Issue's with socketmap (smmapd) under heavy load?
Andrzej Filip
anfi at priv.onet.pl
Sun Jul 18 02:31:17 EDT 2004
Lenny wrote:
> I did 2 things to try and see if this was ident related. First, I
> installed an
> identd server on each frontend and backend.
>
> That had a huge affect on my pop login times, by the way, although I'm
> unsure
> why (for the better, that is). Unfortunately, once there were at least 100
> sendmails going mail delivery still took over 30 seconds per message.
>
> The second thing I tried was out of the sendmail.org faq. I added the
> following
> to my .mc file and remade sendmail.cf:
>
> define(`confTO_IDENT',`0s')dnl
>
> This doesn't seem to have had any affect either. :( All strace's still show
> waiting for the socketmap socket. Is there any further debugging I could do
> besides an strace that might help point me in the right direction?
Have you checked if your sendmail is compiled with TCPWRAPPERS ?
[ RedHat does it ]
confTO_IDENT controls sendmail ident query, liwrap ident queries are unafected
by the setting.
BTW you cyrus may hit limit on number of open files desciptors [open files
(-n) 1024], sendmail keeps one "socket map" session opened per per one active
SMTP session. Try to increase limit of open file descriptors to 2048 and check
if it makes the difference.
> Quoting Andrzej Filip <anfi at priv.onet.pl>:
>> 30s delay make ident query the primary suspect.
>>
>> 1) Take a look at the link below
>> http://www.sendmail.org/faq/section3.html#3.12
>> Q3.12 -- Why do connections to the SMTP port take such a long time?
>>
>> 2) Check if you sendmail is compiled with TCPWRAPPERS
>> echo "" | sendmail -bt -d0.1 | grep TCP
>>
>> TCPWRAPPERS uses external library which does its own ident lookups.
--
Andrzej [en:Andrew] Adam Filip anfi at priv.onet.pl anfi at xl.wp.pl
http://anfi.homeunix.net/ http://anfi.homeunix.net/sendmail/#cyrus
---
Cyrus Home Page: http://asg.web.cmu.edu/cyrus
Cyrus Wiki/FAQ: http://cyruswiki.andrew.cmu.edu
List Archives/Info: http://asg.web.cmu.edu/cyrus/mailing-list.html
More information about the Info-cyrus
mailing list