nbpemreq in 6.5.3

[Jay_Son]

I used to be able to specify "bpconfig -tries 0" which stops all scheduled backups (in 5.1), jobs near completion (while tries are set to 0) will allow the library (robot) to remove the media from the drive and into its perspective slot in the library. Now, with 6.5.3, utilizing "nbpemreq  -suspend_scheduling" stops scheduled jobs but also halts the library from completing its task (eg. setting media back to its slot when pending job(s) finally completes while in "nbpemreq  -suspend_scheduling" mode). When I see that the backup job is complete through my Activity Monitor, I click on my Device Monitor and realize that the robot has NOT emptied the drive(s). What i have to do is set "nbpemreq -resume_scheduling". With this said, I will not be able to maintain the library because other jobs just kicked off and forces me to "Cancel" the jobs. Needles to say, its somewhat inconvenient.

My co-worker is also experiencing the inconvenience of having to do a manaul catalog backup while in "nbpemreq  -suspend_scheduling". He has to  "nbpemreq -resume_scheduling" in order to kick off a "manual" catalog backup. there is not much documented in the "NetBackup Commands Unix and Linux.pdf" file.

What command(s) can we use to mimic what we did when using bpconfig -tries 0?

[Android]

There is a command option for nbpem "-terminate".  I would assume that this would stop the policy execution manager but not any active jobs, so that active jobs will complete and no new jobs would start since nbpem is stopped. 

However there is no detail in the documentation on exactly how this will affect active jobs.  It merely states, from the commands admin guide:

/usr/openv/netbackup/bin/nbpem [-console] [-terminate]
-terminate
Enables you to stop the NetBackup policy execution manager.

You would have to test to see exactly what sort of behavior will result.

[Anonymous]

nbpemreq -terminate also works.

In a situation like that I would have expected that bptm would have kicked in to unload the tapes, not leave them high and dry still in the robots drives.

I would also do a robtest and query that the drives have tapes in them as Device Monitor is showing.

[Jay_Son]

Stuart

Would you know the affects of running "nbpemereq -terminate" while jobs are running and queued jobs, as well? Will they automatically terminate? will the tapes still be high and dry in the drive(s)?

Thanks
Jay Son

[Android]

From the test I have run in my lab, nbpem -terminate does not cancel active, or queued jobs.  It will prevent new jobs from coming into the queue.   Is that what you were looking for?

[Android]

Jay Son,

OK Here is the definitive answer.  nbpem -terminate will not cancel jobs that are active or queued. It does prevent new jobs from coming into the queue.  Once the active jobs finished running, the tape was ejected from the drive by the robot and placed into it's slot in the library. 

Some things to note:
The only strange behavior I noticed was that if I had a job that had a parent and child, for example a BMR job, the child job finished writing data to tape and ejected the tape, that was then moved to library slot and the child job's state was changed to done with a status 0.  The parent portion of the job remained listed as active until I restarted nbpem (nbpem -start).  As soon as nbpem -start was run, the parent job's status also changed to done and the status was 0. 

Other than that one quirk it seems to behave exactly as you say you would need it to.  I hope this is the information you were looking for and if this is a solution for you please mark this as the solution to your question. 

Cheers!

[Android]

Jay Son,

Did you get a chance to test my results?  If so and it worked go ahead and mark the solution.  I would like to know if this worked for you or not. If not, I'll try to find another possible solution. 

[Jay_Son]

Android

I will edit my script to day to see if this is the case. I like your test results. If I can mimic the same results, this would more than be an accomplishment.

Just one more question, does the "nbpemreq -suspend_scheduling" command keep your tapes high and dry in the library? Its what we use and it does for us. Can you verify that in your lab?

Thanks

[Android]

I just ran another test using the "nbpemreq -suspend_scheduling" command.  the behavior was the same as the "nbpem -  terminate" command.  Jobs currently active and queued finished and the tape was ejected from the drive and moved back to its library slot.

[Jay_Son · Accepted solution]

Thank you for your diligence. Must be our library that's causing the issue. Thank you for your time. Your test results helped much.

Jay Son