At 12:11 +0100 10/20/2001, Philip Hazel wrote:
>On Fri, 19 Oct 2001, John W Baxter wrote:
>
>> At 13:38 +0100 10/19/2001, Philip Ross wrote:
>> >Is there any way to configure how long the period of time exim will
>>hold off
>> >from using a particular mx?
>>
>> It's not really predictable that the same MX will do the same thing again,
>> so it's hard to sharpshoot before the fact.
>>
>> See 53.7 in the spec for exim_fixdb. Or use the blunderbuss approach:
>> toss out all the files in the db directory in the spool directory. Exim
>> will do a little extra work for a while.
>
>Am I missing something here? My answer to the original question would be
>"read about retry rules". If you set, for example,
>
>some.host.name * * F,100d,1m
>
>Exim will only hold off for 1 minute for that host, and won't time out
>till 100 days (for domains routed to that host).
No, you aren't missing anything, but I certainly did. I think. The thread
started with a situation in which host A and host B were MXs for some
address, with A being more favored, and A was unresponsive for a short time
and Exim was using B for additional messages to the address because A's
retry time hadn't been reached.
Your answer implies that the retry hints are stored such that a change in
the retry rules for a host takes place right away (the storage is of the
most recent time tried, not the next time to try). That makes sense: it's
more flexible than storing the next time to try the host. [Yes, I should
and will look that the available documents and at the retry database
itself.]
Thanks!
--John
--
John Baxter jwblist@??? Port Ludlow, WA, USA
