You are viewing a plain text version of this content. The canonical link for it is here.
Posted to server-user@james.apache.org by Jerry Malcolm <te...@malcolms.com> on 2016/08/14 22:40:25 UTC
Training SpamAssassin with JAMES
I am using JAMES with a mySQL mail repository. I would like to be able
to train the Bayesian component of SpamAssassin using sa-learn or some
equivalent. All the documentation I can find on sa-learn assumes MBox
format files for the folders containing spam/ham. To this point I
can't find any documentation related to integrating sa-learn with JAMES
using a mySQL repository.
Long ago before I was using SpamAssassin, I was using the
BayesianAnalysisFeeder mailet. But that is not for SpamAssassin, is
it? Is there an equivalent somewhere? Or just in general, what is the
typical way everyone who integrates SpamAssassin with JAMES going about
the process of training SpamAssassin?
BTW... I'm using JAMES 3.0 B5. But I'm assuming that the answer applies
to all versions.
Thanks.
Jerry
---------------------------------------------------------------------
To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
For additional commands, e-mail: server-user-help@james.apache.org
Re: Training SpamAssassin with JAMES
Posted by cryptearth <cr...@cryptearth.de>.
Hey Jerry,
this may not really help you, but I want to give a reply.
I'm running James 3-RC2 with Amavis, ClamAV and SpamAssassin on openSUSE
Tumbleweed and anything works fine. As I use my server for personal
purpose only I don't get any spam (even my domain-name is really
widespread). I just call sa-update and sa-compile once in a while (usual
every 14 days on my regular patch day) and don't have any issues. I
don't if it is working at all as I didn't got any spam nor infected mail
yet to test the system (also: SPF currently not enable - have to change
this) if it's able to protect me against unwanted mails, or, if the docs
are right, would just let in anything as there are no training data on
wich in could detect spam.
Also, as far as I understand the connection between james an other
protection services, james only asks them "Is this mail ok?" and only
marks them (or maybe moves them to different folders - depends how you
set up the rules) if it gets a negative response. So it seems there is
no way in the other direct to access james from the other protection
software - at least it doesn'T seem to be intented.
I can't really understand the reply you should get a "real" MTA, cause
you can also use other MTAs like postfix with a MySQL-backend also. The
main difference, and this is why I chose james, is, that most MTAs save
mail-data in filesystem and using database for user- and
domain-management only. Only very few MTAs like james or Citadel/UX
store also the mails itselfs in the database. So, I guess, storing all
data, mails itselfs included, inside database, is just uncommon and
therefore maybe not widely supported (wonder why, today it's much more
reliable to have a cluster with at least 3 sub-systems: main system,
database-/storage backend, network-/loadbalancer com's - so it's very
common today to store almost anything inside some kind of databases
instead of using flat filesystem).
So, even SA is very commonly widely used (don't know any well known
system wich doesn't use it) it seems the support for database-based
mail-storage isn't and therefore it seems SA itselfs lacks the ability
to access a database. Maybe these folks should think about adding this
10 years ago instead of tryin to tell you "nah, just use another MTA we
support", that's not really helpfull at all.
greetings
Matt
Am 07.07.2017 um 04:43 schrieb Jerry Malcolm:
> I moved to other problems for a while. But my uncaught spam is
> getting worse by the day. SpamAssassin states in its own docs that it
> going to be useless without training (i.e. SA-LEARN). So I'm back
> again with same problem and question. SA-LEARN can't interact with
> the JAMES db mail store. It only deals with file system mail
> folders/files. It baffles me that I cannot find a SINGLE JAMES user
> that is using SpamAssassin with JAMES. Is that really the case, or
> are users just not getting around to responding?
>
> Does ANYBODY know how to train the bayesian filter in SpamAssassin
> integrated with JAMES? If anybody understands the SpamAssassin API
> (assuming there is one), I'll looking into writing a Mailet to do the
> work. I and all my mail clients are drowning in uncaught spam right now.
>
> By chance is there a utility that will dump a JAMES IMAP user folder
> into a file system MBOX folder? I 'guess' I could copy a db folder
> into a file system folder, then run sa-learn on it, then erase it.
> Sounds like duct tape and baling wire... But I HAVE to do SOMETHING.
>
> Please give me some advice.
>
> Thanks.
>
> Jerry
>
>
> On 8/18/2016 2:20 PM, Jerry Malcolm wrote:
>> I tried to get answers to this and other questions on the
>> SpamAssassin forum. They basically told me to pound sand and to get
>> a 'real' MTA instead of JAMES. I'm not throwing out everything just
>> to be able to do better spam filtering. But I'm back here looking
>> for answers. Is ANYBODY using SpamAssassin with JAMES?
