Zombie Trouble Shooting

I have a perl script that becomes a zombie.

It runs fine for days or weeks checking for new data every 60 seconds.
But after a long period of time running on Red Hat 9, Fedora Core 4,
and now Fedora Core 5 it remains in memory as an active process and
maintains its network tcp/ip connections but stops functioning.

Here you see two zombies that stopped functioning days after start
and still in the process queue after several weeks.:
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
kp2a 2866 0.0 0.9 14032 9556 ? S Mar27 0:13 /usr/bin/perl -w ./packet.pl aa2mf
kp2a 2868 0.0 0.9 14024 9456 ? S Mar27 0:05 /usr/bin/perl -w ./packet.pl k5utd

I have tried for several years to find the problem as you can see
by the sequence of OS upgrades! No luck! I included a log
so I know the last good check point which is at the bottom of
an infinite loop:

Mar 29 08:56:45.327 ./packet.pl line 259:
end while remote

I have also included logic to reconnect and restart the remote
network processes should the remote end stop responding.

Can anyone recommend any debugging tools or techniques?

I do not think that perldebug would be appropriate.
tcpdump might help if I arrange for a rotating log
to keep only the last few minutes of traffic.

I might be able to use, I guess, a daily kill SIGHUP
to keep restarting them or starting them fresh every
minute with cron but I would like to solve the mystery!

Thanks.


.

Relevant Pages

  • Re: mIRC Zombie, port 445
    ... "- psexec.exe seems to be a remote tool...unknown..." ... psexec allows you to run commands on a Windows server remotely: ... | I have observed a zombie/trojan on a zombie IRC network that apparently ... | infects vulnerable computers through port 445. ...
    (Incidents)
  • Re: Strange Attack On A Webserver I Work On
    ... they use the webspace of a compromised server to host a remote reverse shell ... They will install some sort of IRC ... or IRC script and attach it to some IRC channel as a zombie. ...
    (Focus-Linux)
  • REPOST: RH9 vs RH8 Zombies
    ... I have had an inordinate number of Zombie Processes on Red Hat 9.0 Vs Red ... I've had to restart the LINKSYSMON daemon almost ... cron.daily work fired of and completed error free....) ...
    (RedHat)
  • RH 8.0 Vs RH9.0 Zombies
    ... I have had an inordinate number of Zombie Processes on Red Hat 9.0 Vs Red ... I've had to restart the LINKSYSMON daemon almost ... cron.daily work fired of and completed error free....) ...
    (RedHat)
  • Re: cron zombie (again?)
    ... > a restart of crond removes the zombie ... is that the parent needs to call wait, ... happening since crond is in a stopped state ...
    (alt.os.linux.suse)