On Mon, 16 Dec 2002, Tim Jackson wrote:
> I've got a server (Exim 4.10, Linux) which is backup MX for a domain. The
> domain's primary MX has got network problems (routing loop I think) and is
> therefore inaccessible. The messages are correctly ending up on my server,
> but they seem to be getting kicked off the queue too soon.
>
> The retry rules are as default:
>
> * * F,2h,15m; G,16h,1h,1.5; F,4d,6h
>
> However, messages seem to be bouncing after only two days. For example, a
> message received by the server for user@??? at 05:48 on
> Saturday morning caused a permanent error message to be generated at 05:48
> this morning, with "No route to host: retry timeout exceeded".
Two days since what? Message arriving? That means nothing. Exim's
retrying is *host* based, not message based. Your retry rule says
"bounce stuff when the host has been down for 4 days". I suspect
that's what it is doing.
> Also, the errors seem to vary between the one above, and "retry time not
> reached for any host after a long failure period". Am I right in
> understanding (from section 31 of the spec) that the latter message is
> generated where each time a delivery for the message in question is
> attempted, the retry time hasn't been reached (because preceding messages
> in the queue have already retried recently)?
Essentially, yes.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