>>
>>
>> On 8/14/2016 5:40 PM, Jerry Malcolm wrote:
>>> I am using JAMES with a mySQL mail repository. I would like to be
>>> able to train the Bayesian component of SpamAssassin using sa-learn
>>> or some equivalent. All the documentation I can find on sa-learn
>>> assumes MBox format files for the folders containing spam/ham. To
>>> this point I can't find any documentation related to integrating
>>> sa-learn with JAMES using a mySQL repository.
>>>
>>> Long ago before I was using SpamAssassin, I was using the
>>> BayesianAnalysisFeeder mailet. But that is not for SpamAssassin, is
>>> it? Is there an equivalent somewhere? Or just in general, what is
>>> the typical way everyone who integrates SpamAssassin with JAMES
>>> going about the process of training SpamAssassin?
>>>
>>> BTW... I'm using JAMES 3.0 B5. But I'm assuming that the answer
>>> applies to all versions.
>>>
>>> Thanks.
>>>
>>> Jerry
>>>
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
>>> For additional commands, e-mail: server-user-help@james.apache.org
>>>
>>>
>>>
>>> -----
>>> No virus found in this message.
>>> Checked by AVG - www.avg.com
>>> Version: 2016.0.7752 / Virus Database: 4647/12832 - Release Date:
>>> 08/18/16
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
>> For additional commands, e-mail: server-user-help@james.apache.org
>>
>>
>> ---
>> This email has been checked for viruses by AVG.
>> http://www.avg.com
>>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
> For additional commands, e-mail: server-user-help@james.apache.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
For additional commands, e-mail: server-user-help@james.apache.org
Re: Training SpamAssassin with JAMES
Posted by Jerry Malcolm <te...@malcolms.com>.
I moved to other problems for a while. But my uncaught spam is getting
worse by the day. SpamAssassin states in its own docs that it going to
be useless without training (i.e. SA-LEARN). So I'm back again with
same problem and question. SA-LEARN can't interact with the JAMES db
mail store. It only deals with file system mail folders/files. It
baffles me that I cannot find a SINGLE JAMES user that is using
SpamAssassin with JAMES. Is that really the case, or are users just not
getting around to responding?
Does ANYBODY know how to train the bayesian filter in SpamAssassin
integrated with JAMES? If anybody understands the SpamAssassin API
(assuming there is one), I'll looking into writing a Mailet to do the
work. I and all my mail clients are drowning in uncaught spam right now.
By chance is there a utility that will dump a JAMES IMAP user folder
into a file system MBOX folder? I 'guess' I could copy a db folder into
a file system folder, then run sa-learn on it, then erase it. Sounds
like duct tape and baling wire... But I HAVE to do SOMETHING.
Please give me some advice.
Thanks.
Jerry
On 8/18/2016 2:20 PM, Jerry Malcolm wrote:
> I tried to get answers to this and other questions on the SpamAssassin
> forum. They basically told me to pound sand and to get a 'real' MTA
> instead of JAMES. I'm not throwing out everything just to be able to
> do better spam filtering. But I'm back here looking for answers. Is
> ANYBODY using SpamAssassin with JAMES?
>
>
> On 8/14/2016 5:40 PM, Jerry Malcolm wrote:
>> I am using JAMES with a mySQL mail repository. I would like to be
>> able to train the Bayesian component of SpamAssassin using sa-learn
>> or some equivalent. All the documentation I can find on sa-learn
>> assumes MBox format files for the folders containing spam/ham. To
>> this point I can't find any documentation related to integrating
>> sa-learn with JAMES using a mySQL repository.
>>
>> Long ago before I was using SpamAssassin, I was using the
>> BayesianAnalysisFeeder mailet. But that is not for SpamAssassin, is
>> it? Is there an equivalent somewhere? Or just in general, what is
>> the typical way everyone who integrates SpamAssassin with JAMES going
>> about the process of training SpamAssassin?
>>
>> BTW... I'm using JAMES 3.0 B5. But I'm assuming that the answer
>> applies to all versions.
>>
>> Thanks.
>>
>> Jerry
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
>> For additional commands, e-mail: server-user-help@james.apache.org
>>
>>
>>
>> -----
>> No virus found in this message.
>> Checked by AVG - www.avg.com
>> Version: 2016.0.7752 / Virus Database: 4647/12832 - Release Date:
>> 08/18/16
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
> For additional commands, e-mail: server-user-help@james.apache.org
>
>
> ---
> This email has been checked for viruses by AVG.
