Determining Cause Of Defunct Processes

[vickersdc]

First things first...

I understand WHY defunct processes are caused, I know what the effect of them is, I understand HOW they are made, but

How do I find out what is causing them? I have no core files to check as none are being created, but I basically have a system that spirals into oblivion creating defunct processes until it can create no more.

As far as "ps av" is concerned there are no processes hogging either CPU or memory. vmstat shows run queue figures in excess of 30 !!! with us+sy at 100% (which would tend to indicate a looping process - but thats not what the ps command is telling me).

Can it be that the system was heavily loaded before (it is a fairly hard run system) to such an extent that the kernel started to kill off processes and in doing so caused the error I now have?

I've also reduced both minperm and maxperm.

Any ideas???

 

[fcail]

Hello,

You can try this command line :
ps -ef | sort -r -n +3 | head -n 15

The lines are order by resources of CPU taken.

Regards.

Fcail

 

[vickersdc]

So the outcome appears to be a job that was run by cron every 5 minutes or so - hence the reason that it seemed to not appear with the ps command so far.

I'm still slightly baffled as to why it was running, or rather why it's suddenly decided to mess up, so now the task of ascertaining that begins.

Thanks...