The cron daemon on one of my HP boxes is giving the following error:
I checked the queuedefs man page, and it has left me a bit confused. The man page says:
So this means that no more than 100 cronjobs can be run at one time (we do not use at and batch on this system).
The /var/adm/cron/queuedefs file does not have a line for cron, so it should be picking up the default value. But in this post, Perderabo says that the default value is also 100. If that is the case, have I already reached the maximum? And if so, do I have any choice, other than having a cronjob that restarts cron every day or so?
-EDIT:
Here's the uname -a o/p:
HP-UX xxxxx B.11.11 U 9000/800
Restarting cron every day or so isn't going to help - cron will just start getting up to the max again - it's all about scheduling and making sure your scripts end. On that other post I mentioned that I had never seen the issue (back in 2002) - well, the same DBA caused it to happen, opened up a Sev 1 ticket blaming the OS for his lousy code which never died - he ran the same script every minute without checking to see if the last one finished - he easily got up to 100 jobs running and the "c queue max run limit reached" message starting popping up.
You need to look at the scripts and scheduling of the cron jobs - make sure that there is a clean exit so you don't build up to 100 jobs. The system DOES reschedule - once the max of 100 jobs goes down, it will start running all the jobs waiting (unless it hits 100 again). It's actually a pretty good way of doing in in my opinion. Keeps my favorite DBA from killing the system every day!
Look up the pid of the cron process. You will have 100 processes with that number as the ppid. This not a matter of a clean exit. They are still running.
Or maybe you have too much stuff in another queue? Notice the language on "man queuedefs": "the total number of jobs that can be run on all the queues is limited to 100". I have never really known whether or not to believe that sentence.
Thanks for the inputs. We did move a few cronjobs around, and I added a line in the queuedefs file: c.100j2n60w. One question though: does a script count as one 'job'? So if I have 5 find commands running through cron and I put them in a single script, I will have reduced the number of jobs by 4 right?
cron is waiting for its child process to finish. It doesn't matter what that child does including running other processes. Want to cheat? Write a cron script that launches a second script in the background and immediately exits.
#! /usr/bin/ksh
/path/to/real/script &
exit 0
I have never tried this, but I'm guessing it will fake cron out.