Re: [Exim] lowest MX record - strange situation

Top Page
Delete this message
Reply to this message
Author: Dmitry Alyabyev
Date:  
To: exim-users
Subject: Re: [Exim] lowest MX record - strange situation
Hello

At first, thank you Philip for attempts to help.

On 12 Jul 2000 13:15:50 +0300, Philip Hazel wrote:

>What is the point of having the same host at different priorities?


MTA will try to deliver the message one more time to the same host.
It can be useful if final relay peaks for short time.

>> >> 123.com preference = 30, mail exchanger = relay2.mydomain.com
>> >> 123.com preference = 20, mail exchanger = mail.123.com
>> >> 123.com preference = 10, mail exchanger = mail.123.com
>> >> 123.com preference = 20, mail exchanger = relay.mydomain.com
>
>If you take this list at face value, it says
>
> "Try mail.123.com; if that is down, try mail.123.com and
> relay.mydomain.com in either order; if they are both down, try
> relay2.mydomain.com."


Yes, it is.
But I think if all lowest relays are down and the Exim-host sees
itself as "first-alive-relay" it have to try deliver to lowest MX
relays time to time by the MX-list.

>That is silly. Why try mail.123.com twice?


Please see above.

>> IFAIK that is normal behavior when you put one host several times with
>> different preference and have several hosts with the same pref. as
>> well.
>
>No, that is not normal practice. One host usually appears only once.


It's not the usual but it's not wrong configuration, isn't it ?

>It should not be a problem with Exim. Exim just optimizes by skipping
>the higher preference, because it has already tried that host.


There is the essential difference with my point of view.
Could you explain me why Exim will try to deliver the message by other
way instead to try deliver it to died relays, stops to do it after
retry-timeout and send error message to sender ?

>Please understand: I do not think this is the cause of your problem. I
>just pointed it out as something that was unusual. Of course, it
>*might* be the cause - it is a rare situation, so the code in Exim is
>not often tested, and any code may contain bugs. If it really is the
>cause, then Exim needs to be fixed, but as I said, I don't think it is
>very likely to be.
>
>However, when I get time I will set up a test like this just to make
>sure Exim is working as it should be.


Thank you one more time, Plilip - Exim really good and powerful MTA.
I'm just trying to understand who's wrong - I or exim :-)

--
Dimitry