Needless mounting tape during online catalog backup

[EvgenyL]

Hi, I'm on NetBackup 6.5 and tape library with 4 drives

An online catalog backup generates three jobs: a parent job, a child job for NetBackup DB, and a child job for catalog images and configuration data. These child jobs running consistently. (No more another jobs. Only online catalog backup.)

Tape TAPE001 and drive 1 are used in the first child job. The first child job finish. After that, the second child job starts and the same tape TAPE001 migrates to ANOTHER drive for writing data. 

Is it normal procedure for online catalog backup? Needless unmounting/mounting tape.

(in my bp.conf set NBRB_MPX_GROUP_UNLOAD_DELAY=300 )

Any advice appreciated. Thanks

[Mark_Solutions]

I have seen this happen and it does seem the normal behavoir as each job is an individual jobs and NetBackup is just load balancing the drive usage (though not the robot arm!!)

If you have disk available run the catalog backup to disk and the duplicate it (SLP or Disk Staging) and it will avoid this

Hope this helps

[Andy_Welburn]

Thought this was going to be about using multiple tapes as opposed to drives! So I presume from what you are saying that you have a STU for your catalog that restricts it down to one drive only so the second queues until the first finishes.

Looking at our last catalog, the first job is granted the media and drive at 02:02 & starts writing once loaded & positioned.

This job ends writing at 02:04:37. The next job starts 'immediately' at 02:04:45 and is granted the same media and the same drive a second later. Immediately it starts to write at the next file position on the tape.

We do not have the NBRB_MPX_GROUP_UNLOAD_DELAY value set in our bp.conf but retrieving it using bpgetconfig shows that it is ussing the default value of 10.

(Using NB6.5.6 on Solaris9 master with 4 drive library)

***EDIT***

Going to my first point "I presume from what you are saying that you have a STU for your catalog that restricts it down to one drive only so the second queues until the first finishes."

If you're not doing this, are you limiting it via max number of partially full media? Maybe worth trying a "limited" STU?

[VirtualED · Accepted solution]

There is a media unmount delay that is by default set to 180 second.  Assuming that has not change, then the only reason we would request an unmount of the tape is if there is a tape alert on the hardware or software level.

And example of a software level tape alert is that you have not ran a cleaning job via NetBackup.

Run the following from the media servers you are seeing this on:

volmgr\bin\tpclean -L

If it is saying the drive needs cleaning, then run a cleaning job via NetBackup.

[Andy_Welburn]

Good point.

[EvgenyL]

.. no more needless tape remounting during catalog backup!!!

Thanks for advice!