Carsten,
i am using amavis with postfix by sending mails to amavis via
smtpd_proxy_filter.
But amavis runs in a timeout, and i don`t know why.
Until now, it appears only with one single message, but I am afraid,
that this could happen more often in the future.
Jun 19 20:02:15 my-postfix postfix/smtpd[2578]: connect from
mail.foobar.org[111.222.111.222]
Jun 19 20:02:17 my-postfix postgrey[3156]: action=pass, reason=triplet
found, client_name=mail.foobar.org,
client_address=111.222.111.222, [email protected],
[email protected]
Jun 19 20:02:17 my-postfix postfix/smtpd[2578]: NOQUEUE:
client=mail.foobar.org[111.222.111.222]
Jun 19 20:02:18 my-postfix amavis[679]: (00679-04) ESMTP::10024
/var/lib/amavis/tmp/amavis-20140619T194021-00679-ZfLFYOVy:
<[email protected]> -> <[email protected]> SIZE=9833669
Received: from mx.provider.de ([127.0.0.1]) by localhost
(my-postfix.provider.de [127.0.0.1])
(amavisd-new, port 10024) with ESMTP for <[email protected]>; Thu,
19 Jun 2014 20:02:17 +0200 (CEST)
Jun 19 20:11:02 my-postfix amavis[679]: (00679-04) (!)ESMTP ABORTING:
Connection broken during DATA:
at (eval 108) line 205, <GEN54> line 1841.
Jun 19 20:11:02 my-postfix amavis[679]: (00679-04) size: 9833669,
TIMING [total 524713 ms] -
SMTP greeting: 5 (0%)0, SMTP EHLO: 1 (0%)0, SMTP pre-MAIL: 1 (0%)0,
lookup_sql: 2 (0%)0,
SMTP pre-DATA-flush: 1015 (0%)0, rundown: 523689 (100%)100
Jun 19 20:11:02 my-postfix amavis[679]: (00679-04) (!)ESMTP: NOTICE:
ABORTING the session:
Connection broken during DATA: at (eval 108) line 205, <GEN54> line
1841.
Jun 19 20:11:02 my-postfix postfix/smtpd[2578]: lost connection after
DATA (506193 bytes) from mail.foobar.org[111.222.111.222]
Jun 19 20:11:02 my-postfix postfix/smtpd[2578]: disconnect from
mail.foobar.org[111.222.111.222]
Any suggestions?
The message size is fairly large (10 MB), although unless you have a
really ancient version
of amavis it should be able to receive SMTP data at about 10 MB/s.
(using TLS on a
connection from MTA to amavisd can drop speed to about 1 MB/s, but that
is not you case).
I hope you are using postfix option 'smtpd_proxy_options=speed_adjust'
on your
postfix smtpd service, otherwise the proxy content filter is at a mercy
of potentially very slow client - which could explain the case above.
Combined with postscreen the master.cf entry could look like:
smtp inet n - n - 1 postscreen
smtpd pass - - n - 150 smtpd
-o smtpd_proxy_filter=inet:[::1]:10024
-o smtpd_proxy_options=speed_adjust
-o smtpd_proxy_timeout=600
Mark
