Re: [Exim] Long time down MTAs

トップ ページ
このメッセージを削除
このメッセージに返信
著者: Philip Hazel
日付:  
To: Martin Stewart
CC: exim-users
題目: Re: [Exim] Long time down MTAs
On Wed, 4 Aug 1999, Martin Stewart wrote:

> How does Exim know that the remote MTA has come back up.
> I guess that a successful delivery will flag the MTA as available
> again.


Yup.

> However if the exim system it secondary MX for the MTA,
> once the MTA comes back up no further mail will arrive. Does the
> "flag" that says an MTA is unavailable expire should no
> unsuccessful delivery attempts be made.


Nope. Except that really old data should get removed by regular runs of
exim_tidydb.

But does it matter? It isn't in fact a flag saying the MTA is available.
It is a time before which it isn't sensible to try that host again.
Suppose nothing happens for a long time. If you look at the retry data
using exim_dumpdb or exinext, you will see all three times ("first
failed", "last tried", "when next to try") are long ago. So if another
message does arrive, it will attempt delivery. Of course, if that should
fail Exim will think the MTA has been down all the time... Hmm. I can
see that that might not be ideal for a secondary MX.

I suppose one could have some kind of threshold to reset the retry data
if a host hasn't actually been tried in the last n days or whatever. In
that case, a continuous stream of messages for one down host would all be
quickly bounced, but if there was a long gap, it would reset its retry
clock. Perhaps this time should be equal to the final retry time (i.e. 4
days typically). Opinions?

A temporary answer is to set really long retry times for domains for which
you are secondary, so they don't get into this state.

-- 
Philip Hazel            University of Cambridge Computing Service,
ph10@???      Cambridge, England. Phone: +44 1223 334714.