[rt-users] EX_TEMPFAIL / receiving multiple copies of mail

[bill at daze.net]

Hello,

Sorry this is a bit long.  The basic issue is I believe there may be a bug
somewhere since we got multiple copies of messages today after bringing RT
back up from a FastCGI lost communications error.  RT correctly issued
EX_TEMPFAIL responses, but when RT came back up, we received one copy of
the message for each EX_TEMPFAIL response that was sent and one copy for
the successful send.

Some Background:

We are running RT3.0.0 (built from source) with Apache and FastCGI
(installed from FreeBSD ports).  This is our first time using FastCGI and
we are encountering some lost communications issues with it.  A couple of
time over the past three days, RT has quit responding and the only
solution is to restart apache.  After this happened the first time, I
added "-idle-timeout 90 -processes 5" to my FastCGIServer directive for RT
to increase the timeout (it's possible that it may take longer than 30
seconds to send an e-mail) and increased the processes to hopefully
improve access for multiple users.  Last night it stopped responding
again, so I increased the timeout to 300 as a bandaid.  I know this won't
correct the problem, it will just make our users wait 5 minutes before
figuring out there is a FastCGI problem.  The FastCGI problem is something
I need to research, however if anyone has insight about this, please feel
free to reply.

Typical FastCGI error in the apache error log:

[Wed Apr  2 08:52:53 2003] [error] [client 130.94.96.248] FastCGI: comm
with server "/usr/local/rt3/bin/mason_handler.fcgi" aborted: idle timeout (90 sec)

Here are the maillog entries (stripped of identifying information to
protect the privacy of the innocent) of one of the messages.  We received
this message 4 times.  Note there were 3 EX_TEMPFAIL's and one successful
send.

% grep "h32GAWdo015894" /var/log/maillog

Apr  2 08:10:35 rt3 sm-mta[15894]: h32GAWdo015894: from=<>, size=1367,
class=0, nrcpts=1, msgid=<200304021625.h32GPCO99599@>, proto=ESMTP,
daemon=MTA, relay=<>

Apr  2 08:12:06 rt3 sm-mta[15895]: h32GAWdo015894:
to="|/usr/local/rt3/bin/rt-mailgate --queue queue-name --action correspond
--url http://www.example.com/",  ctladdr=<> (26/0), delay=00:01:34,
xdelay=00:01:31, mailer=prog, pri=31621, dsn=4.0.0, stat=Deferred: prog
mailer (/bin/sh) exited with EX_TEMPFAIL

Apr  2 08:13:44 rt3 sm-mta[15900]: h32GAWdo015894:
to="|/usr/local/rt3/bin/rt-mailgate --queue queue-name --action correspond
--url http://www.example.com/", ctladdr=<> (26/0), delay=00:03:12,
xdelay=00:01:31, mailer=prog, pri=121621, dsn=4.0.0, stat=Deferred: prog
mailer (/bin/sh) exited with EX_TEMPFAIL

Apr  2 08:46:47 rt3 sm-mta[16006]: h32GAWdo015894:
to="|/usr/local/rt3/bin/rt-mailgate --queue queue-name --action correspond
--url http://www.example.com/", ctladdr=<> (26/0), delay=00:36:15,
xdelay=00:01:33, mailer=prog, pri=211621, dsn=4.0.0, stat=Deferred: prog
mailer (/bin/sh) exited with EX_TEMPFAIL

Apr  2 09:13:01 rt3 sm-mta[16281]: h32GAWdo015894:
to="|/usr/local/rt3/bin/rt-mailgate --queue queue-name --action correspond
--url http://www.example.com/", ctladdr=<> (26/0), delay=01:02:29,
xdelay=00:00:03, mailer=prog, pri=301621, dsn=2.0.0, stat=Sent



--
"If Verisign is the 'Value of Trust',
  then trust ain't worth much any more." - me

Spam Trap Mail Key: ASK and you shall receive