cancel
Showing results for 
Search instead for 
Did you mean: 

Stopping Netbackup

Venki009
Level 4

Hi Team,

When I recycling the services(stopping netbackup on master).All process are stopping mean I used bp.kill_all to stop servcies but it doesnt work and i used manually killing kill -9 , it also not working.

When Iam using bp.kill it showing waiting to terminate bptm and bpdm processess.bptm and bpdm process are not stopping.

And One more thing the reason to recycle is when i intiate manual backup it not initiating backup.

Thanks in advance

1 ACCEPTED SOLUTION

Accepted Solutions

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

" waiting to terminate bptm and bpdm processess " means you have backups that are running.
More than just processes need to be terminated - port connections need to be terminated.
The termination instructions need to be sent to clients as well and all of this takes time.

 

Best to kill or suspend all running backups before stopping NBU to ensure clean termination of server and client processes and unload of tape drives (if backups are written to tape).
Killing bptm and bpdm processes may leave the environment in a bit of a mess with port connections not closed and tapes left in drives.

 

View solution in original post

7 REPLIES 7

Yasuhisa_Ishika
Level 6
Partner Accredited Certified

When a process can not be terminated by SIGKILL(signal #9), it is most likely that this process is calling some system call and it is blocked by any reason in kernel side. Are there any messages repeated in OS log?

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

" waiting to terminate bptm and bpdm processess " means you have backups that are running.
More than just processes need to be terminated - port connections need to be terminated.
The termination instructions need to be sent to clients as well and all of this takes time.

 

Best to kill or suspend all running backups before stopping NBU to ensure clean termination of server and client processes and unload of tape drives (if backups are written to tape).
Killing bptm and bpdm processes may leave the environment in a bit of a mess with port connections not closed and tapes left in drives.

 

mph999
Level 6
Employee Accredited

I'd recommend stopping NBDB before killing processes ...

/usr/openv/db/bin/nbdbms_start_server -stop

... else you run the risk of corrupting the DB / Transaction log

Then kill whatever you like ...

Venki009
Level 4

Hi Team,

One more question,

When I intiate manully  the backups there not starting.

Thanks in advance

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

Is this now after you you stopped NBU and did a kill -9 of remaining processes?

Understand what I was trying to say about the fact that killing processes in this way may leave the environment in a 'bit of a mess'?

Did all processes stop cleanly?

Have you checked that all processes are running after you started NBU again?

Best to start a new discussion for this issue.... We may need to dig into nbpem, look for left-over lock files and/or pempersist, etc as a result of forceful killing of processes.

watsons
Level 6

I remember there's once I was unable to kill some bptm, and some tldd processes, even with "kill -9" that was on an old Solaris box.

Ended up rebooting the server to clean up the hung/defunct process. That was some very old Netbackup version though.

bbahnmiller
Level 4

I'm still waiting for the "quiesce" button for recycling NetBackup. Our process usually goes somewhat like this:

  1. Suspend scheduling
  2. Deactivate all media servers.
  3. Pause all running backups.
  4. Kill any running backups that cannot be paused.
  5. Set SLP's inactive
  6. Kill any active SLP's
  7. Kill any running duplications (separate from SLP's)
  8. If you have tape drives, verify that tape drives are empty

(I'm probably forgetting something.)

Then you can run "netbackup stop" or "bpdown". If everything is quiesced, NetBackup will stop more cleanly, with few, if any errors. I think both 7.5 and 7.6 have been a lot better about cleanly shutting down than previous versions have.