On Wed, 5 Jul 2006, Marc Sherman wrote:
> A relatively simple solution/workaround might be to store a hash of the
> complete set of retry rules, and if a change is detected, clear out the
> retry database. It would still be wrong, but at least it would err on
> the side of retrying instead of bouncing.
Doing this unconditionally wouldn't really be right, I feel. If, for
example, you add a retry rule that adds a short timeout for one
particular host, why should all the rest of the retry data be deleted?
> At the very least, the FAQ should suggest manually clearing the retry
> database whenever retry rules are modified.
I'll put a note in the manual.
--
Philip Hazel University of Cambridge Computing Service
Get the Exim 4 book: http://www.uit.co.uk/exim-book
