Thanks for your reply, Nigel. Much appreciated.
On 11/01/05, Nigel Metheringham (Nigel.Metheringham@???) wrote:
> On Tue, 2005-01-11 at 10:24 +0000, Rory Campbell-Lange wrote:
> > The single issue that doesn't make any sense to me is that when I stop
> > exim (/etc/init.d/exim4 stop), there are still running exim4 processes 3
> > seconds later.
>
> Although I don't know exactly how debian does its startup/shutdown
> scripts, I would expect "/etc/init.d/exim4 stop" to stop the running
> daemon. The other exim processes - ie those handling messages (in
> particular delivery processes which can be quite long lived if connected
> to slow remote sites), queue runners etc will still be running.
This make sense. The shutdown script just looks for the pidfile and then
shuts down the associated process.
> 3 seconds sounds like a very small amount of time for processes involved
> in complex remote operations and database lookups (ie DNS) to complete.
I suppose I should wait for something like the time interval set in
"smtp_receive_timeout"? Or maybe "callout_domain_negative_expire"?
This thread began because my server was experiencing too many
connections. I get lots and lots of outgoing connections hanging in the
following state shown below (CLOSE_WAIT).
(lsof -i output)
exim4 6304 Debian-exim 0u IPv4 655292 TCP myserver:smtp->mail2.myisp.net:49602 (CLOSE_WAIT)
exim4 6304 Debian-exim 2u IPv4 655292 TCP myserver:smtp->mail2.myisp.net:49602 (CLOSE_WAIT)
exim4 6304 Debian-exim 3u IPv4 655313 UDP myserver:35888->ns1.myisp.net:domain
Does this look like a dns problem? Manual, non-cached lookups seem to complete
very rapidly.
Thanks
Rory
--
Rory Campbell-Lange
<rory@???>
<
www.campbell-lange.net>
