[GEMPAK #MEQ-561539]: GEMPAK - Problems with GEMPAK and msgget: no space left on device

Christian,

When you get the msgget error below, it is signalling that you have had
a number of failures somewhere in you processing so that either gplt or device
drivers are not getting closed down.

When running the cleanup script, it attempts to locate all gplt processes,
as well as their children which would be device drivers such as gf, ps, gif, 
png etc.
You may have to run cleanup more than once if the list of processes is longer 
than your 
system can handle.

After running cleanup, see if you can find any gplt or device driver processes 
still running.
If so, then its possible that the "/bin/ps" output is not formated as expected 
on your
system. If that is the case, kill those processes by hand. Also run the "ipcs" 
commans and 
look at the message queue and semaphore id sections to make sure that cleanup 
is running the
"ipcrm" command as needed to remove those queue and semaphore ids.

Since you said that this is happening more frequently, you should determine if 
one
script in particular is failing, and where the error is located so that you
can prevent that from happening. You shouldn't have to resort to rebooting, but
it is possible that the cleanup script has to format the /bin/ps and ipcs
command output differently for your system in particular so that
you don't have to manually kill off processes and remove the queue / semaphore 
ids.

Steve Chiswell
Unidata User Support





> Institution: Universite du Quebec a Montreal
> Package Version: 5.9.4
> Operating System: 2.6.17-1.2142_FC4 x86_64 GNU/Linux
> Hardware Information: 4 Gb RAM
> Inquiry: Hello,
> 
> Since one or two months, I get more often these error messages from GEMPAK :
> 
> GDCNTR> [GDCNTR -3]  Fatal error initializing GEMPLT.
> Error in message send = 22
> itype, ichan, nwords,0,-1,3
> msgget: No space left on device
> msgget: No space left on device
> semget: No space left on device
> [GEMPLT -101]  NOPROC - Nonexistent executable.
> 
> I have then to reboot the machine to get it to work again. I do tried to run 
> manually the script to cleanup the GPLT processes and message queues (it is 
> also in the crontab), but no success.
> 
> I have this mounted for shm:
> /dev/shm               2030040         0   2030040   0% /dev/shm
> 
> Any idea why it happens and how to prevent it?
> 
> Many thanks!
> 
> Christian Page
> 
> 
> 
> 


Ticket Details
===================
Ticket ID: MEQ-561539
Department: Support GEMPAK
Priority: Normal
Status: Closed