, NetBackup is unable to find a storage unit to use."

[watsons]

If max. concurrent drive for master server is 4, the occupied drives for it won't go beyond 4.

If you want master server to make use of all 10 drives, incl. those 5 shared drives, configure max. concurrent drives to 10 then. I assumed you have SSO license.

[RiaanBadenhorst]

Between your own post and Watson's recommendation you have the answer! But keep in mind that the master could then "steal" all the drives.

[Gosiame]

Thanks for all your input, i'll give feedback tomorrow after tonight's backups. I did the necessary changes as recommended.

[Marianne]

If I understand correctly - 5 drives are visible on the Media Server. The same 5 drives are seen by the Master plus 5 more. This means in NBU there are 5 drives configured for the Media Server and 10 drives on the master. The STU config allows the Master to use ANY 4 drives and the Media server can use all 5 drives that is configured for it.

This means that the master server can use any 4 of the 5 shared drives and not only the 5 non-shared, forcing the media server to queue backup jobs due to insufficient available drives (in use by the master).

My suggestion:

Share all 10 drives between the servers and allow each of them to use any 5 drives.

Another option would be to dedicate 5 drives each to the two servers.

[Gosiame]

That's true Marianne, the media server has 5 drives allocated to it but the master server must see all 10 drives. When running backups and the non shared drives are occupied, then the backups won't mount tapes on the shared drives instead it will queue the backups. The changes done on the STU config now is as follows: Max concurrent write drives on the master server is 10 and on the media server is 5. I'll be able to comment on this tomorrow morning after tonight's backups. (Multichoice)

[Marianne · Accepted solution]

Please let us know.

Another possible reason for drives not being used is that the resource broker on the odd occation does not release resource allocations. This 'stuck' resource allocation will prevent drives from being used.

To check/verify, run 'nbrbutil -dump' (output can be quite large, so send output to a file).

Look at the bottom section for MDS Allocations. Carefully compare allocations with devices actually in use.

If you see MDS allocation for drives that know for a fact are not in use, you can release the allocation:

nbrbutil  -releaseMDS <mdsAlocationKey>

[Gosiame]

Hi,

This call can be closed now, i've monitored my drives for past days since i logged this call, my backups are running like a charm after doing the changes recommended. I don't experience unused drives and queued jobs anymore. Thanks to everybody that contributed their expertise.

[Andy_Welburn]

Please mark one post as the solution for your issue (probably that which assisted most - Riaan or watsons?) to allow other users to find a resolution to their issue if suffering from the same problem.

Marking a solution will also allow the post to "lock" (close) after a month without responses.