[Exim] socket read time out problem?

Pàgina inicial
Delete this message
Reply to this message
Autor: Jon Loeliger
Data:  
A: exim
Assumpte: [Exim] socket read time out problem?
OK, I'm not sure what is going on here, but I have
some theories.

In the last week or so, and not before, I've started to
see a whole bunch of these messages when I send mail:

    post: problem initializing server; [BHST] read from socket timed out
    Letter saved in dead.letter


I am using MH as MUA to "post" mail, with exim as MTA:

    chrome 3218 % exim -bV
    Exim version 3.32 #1 built 01-Aug-2001 13:35:57
    Copyright (c) University of Cambridge 2001


All on a FreeBSD system:

    chrome 3220 % uname -a
    FreeBSD chrome.jdl.com 4.3-RELEASE FreeBSD 4.3-RELEASE #0: Sat Apr 21
    10:54:49 GMT 2001 jkh@???:/usr/src/sys/compile/GENERIC i386


Now, when I see these problems, I semi-poll my 'netstat -a' info
and often see several lines like this:

Active Internet connections (including servers)
Proto Recv-Q Send-Q  Local Address          Foreign Address        (state)
tcp4       0      0  chrome.smtp            abv-sfo1-acmta4..38118 TIME_WAIT
tcp4       0      0  chrome.smtp            200-171-103-232..1775  TIME_WAIT


and this:

tcp46      0      0  *.smtp                 *.*                    LISTEN
tcp4       0      0  *.smtp                 *.*                    LISTEN


And the process from "ps auxwww" posting the mail is:

    USER     PID %CPU %MEM   VSZ  RSS  TT  STAT STARTED      TIME COMMAND
    jdl    44617  0.0  2.5  1076  684  p0  I     1:58PM   0:00.03 post -library /home/jdl/.mh -noverbose -alias ~jdl/.mhrc /home/jdl/.mh/drafts/18


And the running exim is this:

    exim   44618  0.0  3.2  3232  896  ??  I     1:58PM   0:00.03 /usr/local/sbin/exim -bd -q30m



My Theory:

    I have just exceeded some threshold somewhere.  In particular,
    I suspect that I've just finally added one too many IP addresses
    to my host_reject_recipents flat file, and it now takes just
    a bit too much time to search it on each and every connection.
    Thus, each exim process running now hangs out just a bit longer,
    and the main exim daemon is not willing to fork off another exim
    to handle my sudden demand to post a new piece of mail.  The net
    effect is that "the server isn't responding" (in time).
    A contributing factor here to timing could be my use of several
    RBLs for anti-spam checking.


I haven't been able to find this problem or message in the
exim FAQ (yet?), probably because it is actually MH's "post"
executable that is complaining, not exim. exim just appears
to be not responding.

Do you think I will be able to solve this issue by having exim
be willing to fork off more processes? And if so, how do is
there a config value I should double somewhere? Can I get
pointed towards the right section of the manual?

Thanks,
jdl