> http://www.avg.com
>
---------------------------------------------------------------------
To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
For additional commands, e-mail: server-user-help@james.apache.org
AW: Training SpamAssassin with JAMES [unsigned]
Posted by Bernd Waibel <BW...@intarsys.de>.
Hello Jerry
we are using SpamAssassin with James2.
We did not switch to v3 yet but we are on the way.
We are running james on linux, and the linux system is responsible for sa updates.
We do not use sa-learn, as I know.
It is as someone else said: We first stream the mail to sa and get back a spam level. Which is used in a second condition to decide if we discard the mail, mark it as spam in subject, or let it flow through.
We do not have a quarantain and no user interaction. Would be nice but not implemented.
SA is just 1 of 3 spam filters in row. So it just filters the mass out of the incoming mails. And thats a lot.
I am currently on holydays so I do not have access to the configs. So I could not offer technical background. If you are interested in the V2 solution and are able to wait till begin of September, please let me know.
We will switch to V3 and we will use SA there too. So it is a needed feature for us.
Best regards
Bernd Waibel
-------- Ursprüngliche Nachricht --------
Von: Jerry Malcolm <te...@malcolms.com>
Datum: 18.08.16 21:20 (GMT+01:00)
An: server-user@james.apache.org
Betreff: Re: Training SpamAssassin with JAMES
I tried to get answers to this and other questions on the SpamAssassin
forum. They basically told me to pound sand and to get a 'real' MTA
instead of JAMES. I'm not throwing out everything just to be able to do
better spam filtering. But I'm back here looking for answers. Is
ANYBODY using SpamAssassin with JAMES?
On 8/14/2016 5:40 PM, Jerry Malcolm wrote:
> I am using JAMES with a mySQL mail repository. I would like to be
> able to train the Bayesian component of SpamAssassin using sa-learn or
> some equivalent. All the documentation I can find on sa-learn assumes
> MBox format files for the folders containing spam/ham. To this point
> I can't find any documentation related to integrating sa-learn with
> JAMES using a mySQL repository.
>
> Long ago before I was using SpamAssassin, I was using the
> BayesianAnalysisFeeder mailet. But that is not for SpamAssassin, is
> it? Is there an equivalent somewhere? Or just in general, what is
> the typical way everyone who integrates SpamAssassin with JAMES going
> about the process of training SpamAssassin?
>
> BTW... I'm using JAMES 3.0 B5. But I'm assuming that the answer
> applies to all versions.
>
> Thanks.
>
> Jerry
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
> For additional commands, e-mail: server-user-help@james.apache.org
>
>
>
> -----
> No virus found in this message.
> Checked by AVG - www.avg.com<http://www.avg.com>
> Version: 2016.0.7752 / Virus Database: 4647/12832 - Release Date:
> 08/18/16
---------------------------------------------------------------------
To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
For additional commands, e-mail: server-user-help@james.apache.org
Re: Training SpamAssassin with JAMES
Posted by Jerry Malcolm <te...@malcolms.com>.
I tried to get answers to this and other questions on the SpamAssassin
forum. They basically told me to pound sand and to get a 'real' MTA
instead of JAMES. I'm not throwing out everything just to be able to do
better spam filtering. But I'm back here looking for answers. Is
ANYBODY using SpamAssassin with JAMES?
On 8/14/2016 5:40 PM, Jerry Malcolm wrote:
> I am using JAMES with a mySQL mail repository. I would like to be
> able to train the Bayesian component of SpamAssassin using sa-learn or
> some equivalent. All the documentation I can find on sa-learn assumes
> MBox format files for the folders containing spam/ham. To this point
> I can't find any documentation related to integrating sa-learn with
> JAMES using a mySQL repository.
>
> Long ago before I was using SpamAssassin, I was using the
> BayesianAnalysisFeeder mailet. But that is not for SpamAssassin, is
> it? Is there an equivalent somewhere? Or just in general, what is
> the typical way everyone who integrates SpamAssassin with JAMES going
> about the process of training SpamAssassin?
>
> BTW... I'm using JAMES 3.0 B5. But I'm assuming that the answer
> applies to all versions.
>
> Thanks.
>
> Jerry
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
> For additional commands, e-mail: server-user-help@james.apache.org
>
>
>
> -----
> No virus found in this message.
> Checked by AVG - www.avg.com
> Version: 2016.0.7752 / Virus Database: 4647/12832 - Release Date:
> 08/18/16
---------------------------------------------------------------------
To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
For additional commands, e-mail: server-user-help@james.apache.org