Cron goes mad

[baumi]

Hello

Thought that I've seen a description of this problem before, but I was not able to find it, so...

We're experiencing a strange behaviour of the cron daemon on several of our AIX systems. Cron runs ok for a random amount of time (weeks, months) and suddenly stops to execute cron jobs. The process ist still there, no childs associated. In this state, it doesn't consume a noticeable amount of cpu cycles.

When I try to kill cron (Signals 1 or 9), cpu consumption goes up to 100%, but the process cannot be killed. Only a reboot helps.

It's really annoying to have to reboot systems that are otherwise extremely stable because of such a 'minor' problem.

The problem doesn't seem to be hardware related. I've seen it on 7043-150, 7025-F30, 7025-F50, 7026-H80 systems with different applications running on them. The crontabs are quite different. All systems have AIX 4.3.3 installed.

Any help will be greatly appreciated.

- Urs

 

[DSMARWAY]

hi,

What maintenance level have you got installed ?

are there any errors listed in /var/adm/cron/log

something like any max processes reached by cron

Are there any errors reported in mailx ?


what jobs do you run from cron , are your entries correct?


when you kill cron are there any jobs that cron is running or will run ?

HTH

 

[DSMARWAY]

hi,

What maintenance level have you got installed ?

are there any errors listed in /var/adm/cron/log

something like any max processes reached by cron

Are there any errors reported in mailx ?


what jobs do you run from cron , are your entries correct?


when you kill cron are there any jobs that cron is running or will run ?

HTH

 

[baumi]

Hi,

No, there is nothing unusual in /var/adm/cron/log. Just the standard messages like

! *** cron started *** pid = 4968 Mi 1 Okt 13:25:12 2003

Nothing in mailx, too.

The installed maintenance level on most of the problematic machines is 4330-10.

Right now, I have to reboot a machine because of a 'mad cron' :-( Is there anything useful (well, besides the normal things like process list etc.) I could check while cron is consuming 100% CPU?

- Urs

 

[DSMARWAY]

hi,

You can check , what programs you are running from cron.

Check you are not running any scripts that do big searches ,
cp large files etc..
if possible i.e. you don't have important jobs running in cron , comment all entries and then try stopping and restarting cron . Check if CPU usage is high.


if cron stops check under /var/locks there is a LCK..cron file with the process id of cron , check if this is correct
there is a link /etc/locks to /var/locks , check if this is valid , if not create a link.


you can use pa aux | head -10 ( see top ten processes on system, consuming alot of CPU time )

HTH