On Wed, 22 Nov 2000, Dave C. wrote:
> Since no on else seems to have answered, I will volunteer that AFAIK,
> there is no way to tell exim to treat the 5xx codes as temporary (at
> least, not without changing the code, which is possible if you can hack
> C, since exim is opensource..)
Correct. I have bent the rules plenty of times to add facilities to Exim
for coping with broken systems of various kinds, but this one seems to
me to be a step too far. I have seen from a server end just how much
trouble arises from clients that do *not* treat 5xx codes as permanent
errors (see quite a lot of comment in the manual in various places).
I would urge fixing the underlying problem, not kludging round it.
On Wed, 22 Nov 2000, David Searles wrote:
> I'm not asking to change the SMTP specification OR to implement new features
> in exim. I'd just like to know if we (just our little corner of the exim
> user community) can choose to treat 550's as temporary failures, because in
> our configuration with our ISP that's exactly what they are; temporary.
The whole point of standards is that they facilitate interworking. A
server that gets a temporary error should follows the standard and send
a 4xx, not a 5xx error. If two entities that are communicating don't
speak the same language, there's little hope of a meaningful
conversation.
> I had hoped some type of retry rule (or similar) could deal with this
> situation. From what I've read in the docs, it appears the retry rules deal
> with connection failures and other transient errors only.
Correct. They deal with temporary errors. Permanent errors are treated
as permanent!
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
