some at jobs stuck, not running

[Hanry Zhou]

If I issue multiple jobs (>10), ex, at now, some of them will not be kicked off, and won't be run. Can you please let me know what the possible reason, and how to fix it?

I remember there is a limit on the number of at jobs you can run, but it could be tuned. I can't remember the detail.

thanks!

[Steven E. Protter]

swapinfo -tam

If you are out of memory jobs will not run.

Also check the kernel parameters nfiles and nproc

If you cant open a process the job won't run.

shmmax

If the job requires shared memory and there is not enough, same possible symptom.

SEP

[Hanry Zhou]

Thanks for the message.

I believe this is not parameters (you listed here) related, since I used to able to run a lot of other jobs same time, and won't have any problem.

This server is not heavily loaded, these at jobs are only jobs I'm running on this server. So I think it is at command specific problem.

[Michael Schulte zur Sur]

Hi,

In Tru64 Unix the crond has a limit of running parallel processes and this is configured in a config file. I wonder, if HP-UX has the same feature? Will look it up, as soon as in the office.

greetings,

Michael

[Michael Schulte zur Sur]

Hi,

have a look at /usr/lib/cron/log and /usr/lib/cron/queuedefs.

greetings,

Michael

[Elmar P. Kolkman]

On most HP-UX systems I've seen the logfile is in /var/adm/cron, though I see that /usr/lib/cron is a symlink.

You could also check out the mail file of the user for messages concerning the jobs that don't run (correctly).

The jobs that don't run, do they appear in the run queue? (at -l)

[Pete Randall · Accepted solution]

I suspect Michael has pointed you in the right direction with queuedefs. Do a man on queuedefs and check the contents of /usr/lib/cron/queuedefs. You're probably running into the max number of at jobs.


Pete

[Dieter Degrendele_1]

Hi,

Are you sure that the jobs terminate in a proper way? If this is not the case, you will see that cron jobs will fail after a certain time. Queudefs indeed.

Regards,
Dieter Degrendele