OSEC

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
 
Subject: timeout after DATA
From: Rexes Villa (rvillaskyinet.net)
Date: Tue Aug 08 2000 - 19:46:15 CDT

EHLO

        I have some weird observation with the mail system that i'm
        currently taking care of. I am running postfix and have these
        messages in my logs:

server 0(mx):
Aug 9 05:59:57 mx postfix/smtp[8721]: 439543BE37: to=<usermydomain.com>,
relay=pop.mydomain.com[ip.ip.ip.ip], delay=144262, status=deferred (conversation
with popx3.mydomain.com[ip.ip.ip.ip] timed out while sending message body)

server 1(pop):
Aug 9 04:14:46 pop3 postfix/smtpd[18332]: timeout after DATA from
mx.mydomain.com[ip.ip.ip.ip]

        my system have this kind of setup
                              ----------
                             | INTERNET |
                              ----------
                                  |
                                  |
                              __________
                             ( mx )
                              ----------
                                  |
                                  |
                              ----------
                             | pop |
                              ----------

        so mails coming from different sending mail server would first pass
        throug the machine called mx. the machine mx would then send it to
        the server with the mailboxes of the recipients on the machine called
        pop. both mx and pop are running postfix.

        over the weekend i noticed a few queued messages on mx, which normally
        doesn't have. and when i do a mailq on mx, the queue listing would
        have something like
        (conversation with pop.mydomain.com[ip.ip.ip.ip] timed out while sending
message body)

        I know there were a few posting like this a weeks ago, but still
        i would like to share my case since the previous discussion involved
        something about 'IP MTU path discovery' (am i right?) and would
        require something on the router, but those two servers that i got
        are within the same local network... will those MTU stuff still be
        responsible on this case??

        A few more points that may help you help me =):

        1) Not all messages on mx would end up queued due to the "timeouts"
        2) the server pop got a queue of almost 4000 request, most of them
        due to "Over quota"
        3) mx got a maximum number of smtpd processes set to 75 and 200
        smtp process. pop got 200 for smtpd and 75 for smtp.

        Anyone got any idea? 

-- 
----------------------------------------------------------------------------
Rexes P. Villa						  rvillaskyinet.net
----------------------------------------------------------------------------
"when you have eliminated the impossible, whatever remains, however 
   improbable, must be the truth."