Sieve failure on deliverdb corruption...

[Ken Murchison]

Rob Mueller wrote:
> Over night our delivery DB (DB3 - db3-3.3.11-6) croaked. This is the first
> time I've seen something like this. I've included the first error and some
> subsequent information from the log below.
> 
> Now I presume this is a DB3 problem and not a cyrus issue, but what really
> annoyed me is that all subsequent sieve filtering failed. This annoyed quite
> a few people who have forwarding to their SMS provider setup for important
> messages.
> 
> Does sieve REALLY require access to the duplicate delivery DB to be
> successful? Is there any chance it can be turned into a non-fatal error?

Its needed for reject and redirect to prevent mail loops and by vacation 
to make sure its only responds to a sender once.

We could disable these actions when deliver.db isn't available, but this 
would cause any scripts that use them to fail and you'd see the same 
results.

If we just ignored the fact that deliver.db isn't available, then we'd 
get people complaining about vacation sending out multiple responses, or 
the same message being redirected more than once, etc.

-- 
Kenneth Murchison     Oceana Matrix Ltd.
Software Engineer     21 Princeton Place
716-662-8973 x26      Orchard Park, NY 14127
--PGP Public Key--    http://www.oceana.com/~ken/ksm.pgp