From: Ralf Hildebrandt on
* brian <postfix-list(a)logi.ca>:
> On 10-05-26 03:31 PM, Matt Hayes wrote:
> >
> >I wonder if using something like postscreen from the 2.8-snapshots would
> >help to curtail some of the resource usage.
> >
>
> Thanks, I'll check it out. However, I'd feel more optimistic about it
> if it was named prescreen ;-)

It's postfix, not prefix.
But then -- postscreen is using an RBL...

--
Ralf Hildebrandt
Geschäftsbereich IT | Abteilung Netzwerk
Charité - Universitätsmedizin Berlin
Campus Benjamin Franklin
Hindenburgdamm 30 | D-12203 Berlin
Tel. +49 30 450 570 155 | Fax: +49 30 450 570 962
ralf.hildebrandt(a)charite.de | http://www.charite.de


From: Ralf Hildebrandt on
* "Jan-Kaspar Münnich" <lists(a)jan-muennich.de>:

> In general RBLs work fine against these dictionary attacks. But in this
> special case where not one address exists at the targeted domain, I
> doubt that RBLs would decrease server load, since that would add one
> more DNS lookup. I wouldn't see a big problem there, even thousands of
> 554s normally don't stress Postfix too much.

* One could also turn off postfix.
* Or disable smtpd
* Or (if there's a spare IP) point the mx to the spare IP and run
smtp-sink there!

--
Ralf Hildebrandt
Geschäftsbereich IT | Abteilung Netzwerk
Charité - Universitätsmedizin Berlin
Campus Benjamin Franklin
Hindenburgdamm 30 | D-12203 Berlin
Tel. +49 30 450 570 155 | Fax: +49 30 450 570 962
ralf.hildebrandt(a)charite.de | http://www.charite.de


From: Wietse Venema on
Jan-Kaspar M�nnich:
> On 26.05.2010, at 21:01, Matt Hayes wrote:
>
> >> Is there
> >> something more I can do to mitigate the stress on the server?
> >
> > You could look into using RBLs such as spamhaus etc.
>
> In general RBLs work fine against these dictionary attacks. But
> in this special case where not one address exists at the targeted
> domain, I doubt that RBLs would decrease server load, since that
> would add one more DNS lookup. I wouldn't see a big problem there,
> even thousands of 554s normally don't stress Postfix too much.

With postscreen (Postfix 2.8) RBL lookup happens before SMTPD
so it takes away system load.

Wietse

From: Matt Hayes on
On 5/26/2010 4:21 PM, Ralf Hildebrandt wrote:
> * brian <postfix-list(a)logi.ca>:
>> On 10-05-26 03:31 PM, Matt Hayes wrote:
>>>
>>> I wonder if using something like postscreen from the 2.8-snapshots would
>>> help to curtail some of the resource usage.
>>>
>>
>> Thanks, I'll check it out. However, I'd feel more optimistic about it
>> if it was named prescreen ;-)
>
> It's postfix, not prefix.
> But then -- postscreen is using an RBL...
>

postscreen doesn't require you to use RBL's during its checks, however,
you have the ability to do so. The nice thing about doing RBL checks in
postscreen is it stops connections from getting to the SMTPD, thus
reducing system load.

-matt

From: Ralf Hildebrandt on
* Matt Hayes <dominian(a)slackadelic.com>:

> postscreen doesn't require you to use RBL's during its checks,

Ah yes, the earlytalking and all.

> however, you have the ability to do so. The nice thing about doing RBL
> checks in postscreen is it stops connections from getting to the SMTPD,
> thus reducing system load.

That's how I'm using it here. It's amazing :)

--
Ralf Hildebrandt
Geschäftsbereich IT | Abteilung Netzwerk
Charité - Universitätsmedizin Berlin
Campus Benjamin Franklin
Hindenburgdamm 30 | D-12203 Berlin
Tel. +49 30 450 570 155 | Fax: +49 30 450 570 962
ralf.hildebrandt(a)charite.de | http://www.charite.de