Hi

Thanks for the reply.

The basic setup is as follows.

We have incoming SMTP servers running Exim 4.30 with the latest ExiScan
patch. Mail is then palmed off to the spamd run on another server. Mail
is then scanned and a report is supplied back to the Exim MTA. The only
monitoring I am doing is a ps check counting the number of spamd procs
that are running, if there is 0 = problem or >251 = problem. I don't
think that is the prob seeing that when I did used to check the port, it
would crash with an error such as this. The spamd server(s) is also
behind a firewall and the only connection made to it is from the
incoming smtp servers. 

Feb 25 17:30:21 01-spam spamd[24273]: Use of uninitialized value in
subroutine entry at
/usr/lib/perl5/5.8.1/i386-linux-thread-multi/Socket.pm
line 370.


I can send more of the strace if you would like as I have about 11GB of
it :) I can aslo include debug info for the spamd process. I did review
the debug output from the spamd proc when it died and it looks likes
like it would normally complete a message, just the spamd dies so it
never spawns any more probs... Ill keep looking.


This particularly happens when I enable health monitoring from a Radware
Load Balancer Switch. What can be done about spamd dying when a simple
nmap or telnet to port 783 takes place.

Thanks


--

Regards

Richard Mayhew
Unix Specialist

MWEB Business
Tel:  + 27 11 340 7200
Fax:  + 27 11 340 7288
Website: www.mwebbusiness.co.za

-----Original Message-----
From: Nick Leverton [mailto:[EMAIL PROTECTED] 
Sent: 25 March 2004 02:53 PM
To: Richard Mayhew
Cc: [EMAIL PROTECTED]
Subject: Re: SPAMD Keeps Dying - UPDATE

On Thu, Mar 25, 2004 at 02:29:54PM +0200, Richard Mayhew wrote:
> Hi All,
> 
> I seem to have narrowed the problem down to this. I did a strace on
the
> main spamd process and watched it for sometime.
> Just before the spamd dies, I get this output. Bearing in mind nothing
> is logged when spamd is in debug mode.
> 
> Going through it as best I can see it seems to be referring to a
LOCALE
> file which does not exist.
> What seems strange is that its happening on both systems, both Linux
and
> FreeBSD. I can't see why these files would be missing.
> 
> Can anyone shed some light on to this?

I haven't kept the earlier messages in the thread, but I suspect the
locale stuff is a bit of a red herring.  It's trying to find appropriate
error messages, but should still fall back to the standard ones.

The trace seems to show a connection that closed early, which it's
trying to report.  Is it possible something is making connections to
spamd's port and then exiting without doing a scan ?  That is known to
cause spamd to die.  It may be innocent and inadvertent (perhaps you
have
something monitoring it), or it may be down to an nmap or similar scan.
You could tcpdump connections to port 783 to check this out.

Nick

Reply via email to