Mon, 30 Jun 2008 17:56:39 +0200, Steve :
Re: [dspam-users] How to force whitelisting ?
> -------- Original-Nachricht --------
> > Datum: Mon, 30 Jun 2008 12:46:28 +0200
> > Von: Bonnetot Jean-Daniel <[EMAIL PROTECTED]>
> > An: [email protected]
> > Betreff: Re: [dspam-users] How to force whitelisting ?
> 
> > 
> > Sat, 28 Jun 2008 00:01:27 +0200, Steve :
> > Re: [dspam-users] How to force whitelisting ?
> > > -------- Original-Nachricht --------
> > > > Datum: Fri, 27 Jun 2008 14:34:28 +0200
> > > > Von: Bonnetot Jean-Daniel <[EMAIL PROTECTED]>
> > > > An: [email protected]
> > > > Betreff: Re: [dspam-users] How to force whitelisting ?
> > > 
> > > > 
> > > > I am the only one which asking how to force whitelisting ?
> > > > 
> > > Looks like. Anyway... you could force whitelisting by directly
> > > incrementing the whitelist token in the storage. For example in
> > > MySQL you could do this (I echo the SQL statement, but I think
> > > you get the point).
> > > 
> > > Assuming the DSPAM user id where you want to force whitelist
> > > token is 4. Assuming the message you want to extract the From
> > > header is in /tmp/test.msg. Assuming you need 10 innocent hits
> > > for getting a From header line to be whitelisted, then something
> > > like this below would print out the SQL statement you need to
> > > execute in order to get that particular From line whitelisted:
> > > 
> > > _dspam_user_id="4";
> > > 
> > > _message="/tmp/test.msg";
> > > 
> > > _from_token="$(dspam_crc "$(sed '/^$/q;' ${_message} | sed -n
> > > 's:^\(From\)\:[\t ]*\(.*\):\1*\2:gIp')" | sed
> > > 's:^.*CRC\:[\t ]*\([0-9]*\)[\t ]*$:\1:')";
> > > 
> > > echo "INSERT INTO \`dspam_token_data\` (\`uid\`, \`token\`,
> > > \`spam_hits\`, \`innocent_hits\`, \`last_hit\`) VALUES
> > > (${_dspam_user_id}, '${_from_token}', 0, 10, NOW()) ON DUPLICATE
> > > KEY UPDATE innocent_hits=innocent_hits+10, last_hit=NOW();"
> > 
> > Great, this is faster and stronger than generate 20 mails and set
> > them "innocent".
> > 
> > > 
> > > 
> > > 
> > > If you want to have a huge bias towards whitelisting for that
> > > token/user, then increase the innocent counter by 1'000 or even
> > > more.
> > > 
> > > I want to emphasise that this is a hack and that you are polluting
> > > your statistical data with the above mentioned method. DSPAM is
> > > not made for such hacks. Probably better would be to completely
> > > ignore DSPAM on the MTA level for the addresses you want to
> > > whitelist.
> > > 
> > > Keep in mind that DSPAM calculates the CRC/token on the whole From
> > > header line. And it is CaSeSenSiTiVe! :)
> > > 
> > > 
> > > As you see below, all the different From lines for the same sender
> > > produce different tokens:
> > > 
> > > From: Bonnetot Jean-Daniel <[EMAIL PROTECTED]>
> > >   TOKEN: 'From*Bonnetot Jean-Daniel
> > > <[EMAIL PROTECTED]>' CRC: 12604099113258623031
> > > 
> > > From: Jean-Daniel Bonnetot <[EMAIL PROTECTED]>
> > >   TOKEN: 'From*Jean-Daniel Bonnetot
> > > <[EMAIL PROTECTED]>' CRC: 14562488709064319461
> > > 
> > > From: Bonnetot Jean-Daniel <[EMAIL PROTECTED]>
> > >   TOKEN: 'From*Bonnetot Jean-Daniel
> > > <[EMAIL PROTECTED]>' CRC: 12594531299795796023
> > > 
> > > From: Bonnetot Jean-Daniel <[EMAIL PROTECTED]>
> > >   TOKEN: 'From*Bonnetot Jean-Daniel
> > > <[EMAIL PROTECTED]>' CRC: 12609785709371195447
> > > 
> > > From: <[EMAIL PROTECTED]>
> > >   TOKEN: 'From*<[EMAIL PROTECTED]>' CRC:
> > > 8608927262636099104
> > > 
> > > From: [EMAIL PROTECTED]
> > >   TOKEN: '[EMAIL PROTECTED]' CRC:
> > > 5384785331067481214
> > 
> > Yes, I saw DSPAM is case sensitive and From field are not always the
> > same for one email address.
> > DSPAM might read "From" line with specials rules added of the
> > present token engine like:
> > - case insensitive
> > - token build with "[EMAIL PROTECTED]", everything else is
> > ignored.
> > 
> So you are purposing to use just the "From" part and then the email
> address in lowercase without any other element of the from line.
> Right? Doing so will weaken DSPAM. It is much easier for me to fake
> "From: <[EMAIL PROTECTED]>" then to fake "From: Bonnetot
> Jean-Daniel <[EMAIL PROTECTED]>".

Yes, but this is the price for a managed whitelist ;)
I think this addon may be added and not be a replacement of the
present engine for the From line.

> 
> 
> > > 
> > > 
> > > Keep as well in mind, that the above command just increased the
> > > innocent hit on the whitelist token but did not update the
> > > statistics for the user in question. From a statistical viewpoint
> > > this is not so super good (depending on how much weight you want
> > > to give to the whitelist token).
> > > 
> > > Best approach for forcing whitelisting would be to modify DSPAM to
> > > allow whitelisting from the command line with parameters. This
> > > approach would ensure that all the different storage backends
> > > would be able to get updated whitelisting. But I don't think that
> > > it would be a wise idea to go that path (I mean the forced
> > > whitelisting).
> > > 
> > > Do others on the list see a huge benefit if we would have forced
> > > whitelisting possibility from the command line?
> > 
> > I think this is a good idea !
> > 
> I would see a benefit in that only if the web interface and the other
> end-user tools would have a possibility to interact with the
> whitelisting. Just adding forced whitelisting into DSPAM is from my
> viewpoint useless. The reason why I think that it is useless is:
> - The whitelisting could be done more elegant on the MTA level (why
> even bothering to start DSPAM when you already on the MTA level know
> that you don't want to filter that email).
> - Forced manual whitelisting in the content filter (aka DSPAM) would
> pollute the statistical data of DSPAM and it would fire up the
> content filter for nothing.
> - The current implementation for whitelisting is good enough for most
> end users and it is automatic. The user does not need to maintain the
> whitelisting as DSPAM does everything for him/her.
> 

Whitelisting on the MTA level may be harder to manage depending on your
network architecture. 
I have some people who send report when DSPAM make a mistake but every
email are different, so one email address could take long time before
being whitlisted.
In the first times of using a email address, the user's dictionary is
building, this can take some time. People don't want to report many and
many mail for corresponding with others.

> 
> > > 
> > > BTW: DSPAM does not have domain whitelisting possibility. Would
> > > however not be a ultra huge development task to get that
> > > implemented in DSPAM.
> > 
> > Yes, domain whitelisting is too importante for me.
> > 
> You mean domain whitelisting done the same way as it is now done with
> the normal whitelisting functionality in DSPAM or do you mean one
> where the user can/has to interact with DSPAM in order to populate
> the domain based whitelisting?
> 

I talk about a managed whitelist.

!DSPAM:1011,4869130c150922805823753!


Reply via email to