You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by Hans-Werner Friedemann <h-...@vds-herzberg.de> on 2010/09/16 10:17:51 UTC

SA-Learn per remote

Hi @ all
 
sorry, but I have to pick up this issue once again.
 
I´ve installed Spamassassin with bayes on a SuseLinux-Client.
My mailserver (hmailserver) is running on a Windows machine.
 
The Mailserver has a mailbox for spam and ham inside of themself, where 
Users can send mails to.
 
I want to perform a Cronjob on my Spamassassin-PC which starts sa-learn.
The difficulty is, that sa-learn must access the mailboxes on my hmail-Server (Windows......)
 
sa-learn --spam --mbox \\Server...\Share\...... !?
 
Thanks in advance

RE: SA-Learn per remote

Posted by Giampaolo Tomassoni <Gi...@Tomassoni.biz>.
From: Hans-Werner Friedemann [mailto:h-w.friedemann@vds-herzberg.de] 
> 
> Hi @ all
> 
> sorry, but I have to pick up this issue once again.
> 
> I´ve installed Spamassassin with bayes on a SuseLinux-Client.
> My mailserver (hmailserver) is running on a Windows machine.
> 
> The Mailserver has a mailbox for spam and ham inside of themself, where 
> Users can send mails to.
> 
> I want to perform a Cronjob on my Spamassassin-PC which starts sa-learn.
> The difficulty is, that sa-learn must access the mailboxes on my
hmail-Server (Windows......)
> 
> sa-learn --spam --mbox \\Server...\Share\...... !?
> 
> Thanks in advance

Linux can't generally handle URIs like \\Server\Share\Whatever: they are
windows-specific...

You may use Samba instead. In the Linux client, issue:

	mount -tsmbfs //Server/Share/Whatever /mnt/localmountpoint
-ousername=USERNAME,password=PASSWORD

to mount your maildir on the /mnt/localmountpoint mount point, then use
sa-learn on it.


Re: SA-Learn per remote

Posted by John Hardin <jh...@impsec.org>.
On Thu, 16 Sep 2010, Hans-Werner Friedemann wrote:

> I�ve installed Spamassassin with bayes on a SuseLinux-Client.
> My mailserver (hmailserver) is running on a Windows machine.
>
> The Mailserver has a mailbox for spam and ham inside of themself, where 
> Users can send mails to.
>
> I want to perform a Cronjob on my Spamassassin-PC which starts sa-learn. 
> The difficulty is, that sa-learn must access the mailboxes on my 
> hmail-Server (Windows......)
>
> sa-learn --spam --mbox \\Server...\Share\...... !?

Ignoring for the moment the UNC filepaths, what format is the mailbox file 
you want to read? Windows-native binary mailbox file formats likely aren't 
going to work. (I see the --mbox, I just want this point to be explicitly 
considered...)

The historical recommendation has been to configure the Windows mail 
server to have publicly-accessible shared learn-as-ham and learn-as-spam 
_folders_ that all users can access (vs. learning mailboxes) and to expose 
those folders via POP or IMAP; to train you have the SA server log in, 
retrieve and flush the messages via POP or IMAP to a local MBOX or other 
recognized mail format using one of the available tools like fetchmail, 
and then learn from the local folder of downloaded messages, then age it.

Check very carefully for message mangling in this scenario. Pulling a 
message from a Windows mail server may alter the headers or the format of 
the body in a significant-to-SA manner compared to what was originally 
received.

It is very important that spams and hams to be learned should _not_ be 
forwarded through the mail system to the training mailboxes. This will 
inevitably cause format changes that will hurt training. Messages to be 
learned should be _moved_ from the user's inbox to the train-as-spam 
folder and _copied_ from the spam quarantine to the train-as-ham folder 
and user's inbox folder.

-- 
  John Hardin KA7OHZ                    http://www.impsec.org/~jhardin/
  jhardin@impsec.org    FALaholic #11174     pgpk -a jhardin@impsec.org
  key: 0xB8732E79 -- 2D8C 34F4 6411 F507 136C  AF76 D822 E6E6 B873 2E79
-----------------------------------------------------------------------
   End users want eye candy and the "ooo's and aaaahhh's" experience
   when reading mail. To them email isn't a tool, but an entertainment
   form.                                                 -- Steve Lake
-----------------------------------------------------------------------
  Tomorrow: the 223rd anniversary of the signing of the U.S. Constitution