You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by Reindl Harald <h....@thelounge.net> on 2015/06/02 14:36:07 UTC

shortcircuit dnsbl/uribl

is there a way to skip DNSBL/URIBL if a message hits the rule below, i 
tried to define dnsbl-rules with "priority CUST_DNSBL -450" but that 
don't change anything

given that USER_IN_SPF_WHITELIST score with -100 here there is no real 
point to fire up all the other tests, it's clear anyways that this 
message will pass

meta         SHORTCIRCUIT_NET_HAM (USER_IN_DKIM_WHITELIST || 
USER_IN_SPF_WHITELIST)
priority     SHORTCIRCUIT_NET_HAM -500
shortcircuit SHORTCIRCUIT_NET_HAM on
score        SHORTCIRCUIT_NET_HAM -0.001
describe     SHORTCIRCUIT_NET_HAM Skip tests for SPF/DKIM whitelisted 
senders



Re: shortcircuit dnsbl/uribl

Posted by RW <rw...@googlemail.com>.
On Wed, 03 Jun 2015 11:22:42 +0200
Reindl Harald wrote:

> 
> Am 02.06.2015 um 16:30 schrieb RW:
> > On Tue, 02 Jun 2015 14:36:07 +0200
> > Reindl Harald wrote:
> >
> >> given that USER_IN_SPF_WHITELIST score with -100 here there is no
> >> real point to fire up all the other tests, it's clear anyways that
> >> this message will pass
> >
> > As far as possible spamassassin does network test in parallel with
> > each other and with the local tests. Making one set of network tests
> > conditional on another set slows down the scanning of all mail
> 
> i doubt it would slow down since i could skip around 50 dns requests
> for 30% of all scanned messages 

Skipping those DNS requests would save some network traffic, but not
much else. Within SA itself sending requests and processing responses
are both cheap; the expensive part is waiting for slow responses and
time-outs and that will be short-circuited. 


> so even if the remaining 70% would get a small slowdown it won't
> matter, 90% of all inbound mail don't touch SA at all

Which means that any savings made to network traffic by short-circuiting
would be diluted by the DNS look-ups on the 90%.  Also whether or not
the slowdown is small, comparing it with a hypothetical load that's 10
times your actual load is bogus.


Re: shortcircuit dnsbl/uribl

Posted by Reindl Harald <h....@thelounge.net>.
Am 02.06.2015 um 16:30 schrieb RW:
> On Tue, 02 Jun 2015 14:36:07 +0200
> Reindl Harald wrote:
>
>> given that USER_IN_SPF_WHITELIST score with -100 here there is no
>> real point to fire up all the other tests, it's clear anyways that
>> this message will pass
>
> As far as possible spamassassin does network test in parallel with each
> other and with the local tests. Making one set of network tests
> conditional on another set slows down the scanning of all mail

i doubt it would slow down since i could skip around 50 dns requests for 
30% of all scanned messages if i just could order whitelist_auth with a 
priority before other network tests

so even if the remaining 70% would get a small slowdown it won't matter, 
90% of all inbound mail don't touch SA at all


Re: shortcircuit dnsbl/uribl

Posted by RW <rw...@googlemail.com>.
On Tue, 02 Jun 2015 14:36:07 +0200
Reindl Harald wrote:


> given that USER_IN_SPF_WHITELIST score with -100 here there is no
> real point to fire up all the other tests, it's clear anyways that
> this message will pass

As far as possible spamassassin does network test in parallel with each
other and with the local tests. Making one set of network tests
conditional on another set slows down the scanning of all mail.