Re: Bounce Queue full

Charles Gregory:
>
> My compliments to kdent for his excellent postqueue FAQ.
>
> I am running whatever postfix runs on RedHat 9
> (rpm reports postfix-1.1.12-1).
>
> I've been trying to piece together what I should do for a situation where
> I am seeing numerous NDN's from spam that are 'sticking' in the active
> queue for long periods of time (which causes legitimate mail being
> forwarded to also get stuck as 'active' for an hour or more).

Be a good citizen, and don't accept undeliverable mail, instead
of harassing innocent people about mail that they didn't send.

> When I examine my queue sizes, I see this (du -k):
> 22672 ./active
> 1060 ./bounce
> 15896 ./defer
> 31312 ./deferred
> 1100 ./incoming
> 4 ./maildrop
>
> The bounce queue is very small. It's like these NDN messages are either
> sticking in 'active' status, or they are being treated as regular
> messages and 'deferred'.

There is no such thing as a bounce queue.

Sender notifications are queued and delivered exactly in
the same way as regular mail.

Wietse

> I've added to my main.cf the parameter,
> bounce_queue_lifetime = 6000s
> Which in theory should drop NDN's after 100 minutes.
> But I don't know if my Postfix is recent enough to use this parameter.
> Would it give an error message if it did not?
> I'll set the above to a more reasonable figure once the 'crush' is off.
>
> But with so little in the 'bounce' queue, will the above parameter do what
> it says? I cannot find entries in my logs that show the original mails
> incoming to know if I have a specific 'source' for this heavy load.
> All I see are bunches of queued mail from <>.
>
> One last question: I am using procmail as my local delivery agent.
> Can I use the 'local_recipient_maps' parameter safely to block
> mail? I have alias_maps and NIS unix password system.
> The FAQ was not quite clear on this....
>
> Thanks!
>
> -
>
>
>