[exim] Re: Use case for auto-thaw? (was Re: [exim-dev] Exim …

Góra strony
Delete this message
Reply to this message
Autor: Tony Marques
Data:  
Dla: exim-users
Temat: [exim] Re: Use case for auto-thaw? (was Re: [exim-dev] Exim from mailnull by local "Auto-Submitted: auto-generated" bounces keep bouncing)
On 6/8/05, Marc Sherman <msherman@???> wrote:
> [Moving my question from exim-dev, since this really is a user config
> issue, not a bug at all]
>
> Tony Marques wrote:
> >
> > At this point I would like to suggest that
> >
> > a) auto_thaw shouldn't apply to bounces
> >
> > or that
> >
> > b) auto_thaw should mandate ignore_bounce_errors_after be evaluated at
> > the same time (but first) and work as designed (deleting the bounce
> > after a second failure) -- (so auto_thaw 2h should imply
> > ignore_bounce_errors_after 2h).
>
> That's a good point. I wonder, though, what is the use case for
> auto_thaw? Does anyone here use it for anything useful? I've been
> wracking my brain for the duration of this thread trying to figure out a
> use case for it that wasn't Dangerous, and I really can't.
>
> - Marc


I think the suggestions should have stayed in exim-dev.

Here in exim-users, I would like everyone who does use auto_thaw to
check to make sure you also configure ignore_bounce_errors_after to
use a smaller time window. There seem to be a lot of Exim servers out
there (especially if maybe you use Cpanel) which will repeatedly send
mail from <> despite getting 55x SMTP errors (user unknown/virus/bad
content/etc, all errors that are permanent and shouldn't be retried).

Normally nobody notices this cuz the target rejects the message with a
55x error and there is no source (it's a bounce) and after 2 days it
looks exactly as if it was undeliverable (it only got 27 55x errors
instead of 27 45x errors or no/bad connects).

What seems to be happening is auto_thaw is unfreezing bounces that are
rejected by 55x errors. I consider it a big oversight and it
certainly violates RFCs. I suppose regular messages could be frozen
for a reasons other than 550 errors, but if it was because of a 55x
error they should also not be unfrozen, that is what 55x permanent
errors are for.

Actually regular messages shouldn't get frozen as they should be
returned to the sender. Perhaps this should go back to exim-dev and
we should ask to get rid of auto_thaw entirely. And here, we should
ask why do you need to use auto_thaw?