I am just wondering... we have a script that once in a while will hang, we have went through every line of the code, which has like 30k lines in all, it is a huge program. I had every loop of every type do a kill if it runs wild after like 100k loops, since there is nothing that should get that many loops...

Anyhow, I have tried every thing a user would do to make it hang myself and am not able to duplicate anything they do.

Once one hangs a bunch will start to hang, but even then I cannot get one to hang, so I am figuring it is a script or something that someone is running, maybe a spider attempting to index it or something. What I do in shell is run this command:

ps auxww | egrep [m]ember.cgi | awk '{print $2}' | while read PID; do 
+kill -9 $PID; done
[download]

the name of the file listed in 'top' under the 'command' section is the name of the file: member.cgi

So, that is what that shell command does, it gets all those commands and kills them.

I want to set that on a cron job that runs every so often to kill any of those processes. I ran it while I had 10 people load the website and since it is only a redirect script, not one of them was interrupted by it. So I am confident that it will not hurt any visitors.

What I did was create a script for root to run:

#!/usr/bin/perl

print "Content-type: text/plain\n\n";

`ps auxww | egrep [m]ember.cgi | awk '{print $2}' | while read PID; do
+ kill -9 $PID; done`;

print "done\n";
exit;
[download]

however, when I run it from shell:

perl root_cron.cgi
[download]

I get this error: "kill -n signum | -sigspec pid | jobspec ... or kill -l sigspec"

I only get that only if there are member.cgi scripts hung, if none are hung I only get the output: done

Is there a way with Perl to do this?

I am currently re-doing how the redirect script will work so I will not need this for long, I should have the whole thing re-programmed in the next 3 to 6 months, so I do need it for now.

I would really appreciate any thing you can point me to.

Thanks,

Richard