From: Mike Futerko (no email)
Date: Mon Mar 01 2004 - 16:07:16 EST
Hi Wietse,
> > Here I mean Postfix checks SENDER address and then SENDER domain in the
> > virtual_alias_maps, as shown in the logs attached. So why Postfix checks
> > SENDER address/domain in recipient aliases? What is the reason to check SENDER
> > address and domain in RECIPIENT aliases?
>
> Postfix recent snapshots check SENDER and RECIPIENT. This can stop
> malware, and it is generally a bad idea to send mail with an invalid
> return address.
Yes, I understand and agree that it's bad to allow to send mail with
invalid return address. However logs show that Postfix checks SENDER
address in ALIAS tables, that result at least 2 SQL requests in my
case and anyway these checks fail because my alias maps don't contain
external senders. It is useful to check sender address if sender IP matches
mynetworks, but since it doesn't match - cancel of these checks may safe some
CPU resources. Also I notice that Postfix makes two more SQL requests (ALIAS
checks) after it got Maildir location (virtual_mailbox_maps) which are
unuseful as well I think. So the overall SQL session follows:
1*. check external sender address in virtual_alias_maps ()
2*. check external sender domain in virtual_alias_maps (@external.net)
3. check internal recipient address in virtual_alias_maps ()
4. check internal recipient domain in virtual_alias_maps (@internal.com)
5. check internal recipient address virtual_mailbox_maps ()
6*. check internal recipient address in virtual_alias_maps ()
7*. check internal recipient domain in virtual_alias_maps (@internal.com)
* - marks strange requests, I don't understand what is the reason to do so.
The SQL session between Postfix and PgSQL was dumped by tcpdump-like program
ethereal.
Here maybe I used not very good words (external and internal) to illustrate
sender and recipient, so:
'internal' here means any domain from virtual_mailbox_*
'external' any address/domain from the Internet which don't match mynetworks
nor virtual_mailbox_*
So would it be possible to optimize Postfix before the 2.1 release, if I'm
right with my tests result?
Or what I do wrong getting this if everything is OK with Postfix?
Many thanks,
Mike.
|
|
|