[Veritas-bu] Storage Unit Strangeness
2007-04-24 13:17:19
Subject: |
[Veritas-bu] Storage Unit Strangeness |
From: |
richard.r.green at boeing.com (Green, Richard R) |
Date: |
Tue, 24 Apr 2007 10:17:19 -0700 |
We had this same problem where there were more drives used than was
stated in the storage_unit parameters for backups. We have NetBackup 5.1
MP4 on a Sun box with Solaris 10. We have two storage_units. One for
backups on newer tape drives and another storage_unit for old technology
drives to use for restores of older backups.
After opening a case with Symantec it was determined that we needed to
do the following;
Create a line in /usr/openv/netbackup/db/config/behavior/ called
MAX_DRIVES_THIS_MASTER 12. 12 being the total number o tape drives that
will be used for backups. You will need to determine this number for
your Master. We do this so there is a least one drive available for
restores.
You will then need to stop and restart NetBackup.
>From the case;
The calculation of active jobs and concurrent jobs is based on:
Definitions.
Active jobs is the sum of mpxgrps (multiplex groups) and a multiplex
group is equivalent to a tape drive.
Concurrent jobs is derived as a minimum of either the available drives
in the robot or the maximum concurrent jobs for all storage units in a
robot.
For this problem: max concurrent jobs is 32 x 12 = 384, mpx per
drive is set to 32 and 12 drives in the storage unit.
To solve or try to solve the problem comes down to utilizing the
parameter: mdtm MAX DRIVES THIS MASTER.
Bpsched algorithm will check for a hard limit, mdtm and stop allocating
drives at that point.
Any drives that are not allocated due to the hard limit , remain
available for restore work
Hope this helps.
|
|
|