Re: Sendmail 30 second delay?

[Jonathan Benson <sysadmin ocean com au>]

OK a combined followup here:

- I can tell IDENTD is doing it's thing by the fact the user I'm running telnet
as on sotong.ocean.com.au is showing up in the logs.

-  Unless I'm mistaken the default timeout is 5s not 30s and I've uncommented
the timeout in sendmail.cf (yes and restarted sendmail) in case that was it.
It's NOT IDENTD at fault.

- I normally have identd disabled on the majority of my hosts but have enabled
it for this testing.

-  I've left debugging on and a brief look at the logs shows timestamps that
vary by 1 to 5 seconds which I'd consider normal.

- I have no test case in terms of a message.  The problem is in the initial
connection from a remote host.


Conclusions.

- It may have been a DNS issue.  A reverse DNS lookup on the IP currently
resolves in under a second and at the moment I can't reproduce the problem.

- I can't be bothered risking an alert at 3am just to do further tests if this
reoccurs so I'm just going to leave the timeout on my monitor at 45 seconds and
only bother with it if I notice the problem for other reasons.  :)  It's only a
mail server used for mailing lists and is rather low volume.  I've NEVER seen
the problem with our primary or secondary mail servers.


Thanks to all

Jon

Jonathan Benson wrote:

> Sendmail seems to be taking 30 seconds after a connection to launch the child
> process.
>
> It is NOT Identd as can be seen from the logs with debugging pumped right up:
>  May 31 11:46:17 pingu sendmail[24036]: NOQUEUE: connect from
> IDENT:user sotong ocean com au [203.12.234.9]
>  May 31 11:46:47 pingu sendmail[24036]: SASL: available mech=PLAIN LOGIN
> ANONYMOUS, allowed mech=GSSAPI KERBEROS_V4 DIGEST-MD5 CRAM-MD5
>
> The real pain with debugging this is sometimes things work fine.  Especially
> if a connection has just been made:
> May 31 11:50:22 pingu sendmail[24060]: NOQUEUE: connect from
> IDENT:user sotong ocean com au [203.12.234.9]
> May 31 11:50:22 pingu sendmail[24060]: SASL: available mech=PLAIN LOGIN
> ANONYMOUS, allowed mech=GSSAPI KERBEROS_V4 DIGEST-MD5 CRAM-MD5
>
> ANY ideas and input appreciated.  I monitor my mail servers and this delay is
> causing the monitor to trigger too often.  For now I'm going to increase the
> timeout on the monitor but I'd prefer to fix the problem with sendmail.
>
> Thanks
>
> Jon
>
> --
> Jonathan Benson
> Systems Administrator
> Ocean Internet
> http://www.ocean.com.au/
>
> _______________________________________________
> Seawolf-list mailing list
> Seawolf-list redhat com
> https://listman.redhat.com/mailman/listinfo/seawolf-list

--
Jonathan Benson
Systems Administrator
Ocean Internet
http://www.ocean.com.au/