L1000 Sam Backup Failure

[BruceDFreeman]

Hi,

I am trying to manually run an automated Sam Backup on an L1000. This backup has not been run for 6 months or more but fails with the following screen message:

Starting interactive backup...
br_backup: Invoking fbackup. See /var/sam/log/br_log for details.
fbackup(1421): no history is available for graph file /var/sam/graphZCAa21538 (b
elow level 0)
fbackup(1004): session begins on Wed Nov 10 09:18:08 2010
fbackup(1010): semget failed for the semaphore

Press to continue sam...

And in the br_log :

{ fbackup -0 -u -v -g /var/sam/graphZCAa21538 -I /var/sam/log/br_index.full -c
/etc/sam/br/fbackup_config -d /var/adm/fbackupfiles/dates -f /dev/rmt/0mb } 2>&
1
Full backup failed on beales2 : Wed Nov 10 09:13:04 GMT 2010 .
Exit code = 2

Full backup started on beales2 : Wed Nov 10 09:18:07 GMT 2010 (Scheduled Backup)
{ fbackup -0 -u -v -g /var/sam/graphZCAa21538 -I /var/sam/log/br_index.full -c
/etc/sam/br/fbackup_config -d /var/adm/fbackupfiles/dates -f /dev/rmt/0mb } 2>&
1
Full backup failed on beales2 : Wed Nov 10 09:18:08 GMT 2010 .
Exit code = 2

Has anyone come across this and can point me in the right direction.

Thanks very much,

Bruce

[Dennis Handly]

Try invoking that fbackup command directly?
Could the missing error be:
fbackup(1421): no history is available for graph file /var/sam/graphZCAa21538 (below level 0)

[rariasn]

Hi Bruce:

http://forums11.itrc.hp.com/service/forums/questionanswer.do?threadId=959254

rgs,

[Matti_Kurkela]

>fbackup(1010): semget failed for the semaphore

Hmm... is your system perhaps running out of semaphores?

Run "ipcs -s" to list the current semaphore usage. Compare with the sem* kernel parameters: are you at or close to the limit set by semmni/semmns kernel parameter? If you are, examine how your applications are using semaphores: is the current usage level normal or not?
(If you have glance, start it and press "t" for a report on system tables. Check the utilization level of the semaphore table.)


If the usage level is normal, the kernel sem* parameters are probably set too low: you may need to increase the appropriate sem* parameters and reboot.

If the usage level is higher than expected, you may have an application that leaks semaphores. Regular memory leaks are automatically reclaimed by the OS when the leaky process exits; semaphore leaks are not. Leaked semaphores must be cleaned up manually, using the "ipcrm" command or by rebooting the system.

Another possible cause of extra semaphore usage is that someone has a bad habit of using "kill -9" to stop an application that uses semaphores.

A regular "kill" command will give its target a chance to "clean up" before exiting; a well-designed semaphore-using program should always have a signal handler that removes any semaphores and other shared resources the program has allocated before the program is terminated. If "kill -9/kill -KILL" is used, this signal handler will not be run, and any shared resources used by the program must be cleaned up manually.

MK

[BruceDFreeman]

Thanks for your help. I increased the semmni and semmns kernel parameters and it is running now.

Once again thank you,

Bruce