You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by Martin Skjöldebrand <ma...@skjoldebrand.eu> on 2015/08/11 07:05:06 UTC

UCE not stopped

Hi,

I've run sa-learn on my spam folder (containing over 700 mails) and
random ham folder (at least as much). And still a horrible amount of
UCE (mainly loans and gambling sites, is it a conspiracy? =) ) are
allowed through with a score of 2.4 or something like that. All of it
is in Swedish or "Swedish" (well atleast 98% is). 

How do I stop these? How do I stop sucking at using sa-learn?

/Martin S

Re: UCE not stopped

Posted by Martin Skjöldebrand <ma...@skjoldebrand.eu>.
On Tue, 2015-08-11 at 13:19 +0200, Matus UHLAR - fantomas wrote:

> this just cays how spamd runs (user debian-spamd), but how do you 
> connect
> your mail system to spamassassin?
> Seems I should have been cleared in former message
> 
> and please don't send me personal replies - I read the list (and 
> ignore
> personal replies).

If I understand you corr this time =) it's connected through the
main/master.cf files of Postfix. 
Apologies for the persoal mail, wasn't quite aware how this mail client
worked and/or brain not engaged when replying.

/Martin S

Re: UCE not stopped

Posted by Matus UHLAR - fantomas <uh...@fantomas.sk>.
>On Tue, 2015-08-11 at 09:06 +0200, Matus UHLAR - fantomas wrote:
>
>> 2. how do you run SA? If you are using amavis or other system that
>> uses
>> single account for scoring, you must train that account

On 11.08.15 10:04, Martin Skjöldebrand wrote:
>This is the output of ps aux | grep spamd
>
>root     21807  0.0  1.9 170008 80528 ?        Ss   06:35   0:02
>/usr/sbin/spamd -d --pidfile=/var/run/spamassassin.pid --create-prefs -
>-max-children 2 --username debian-spamd [...]
>debian-+ 21824  0.0  2.2 178248 89544 ?        S    06:35   0:05 spamd
>child
>debian-+ 21825  0.0  1.9 170828 80728 ?        S    06:35   0:00 spamd
>child
>
>The main instance is running as root, (and sa-learn has been run as
>root) but the children are run as debian-spamd but I can't "su" to that
>user, I'm still root.

this just cays how spamd runs (user debian-spamd), but how do you connect
your mail system to spamassassin?
Seems I should have been cleared in former message

and please don't send me personal replies - I read the list (and ignore
personal replies).

-- 
Matus UHLAR - fantomas, uhlar@fantomas.sk ; http://www.fantomas.sk/
Warning: I wish NOT to receive e-mail advertising to this address.
Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu.
I feel like I'm diagonally parked in a parallel universe. 

Re: UCE not stopped

Posted by RW <rw...@googlemail.com>.
On Tue, 11 Aug 2015 10:04:41 +0200
Martin Skjöldebrand wrote:


> The main instance is running as root, (and sa-learn has been run as
> root) but the children are run as debian-spamd but I can't "su" to
> that user, I'm still root.

Use su -m

Re: UCE not stopped

Posted by Martin Skjöldebrand <ma...@skjoldebrand.eu>.
On Tue, 2015-08-11 at 19:23 +0200, Reindl Harald wrote:
> Am 11.08.2015 um 19:17 schrieb Martin Skjöldebrand:
> > On Tue, 2015-08-11 at 14:10 +0200, Martin Skjöldebrand wrote:
> > > On Tue, 2015-08-11 at 10:09 +0200, Reindl Harald wrote:
> > > 
> > > > than fix /etc/passwd and give that user a sehll (as i did for
> > > > spamass-milter user on Fedora), after that you can "su - debian
> > > > -spamd"
> > > > or just use 
> > > > http://wiki.apache.org/spamassassin/SiteWideBayesSetup
> > > 
> > > I've changed my configuration to use the sitewide settings given
> > > above.
> > 
> > After this change it looks like the Bayes rules are killing stuff 
> > as
> > they should. I'll let another day go before calling it a complete
> > success.
> 
> well, traing the bayes which is then used for scanning improves SA 
> greatly :-)

This is true =)
Also, you need to "Do It Right"(TM) for it to be really effective.

/Martin S
-- 
*****************************************************************This address is for technical mail lists only!Please you my main address at the .org domain for all other matters.*****************************************************************

Re: UCE not stopped

Posted by Reindl Harald <h....@thelounge.net>.
Am 11.08.2015 um 19:17 schrieb Martin Skjöldebrand:
> On Tue, 2015-08-11 at 14:10 +0200, Martin Skjöldebrand wrote:
>> On Tue, 2015-08-11 at 10:09 +0200, Reindl Harald wrote:
>>
>>> than fix /etc/passwd and give that user a sehll (as i did for
>>> spamass-milter user on Fedora), after that you can "su - debian
>>> -spamd"
>>> or just use http://wiki.apache.org/spamassassin/SiteWideBayesSetup
>>
>> I've changed my configuration to use the sitewide settings given
>> above.
>
> After this change it looks like the Bayes rules are killing stuff as
> they should. I'll let another day go before calling it a complete
> success.

well, traing the bayes which is then used for scanning improves SA 
greatly :-)


Re: UCE not stopped

Posted by Martin Skjöldebrand <ma...@skjoldebrand.eu>.
On Tue, 2015-08-11 at 14:10 +0200, Martin Skjöldebrand wrote:
> On Tue, 2015-08-11 at 10:09 +0200, Reindl Harald wrote:
> 
> > than fix /etc/passwd and give that user a sehll (as i did for 
> > spamass-milter user on Fedora), after that you can "su - debian
> > -spamd" 
> > or just use http://wiki.apache.org/spamassassin/SiteWideBayesSetup
> 
> I've changed my configuration to use the sitewide settings given 
> above.

