'at now' jobs not working . . .except when it's working

[Johns Manville]

I have an HP-UX 11.11 server that doesn't run an 'at now' job around 2 a.m. sometimes. When I test 'at' jobs during the day by running them 'now,' they run fine.

This server is an Oracle server and is very busy, although not maxed out at 2 a.m. In fact, the server is less busy than it's been for several years, but the 'at' jobs used to work without issue, and are now failing to run during the server busy times.

Shouldn't the 'at' jobs run regardless of how busy the server is? Or could the 'now' timing be an issue where the server is too busy to execute it "now," and when it gets some free CPU cycles and looks at it again, it's past 'now' so it doesn't execute? (Not sure if that made sense)

Any ideas on how to make the 'at now' job guaranteed to run?

[Johns Manville]

Additional info:

The values in /usr/lib/cron/queuedefs are:

hpux05
a.8j1n
b.2j2n90w

I'm not sure what the 8j1n means for 'at' jobs, so maybe someone can explain that? Maybe that setting needs a change to improve reliability?

Thanks.

[James R. Ferguson]

Hi:

Look at the manpages for 'queuedefs'. This will show you how to decipher your posted values.

Regards!

...JRF...

[Johns Manville]

Duh! I should have done that. :-)

[James R. Ferguson]

Hi (again):

I still like your subject line --- it's a classic :-)

...JRF...

[Johns Manville]

Well James, thanks . . . I think. :-)

I wish I could get some ideas on this. I raised the number of simultaneous 'at' jobs from 8 to 28. I'll see if that makes a difference.

Thanks again.

[Patrick Wallek]

Have you checked /var/adm/cron/log and/or /var/adm/syslog/syslog.log for any errors at the time the jobs fail?

[Johns Manville]

I don't see any issues in the logs.

[Bill Hassell]

Perhaps the first thing to do is to verify that the jobs truly did not run. It may be that the jobs ran but had problems and are not properly instrumented to log an early failure, thus appearing to not run.

I would add at least a simple logger command in front of the job, something like:

logger -p local1.notice "starting job21"