Re: [exim] limiting simaltenous connections to certain hosts

Top Page
Delete this message
Reply to this message
Author: Zbigniew Szalbot
Date:  
To: Exim Mailing List
Subject: Re: [exim] limiting simaltenous connections to certain hosts
Hi there again,

Dean Brooks said the following:
> On Thu, Aug 31, 2006 at 06:24:44PM +0100, Stephen Gran wrote:
>
>> If I understand correctly, you want to send emails to people at that
>> provider, but want to rate limit your outgoing connection numbers. I
>> would queue the incoming messages, and then deliver with -qq to send all
>> the messages down a single connection. It adds a slight delay, but not
>> as much as having connections reset, presumably.
>>
>
> I second that solution. We move a significant amount of email, so we
> maintain a separate instance of Exim to handle all outbound mail deliveries.
>
> On that server, we use startup options of "exim -bd -qq10m" to achieve
> results of optimizing deliveries down a single connection whenever possible.
>

OK. I read the docs and it really seems a very nice solution.

I did test it for a while. Here are some results. I need your help
because I am not sure how to interpret them. I sent about 20 emails from
MUA to exim on a single connection and exim protested deffering some of
them (because of the limit in exim config). Anyway, I watched what would
happen.

2006-08-31 20:08:14 J4VJH9-00020O-3X SMTP error from remote mail server
after MAIL FROM:<EMAIL_ADDRESS> SIZE=9583: host mx8.go2.pl
[193.17.41.48]: 421 mx8.go2.pl Error: too much mail from 83.19.156.210
2006-08-31 20:08:14 J4VJH9-00020O-3X == EMAIL_ADDRESS R=dnslookup
T=remote_smtp defer (-45): SMTP error from remote mail server after MAIL
FROM:<EMAIL_ADDRESS> SIZE=9583: host mx8.go2.pl [193.17.41.48]: 421
mx8.go2.pl Error: too much mail from 83.19.156.210

So it looks as if they didn't care messages were being passed with exim
-bd -qq option but then I noticed:

2006-08-31 20:08:22 J4VJHA-00020O-B7 => EMAIL_ADDRESS R=dnslookup
T=remote_smtp H=mx6.go2.pl [193.17.41.46]
2006-08-31 20:08:22 J4VJHA-00020O-B7 Completed
2006-08-31 20:08:24 J4VJH9-00020O-3X => EMAIL_ADDRESS R=dnslookup
T=remote_smtp H=mx6.go2.pl [193.17.41.46]*
2006-08-31 20:08:24 J4VJH9-00020O-3X Completed
2006-08-31 20:08:26 J4VJH9-00020O-6Z => EMAIL_ADDRESS R=dnslookup
T=remote_smtp H=mx6.go2.pl [193.17.41.46]*
2006-08-31 20:08:26 J4VJH9-00020O-6Z Completed
2006-08-31 20:08:28 J4VJH9-00020O-JH => EMAIL_ADDRESS R=dnslookup
T=remote_smtp H=mx6.go2.pl [193.17.41.46]*
2006-08-31 20:08:28 J4VJH9-00020O-JH Completed
2006-08-31 20:08:30 J4VJHA-00020O-3O => EMAIL_ADDRESS R=dnslookup
T=remote_smtp H=mx6.go2.pl [193.17.41.46]*
2006-08-31 20:08:30 J4VJHA-00020O-3O Completed

And my qeustion here is this: does this log means these 5 emails were
delivered on a single connection? From what I have read it seems, they
were not (different IDs) but maybe I am wrong.
If they were not, does this mean the -qq option did not really work out
in this case? Just trying to learn more so I appreciate all comments!
> Our outbound instance of Exim also hands off any email that does not
> deliver within an hour to a second outbound (fallback) instance of
>

Dean - can I ask you to show me how you do it (off or on list)? I mean I
am not sure how I would go about letting emails in queue in an hour an
after that time releasing them to a fallback host. I have already
configured a fallback host (thanks Pierre!).

Thanks again for your support.

Warm regards,

Zbyszek