"startup" of smtp connection timeout

Thread View: comp.mail.sendmail

4 messages

[T] Back to Threads [V] Tree View [G] Group Articles

4 total messages Started by Jeremy Fitz Wed, 12 Feb 2003 03:42

#3596

Author: Jeremy Fitz
Date: Wed, 12 Feb 2003 03:42

22 lines
2442 bytes

Hi:

I have a microsoft (win2k) box running ISA server (firewall).  When I telnet
from this box to a specific unix-based sendmail box (port 25), the unix box
shows the connection sitting in startup phase... until I kill the telnet
session.  This only happens with a minority of mail servers - for the most
part my telnet session usually connects immediately, and I have no problems
sending mail.  NB. I am using telnet for troubleshooting to replace the SMTP
(exchange server) that is having problems sending mail.

Under UNIX,  ps -elf   shows something like
sendmail: startup with aa.bb.cc.dd

In order to debug this problem, I'd like to understand what sendmail is
doing during "startup".  Can anyone point me to information on this topic -
or more importantly what can cause the "startup" phase to hang like this. I
am sure this has something to do with the particulars of the microsoft box
(DNS? - not the unix box) as other hosts have no problems connecting to the
unix box in question.

Thoughts?
Thanks

Re: "startup" of smtp connection timeout

#3597

Author: Neil W Rickert
Date: Wed, 12 Feb 2003 04:39

22 lines
1889 bytes

"Jeremy Fitz" <jeremyfitz@hotmail.com> writes:

>In order to debug this problem, I'd like to understand what sendmail is
>doing during "startup".

Basically, it is doing:

  gethostbyaddr(ip-address)		lookup the hostname
  gethostbyname(hostname)		cross check hostname to see if forged
  identd callback
  run the check_relay ruleset

The identd callback should take no longer than 30 seconds on sendmail-8.9.3
or earlier, and 5 seconds on 8.10.0 and later.

The most likely problems are DNS lookup delays for gethostbyaddr()
or the subsequent gethostbyname(), or DNS lookups done in
the check_relay ruleset (mostly dnsbl lookups).

Maybe there can be a delay if your aliases map needs rebuilding.
Run "newaliases" just in case this is the problem.  >doing during
"startup".  Can anyone point me to information on this topic -

Re: "startup" of smtp connection timeout

#3598

Author: Jeremy Fitz
Date: Wed, 12 Feb 2003 19:28

41 lines
2818 bytes

Neil:

Is the gethostbyaddr executed via a reverse DNS lookup?
- what is done with the result?
- what happens if the reverse lookup doesn't return anything (isn't
populated for my host), or returns the incorrect result?
- could this account for delayed or non-successful "startup"?

How is a "forged" address detected via the gethostbyname() call?

Your help is appreciated.

J


"Neil W Rickert" <rickert+nn@cs.niu.edu> wrote in message
news:b2cj6d$n2c$1@husk.cso.niu.edu...
> "Jeremy Fitz" <jeremyfitz@hotmail.com> writes:
>
> >In order to debug this problem, I'd like to understand what sendmail is
> >doing during "startup".
>
> Basically, it is doing:
>
>   gethostbyaddr(ip-address) lookup the hostname
>   gethostbyname(hostname) cross check hostname to see if forged
>   identd callback
>   run the check_relay ruleset
>
> The identd callback should take no longer than 30 seconds on
sendmail-8.9.3
> or earlier, and 5 seconds on 8.10.0 and later.
>
> The most likely problems are DNS lookup delays for gethostbyaddr()
> or the subsequent gethostbyname(), or DNS lookups done in
> the check_relay ruleset (mostly dnsbl lookups).
>
> Maybe there can be a delay if your aliases map needs rebuilding.
> Run "newaliases" just in case this is the problem.  >doing during
> "startup".  Can anyone point me to information on this topic -
>

Re: "startup" of smtp connection timeout

#3599

Author: Neil W Rickert
Date: Thu, 13 Feb 2003 00:01

26 lines
1947 bytes

"Jeremy Fitz" <jeremyfitz@hotmail.com> writes:

>Is the gethostbyaddr executed via a reverse DNS lookup?

Yes.

>- what is done with the result?

It is used to set ${client_name}

>- what happens if the reverse lookup doesn't return anything (isn't
>populated for my host), or returns the incorrect result?

The failure is indicated in ${client_resolve}, and ${client_name} is
set to [ip-address]

>- could this account for delayed or non-successful "startup"?

The time taken for the lookup is the problem.  Failed lookups often
take longer than successful lookups (retries at several
nameservers).

>How is a "forged" address detected via the gethostbyname() call?

The ip addresses from gethostbyname should include the original ip
address.  Otherwise DNS spoofing might be in use.

Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads