Re: Issues with sa-update - seems to be running but not really updating

[Karsten Bräckelmann <gu...@rudersport.de>]

On Wed, 2009-04-22 at 08:25 -0600, Dan Roberts wrote:
> I am running Spamassassin Version 3.2.4 with Perl Version 5.8.8 on a  
> CentOS Linux 5.2 system.
> 
> As FAQ postings and other notes appeared to suggest running sa-update  
> daily, I set up a cron job to do that.   Though it has updated in the  
> past, it seems most of the time to find nothing to update, [...]

One issue at a time...

> So my question is, should I see greater frequency of updates, and if  
> so, does the info below indicate why I am not?   Any suggestions on  
> how to improve or resolve this?

There is nothing wrong or strange about that. There just is no more
recent update.

Updates as of now are primarily used to fix issues, like dropping
obsolete rules with a low hit-rate or poor S/O ratio but resulting in
FPs. Updates are *not* used to drastically change scores or rules,
because that requires a full GA run on a massive corpus -- and sometimes
also depends on new code.

Also, please keep in mind that updates are pushed manually, by humans
(err, the devs) and thus requires some work and spare time at their
hands...


> [6634] dbg: channel: metadata version = 759778
> [6634] dbg: dns: 4.2.3.updates.spamassassin.org => 759778, parsed as 759778
> [6634] dbg: channel: current version is 759778, new version is 759778, skipping channel

-- 
char *t="\10pse\0r\0dtu\0.@ghno\x4e\xc8\x79\xf4\xab\x51\x8a\x10\xf4\xf4\xc4";
main(){ char h,m=h=*t++,*x=t+2*h,c,i,l=*x,s=0; for (i=0;i<l;i++){ i%8? c<<=1:
(c=*++x); c&128 && (s+=h); if (!(h>>=1)||!t[s+h]){ putchar(t[s]);h=m;s=0; }}}

Re: Issues with sa-update - seems to be running but not really updating

[Karsten Bräckelmann <gu...@rudersport.de>]

On Wed, 2009-04-22 at 09:01 -0600, Dan Roberts wrote:
> I thought with the newer versions, rather than training Spamassassin,  
> the sa-update was the recommended way to update the rule set for  
> thwarting spam.

No. Rule updates are *NO* substitute for training. The Bayes subsystem
is independent and can *only* be enhanced by the same old training using
sa-learn.

Given that -- did you check your Bayes results? Are they getting worse?


> Is there another, perhaps better way to update the rule sets and get  
> spam to start dropping off?

One issue at a time. ;)  See my other post regarding enhancing results
and fixing issues with FNs.

Anyway, no, there is no better way "to update the [stock] rule sets".
Third-party rule-sets are a different topic and some of them actually do
use sa-update themself.

  guenther


[ useless full quote under including the sig -- snipped ]

-- 
char *t="\10pse\0r\0dtu\0.@ghno\x4e\xc8\x79\xf4\xab\x51\x8a\x10\xf4\xf4\xc4";
main(){ char h,m=h=*t++,*x=t+2*h,c,i,l=*x,s=0; for (i=0;i<l;i++){ i%8? c<<=1:
(c=*++x); c&128 && (s+=h); if (!(h>>=1)||!t[s+h]){ putchar(t[s]);h=m;s=0; }}}

Re: Issues with sa-update - seems to be running but not really updating

[Dan Roberts <da...@jlazyh.com>]

Hi back -

I thought with the newer versions, rather than training Spamassassin,  
the sa-update was the recommended way to update the rule set for  
thwarting spam.

Is there another, perhaps better way to update the rule sets and get  
spam to start dropping off?



On Apr 22, 2009, at 8:48 AM, Karsten Bräckelmann wrote:

> On Wed, 2009-04-22 at 08:25 -0600, Dan Roberts wrote:
>> I am running Spamassassin Version 3.2.4 with Perl Version 5.8.8 on a
>> CentOS Linux 5.2 system.
>>
>> As FAQ postings and other notes appeared to suggest running sa-update
>> daily, I set up a cron job to do that.   Though it has updated in the
>> past, it seems most of the time to find nothing to update, [...]
>
> One issue at a time...
>
>> So my question is, should I see greater frequency of updates, and if
>> so, does the info below indicate why I am not?   Any suggestions on
>> how to improve or resolve this?
>
> There is nothing wrong or strange about that. There just is no more
> recent update.
>
> Updates as of now are primarily used to fix issues, like dropping
> obsolete rules with a low hit-rate or poor S/O ratio but resulting in
> FPs. Updates are *not* used to drastically change scores or rules,
> because that requires a full GA run on a massive corpus -- and  
> sometimes
> also depends on new code.
>
> Also, please keep in mind that updates are pushed manually, by humans
> (err, the devs) and thus requires some work and spare time at their
> hands...
>
>
>> [6634] dbg: channel: metadata version = 759778
>> [6634] dbg: dns: 4.2.3.updates.spamassassin.org => 759778, parsed  
>> as 759778
>> [6634] dbg: channel: current version is 759778, new version is  
>> 759778, skipping channel
>
> -- 
> char *t="\10pse\0r\0dtu\0.@ghno\x4e\xc8\x79\xf4\xab\x51\x8a 
> \x10\xf4\xf4\xc4";
> main(){ char h,m=h=*t++,*x=t+2*h,c,i,l=*x,s=0; for (i=0;i<l;i++){ i 
> %8? c<<=1:
> (c=*++x); c&128 && (s+=h); if (!(h>>=1)||!t[s+h]) 
> { putchar(t[s]);h=m;s=0; }}}