You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@spamassassin.apache.org by Henrik K <he...@hege.li> on 2018/10/01 13:29:17 UTC

Razor2 razor_fork

Hey guys, I just committed razor_fork option, which forks the lookup process
and reads in the result later.

http://svn.apache.org/viewvc?view=revision&sortby=date&revision=1842494

I'd appreciate if someone can stress test it too.  I will look into doing
DCC and Pyzor too if things look good.

Also I have no idea what is considered a standard "Windows SpamAssassin"
setup that could be tested with?

Cheers,
Henrik


Re: Razor2 razor_fork

Posted by Henrik K <he...@hege.li>.
On Mon, Oct 01, 2018 at 09:41:55AM -0400, Kevin A. McGrail wrote:
> In trunk?

Sure. :-)

Re: Razor2 razor_fork

Posted by "Kevin A. McGrail" <km...@apache.org>.
In trunk?

On Mon, Oct 1, 2018, 08:29 Henrik K <he...@hege.li> wrote:

>
> Hey guys, I just committed razor_fork option, which forks the lookup
> process
> and reads in the result later.
>
> http://svn.apache.org/viewvc?view=revision&sortby=date&revision=1842494
>
> I'd appreciate if someone can stress test it too.  I will look into doing
> DCC and Pyzor too if things look good.
>
> Also I have no idea what is considered a standard "Windows SpamAssassin"
> setup that could be tested with?
>
> Cheers,
> Henrik
>
>

Re: Razor2 razor_fork

Posted by Henrik K <he...@hege.li>.
On Wed, Oct 17, 2018 at 02:55:30PM +0100, RW wrote:
> On Tue, 16 Oct 2018 18:53:07 +0300
> Henrik K wrote:
> 
> 
> > Trunk now supports pyzor_fork 1.  Please give it a try, I would like
> > to use these as default settings if nothing funny happens.  Seems to
> > be working stable here.
> > 
> 
> These rule currently run with positive priority
> 
> 
> priority DCC_CHECK	10
> priority DCC_REPUT_00_12  10
> priority DCC_REPUT_13_19  10
> priority DCC_REPUT_70_89  10
> priority DCC_REPUT_90_94  10
> priority DCC_REPUT_95_98  10
> priority DCC_REPUT_99_100 10
> priority PYZOR_CHECK	30
> priority RAZOR2_CHECK	20
> priority RAZOR2_CF_RANGE_51_100 10
> 
> 
> I think anyone testing these would want to override the *_CHECK
> priorities to a slightly negative value, so the network queries can
> run in parallel with the ordinary regex rules.

They already automatically adjust to -100 if fork option is used.


Re: Razor2 razor_fork

Posted by RW <rw...@googlemail.com>.
On Tue, 16 Oct 2018 18:53:07 +0300
Henrik K wrote:


> Trunk now supports pyzor_fork 1.  Please give it a try, I would like
> to use these as default settings if nothing funny happens.  Seems to
> be working stable here.
> 

These rule currently run with positive priority


priority DCC_CHECK	10
priority DCC_REPUT_00_12  10
priority DCC_REPUT_13_19  10
priority DCC_REPUT_70_89  10
priority DCC_REPUT_90_94  10
priority DCC_REPUT_95_98  10
priority DCC_REPUT_99_100 10
priority PYZOR_CHECK	30
priority RAZOR2_CHECK	20
priority RAZOR2_CF_RANGE_51_100 10


I think anyone testing these would want to override the *_CHECK
priorities to a slightly negative value, so the network queries can
run in parallel with the ordinary regex rules.

Re: Razor2 razor_fork

Posted by "Kevin A. McGrail" <km...@apache.org>.
On 10/16/2018 1:12 PM, Henrik K wrote:
> Still trunk yes, too many changes around..  I'd just like to be done with
> 3.4, everything is so old and clunky.
+1, I'll roll an RC for 3.4.3 this week if I can so if you have bugs in
3.4.3 that are important to you, please get them in!

Regards,
KAM

Re: Razor2 razor_fork

Posted by Henrik K <he...@hege.li>.
On Tue, Oct 16, 2018 at 12:13:22PM -0400, Kevin A. McGrail wrote:
> On 10/16/2018 11:53 AM, Henrik K wrote:
> > Trunk now supports pyzor_fork 1.  Please give it a try, I would like to use
> > these as default settings if nothing funny happens.  Seems to be working
> > stable here.
> 
> Sounds good.  Is this only in trunk?  I'm currently testing 3.4 getting
> ready for 3.4.3.  Hoping to put some cycles on it this week.

Still trunk yes, too many changes around..  I'd just like to be done with
3.4, everything is so old and clunky.


Re: Razor2 razor_fork

Posted by "Kevin A. McGrail" <km...@apache.org>.
On 10/16/2018 11:53 AM, Henrik K wrote:
> Trunk now supports pyzor_fork 1.  Please give it a try, I would like to use
> these as default settings if nothing funny happens.  Seems to be working
> stable here.

Sounds good.  Is this only in trunk?  I'm currently testing 3.4 getting
ready for 3.4.3.  Hoping to put some cycles on it this week.

-- 
Kevin A. McGrail
VP Fundraising, Apache Software Foundation
Chair Emeritus Apache SpamAssassin Project
https://www.linkedin.com/in/kmcgrail - 703.798.0171


Re: Razor2 razor_fork

Posted by Henrik K <he...@hege.li>.
On Tue, Oct 09, 2018 at 05:30:08PM +0300, Henrik K wrote:
> On Mon, Oct 01, 2018 at 04:29:17PM +0300, Henrik K wrote:
> > 
> > Hey guys, I just committed razor_fork option, which forks the lookup process
> > and reads in the result later.
> > 
> > http://svn.apache.org/viewvc?view=revision&sortby=date&revision=1842494
> > 
> > I'd appreciate if someone can stress test it too.  I will look into doing
> > DCC and Pyzor too if things look good.
> > 
> > Also I have no idea what is considered a standard "Windows SpamAssassin"
> > setup that could be tested with?
> 
> DCC is now async too with dccifd, please test trunk if you can. :-)
> 
> Now there's only Pyzor left slowing things down. I don't even get any hits
> with it, but I guess it's alive as ruleqa shows hits?

Trunk now supports pyzor_fork 1.  Please give it a try, I would like to use
these as default settings if nothing funny happens.  Seems to be working
stable here.


Re: Razor2 razor_fork

Posted by Henrik K <he...@hege.li>.
On Mon, Oct 01, 2018 at 04:29:17PM +0300, Henrik K wrote:
> 
> Hey guys, I just committed razor_fork option, which forks the lookup process
> and reads in the result later.
> 
> http://svn.apache.org/viewvc?view=revision&sortby=date&revision=1842494
> 
> I'd appreciate if someone can stress test it too.  I will look into doing
> DCC and Pyzor too if things look good.
> 
> Also I have no idea what is considered a standard "Windows SpamAssassin"
> setup that could be tested with?

DCC is now async too with dccifd, please test trunk if you can. :-)

Now there's only Pyzor left slowing things down. I don't even get any hits
with it, but I guess it's alive as ruleqa shows hits?