After this change it looks like the Bayes rules are killing stuff as
they should. I'll let another day go before calling it a complete
success.

/martin s

Re: UCE not stopped

Posted by Martin Skjöldebrand <ma...@skjoldebrand.eu>.
On Tue, 2015-08-11 at 10:09 +0200, Reindl Harald wrote:

> than fix /etc/passwd and give that user a sehll (as i did for 
> spamass-milter user on Fedora), after that you can "su - debian
> -spamd" 
> or just use http://wiki.apache.org/spamassassin/SiteWideBayesSetup

I've changed my configuration to use the sitewide settings given above.

/Martin S

Re: UCE not stopped

Posted by Reindl Harald <h....@thelounge.net>.

Am 11.08.2015 um 10:04 schrieb Martin Skjöldebrand:
> On Tue, 2015-08-11 at 09:06 +0200, Matus UHLAR - fantomas wrote:
>
>> 2. how do you run SA? If you are using amavis or other system that
>> uses
>> single account for scoring, you must train that account
>
> This is the output of ps aux | grep spamd
>
> root     21807  0.0  1.9 170008 80528 ?        Ss   06:35   0:02
> /usr/sbin/spamd -d --pidfile=/var/run/spamassassin.pid --create-prefs -
> -max-children 2 --username debian-spamd [...]
> debian-+ 21824  0.0  2.2 178248 89544 ?        S    06:35   0:05 spamd
> child
> debian-+ 21825  0.0  1.9 170828 80728 ?        S    06:35   0:00 spamd
> child
>
> The main instance is running as root, (and sa-learn has been run as
> root) but the children are run as debian-spamd but I can't "su" to that
> user, I'm still root

than fix /etc/passwd and give that user a sehll (as i did for 
spamass-milter user on Fedora), after that you can "su - debian-spamd" 
or just use http://wiki.apache.org/spamassassin/SiteWideBayesSetup


Re: UCE not stopped

Posted by Martin Skjöldebrand <ma...@skjoldebrand.eu>.
On Tue, 2015-08-11 at 09:06 +0200, Matus UHLAR - fantomas wrote:

> 2. how do you run SA? If you are using amavis or other system that 
> uses
> single account for scoring, you must train that account

This is the output of ps aux | grep spamd

root     21807  0.0  1.9 170008 80528 ?        Ss   06:35   0:02
/usr/sbin/spamd -d --pidfile=/var/run/spamassassin.pid --create-prefs -
-max-children 2 --username debian-spamd [...]
debian-+ 21824  0.0  2.2 178248 89544 ?        S    06:35   0:05 spamd
child
debian-+ 21825  0.0  1.9 170828 80728 ?        S    06:35   0:00 spamd
child

The main instance is running as root, (and sa-learn has been run as
root) but the children are run as debian-spamd but I can't "su" to that
user, I'm still root.

/Martin S

Re: UCE not stopped

Posted by Martin Skjöldebrand <ma...@skjoldebrand.eu>.
On Tue, 2015-08-11 at 09:06 +0200, Matus UHLAR - fantomas wrote:
> On 11.08.15 07:05, Martin Skjöldebrand wrote:
> > I've run sa-learn on my spam folder (containing over 700 mails) and
> > random ham folder (at least as much). And still a horrible amount 
> > of
> > UCE (mainly loans and gambling sites, is it a conspiracy? =) ) are
> > allowed through with a score of 2.4 or something like that. All of 
> > it
> > is in Swedish or "Swedish" (well atleast 98% is).
> > 
> > How do I stop these? How do I stop sucking at using sa-learn?
> 
> 1. did you train on ham? SA must be trained on ham to be able to 
> distinguish
> what is spam and what is ham?

Yes I train ham as well.

> 2. how do you run SA? If you are using amavis or other system that 
> uses
> single account for scoring, you must train that account

Ah! That can be it ...
I'll give it a look and see if it changes anything.

/Martin S

Re: UCE not stopped

Posted by Matus UHLAR - fantomas <uh...@fantomas.sk>.
On 11.08.15 07:05, Martin Skjöldebrand wrote:
>I've run sa-learn on my spam folder (containing over 700 mails) and
>random ham folder (at least as much). And still a horrible amount of
>UCE (mainly loans and gambling sites, is it a conspiracy? =) ) are
>allowed through with a score of 2.4 or something like that. All of it
>is in Swedish or "Swedish" (well atleast 98% is).
>
>How do I stop these? How do I stop sucking at using sa-learn?

1. did you train on ham? SA must be trained on ham to be able to distinguish
what is spam and what is ham?

2. how do you run SA? If you are using amavis or other system that uses
single account for scoring, you must train that account

-- 
Matus UHLAR - fantomas, uhlar@fantomas.sk ; http://www.fantomas.sk/
Warning: I wish NOT to receive e-mail advertising to this address.
Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu.
42.7 percent of all statistics are made up on the spot.