Feargal Reilly wrote:
> Have a look in the dbmail-dev archives for the thread from
> Febuary called "MTA return codes from dbmail-stmp", that may
> provide you with some of the background.

Hmm, but I cannot see the changes proposed in that thread, unless those were
to disable dbmail's internal bouncing while not implementing the return
codes to allow the MTA to do them...

> IIRC, it was decided over-quota should be a temporary failure as
> the recipient may clear his mailbox within the retry period. I
> think that's in line with other MDAs.

I don't think that is very common, I usually get those bounce messages
immediately. It is not very likely that a user will clear the mailbox
reasonably fast, especially since he isn't notified, and an MTA cannot
inject a message to the user about it either (since it's full). I also
recall discussion in RFCs, or similar, suggesting this to be a permanent
error, but my memory could fail me.

Someone should at least send a warning to sender.

> I believe a decision was also made that the error messages are up
> to the MTA, not dbmail.

There is at least code left in dbmail's source for generating them.

As far as I can see, there is no way for the MTA to know what kind of error
happened. dbmail-smtp only returns EX_TEMPFAIL and dbmail-lmtpd a "450 ...
TEMP FAIL"; not enough for the MTA to generate a informative bounce message.

Unless I've missed something, there is no way the MTA can distinguish
between "mailbox full" and any of the numerous other kinds of temporary
failures (dns, db, network, etc).

The bottom line, if this is the way dbmail now works, how are we supposed to
use quotas?

Regards,
Robert Andersson

Reply via email to