cron goes mad?

Tom Mitchell mitch48 at sbcglobal.net
Fri Jan 16 10:01:03 UTC 2004


On Thu, Jan 15, 2004 at 08:41:01AM -0500, Dave Alden wrote:

> 
> Hi,
>   I've got the exact same problem.  :-(  The load on this workstation is
> up to 478.  :-)  I received no mail fron cron and there is nothing out
> of the ordinary in any of the logfiles in /var/log...
....
> did a /etc/init.d/crond restart).  I have a process that runs every
> 5 minutes and it's last entry was at 16:40:00 in /var/log/cron.

If you start a process every 5 min that takes 5 min and 2 seconds you
will eventually run out of something....  It takes a little more than
12 hours for you to have two running all the time.  then 25 hours
later you have three....

Since the system will get busy in the small hours (night) things that
finish just fine in the day could break at night.  Make sure you pay
attention to how long things take 7x24 for things like this.  Always
watch for exceptions....

Your launcher might check the load average (parse output of 'w')
and not launch if the system is too loaded....

It is possible to "time" processes with a wrapper. When they take too
long, log or send a message.

Runaway systems are hard to debug.... One trick is to run "script"
when lodged into the trouble machine (over the net) and run shell
scripts that loop on things like a full "ps" listing, "lsof", "df -a",
"w", "netstat -a".

The /proc file-system has lots of info...

Also make sure that these are not zombie process for which some
script/component forgot to do a "wait".

Do not send all output and errors to /dev/null when debugging.


-- 
	T o m  M i t c h e l l 
	mitch48-at-sbcglobal-dot-net





More information about the fedora-list mailing list