not enough usable tokens for Bayes? - SpamAssassin

This is a discussion on not enough usable tokens for Bayes? - SpamAssassin ; I get a weird Bayes result on one of my machines. It tells me there are too few tokens, although it should have much more than enough. SA 3.2.4, Bayes on local db. [29963] dbg: bayes: DB journal sync: last ...

+ Reply to Thread
Results 1 to 4 of 4

Thread: not enough usable tokens for Bayes?

  1. not enough usable tokens for Bayes?

    I get a weird Bayes result on one of my machines. It tells me there are
    too few tokens, although it should have much more than enough.

    SA 3.2.4, Bayes on local db.

    [29963] dbg: bayes: DB journal sync: last sync: 1215920690
    [29963] dbg: bayes: corpus size: nspam = 62748, nham = 42637
    [29963] dbg: bayes: cannot use bayes on this message; not enough usable
    tokens found

    0.000 0 3 0 non-token data: bayes db version
    0.000 0 62748 0 non-token data: nspam
    0.000 0 42637 0 non-token data: nham
    0.000 0 1795535 0 non-token data: ntokens

    Ideas?

    Kai

    --
    Kai Schätzl, Berlin, Germany
    Get your web at Conactive Internet Services: http://www.conactive.com


  2. Re: not enough usable tokens for Bayes?

    On Sun, Jul 13, 2008 at 06:09:03PM +0200, Kai Schaetzl wrote:
    > I get a weird Bayes result on one of my machines. It tells me there are
    > too few tokens, although it should have much more than enough.
    >
    > [29963] dbg: bayes: cannot use bayes on this message; not enough usable
    > tokens found


    This means there weren't enough tokens in the message.

    --
    Randomly Selected Tagline:
    "Inventing 'generic' mechanisms before you get an idea of what are they
    going to be used for is an exercise in masturbation."
    - Alexander Viro

    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.4.8 (GNU/Linux)

    iD8DBQFIeiw8RnAwoQckjjoRAgihAKCTNftoCYff6T/ucIGBvdAHGxPCCwCgwdwk
    /dVSMSfKHc2Jubw4NbxInjI=
    =5lVe
    -----END PGP SIGNATURE-----


  3. Re: not enough usable tokens for Bayes?

    Theo Van Dinter wrote on Sun, 13 Jul 2008 12:24:29 -0400:

    > This means there weren't enough tokens in the message.


    Thanks, Theo, for the quick answer.
    Ok, false alarm? I used a real message and now I get Bayes and a score.
    BUT:

    I ran the lint under 3.2.4 on two different machines. The first shows that
    warning and the other doesn't. I also get BAYES_50 on the second and no
    BAYES on the first. Shouldn't this be the same on both?

    And a second problem on the other machine:
    While troubleshooting I found that I should expire the db on the second
    machine as it contained over 3 million tokens. That's the result:

    [18595] dbg: bayes: expiry starting
    [18595] dbg: locker: refresh_lock: refresh /home/spamd/bayes/bayes.mutex
    [18595] dbg: locker: refresh_lock: refresh /home/spamd/bayes/bayes.mutex
    [18595] dbg: bayes: expiry check keep size, 0.75 * max: 1125000
    [18595] dbg: bayes: token count: 0, final goal reduction size: -1125000
    [18595] dbg: bayes: reduction goal of -1125000 is under 1,000 tokens,
    skipping expire
    [18595] dbg: bayes: expiry completed

    Makes no sense to me.

    0.000 0 3 0 non-token data: bayes db version
    0.000 0 60887 0 non-token data: nspam
    0.000 0 241224 0 non-token data: nham
    0.000 0 3548687 0 non-token data: ntokens
    0.000 0 1136505669 0 non-token data: oldest atime
    0.000 0 1215939142 0 non-token data: newest atime
    0.000 0 1215939367 0 non-token data: last journal sync
    atime
    0.000 0 1147637887 0 non-token data: last expiry atime
    0.000 0 11059200 0 non-token data: last expire atime
    delta
    0.000 0 1206279 0 non-token data: last expire
    reduction count


    Kai

    --
    Kai Schätzl, Berlin, Germany
    Get your web at Conactive Internet Services: http://www.conactive.com


  4. Re: not enough usable tokens for Bayes?

    Kai Schaetzl wrote on Sun, 13 Jul 2008 19:19:51 +0200:

    > And a second problem on the other machine:
    > While troubleshooting I found that I should expire the db on the second
    > machine as it contained over 3 million tokens. That's the result:
    >
    > [18595] dbg: bayes: expiry starting
    > [18595] dbg: locker: refresh_lock: refresh /home/spamd/bayes/bayes.mutex
    > [18595] dbg: locker: refresh_lock: refresh /home/spamd/bayes/bayes.mutex
    > [18595] dbg: bayes: expiry check keep size, 0.75 * max: 1125000
    > [18595] dbg: bayes: token count: 0, final goal reduction size: -1125000
    > [18595] dbg: bayes: reduction goal of -1125000 is under 1,000 tokens,
    > skipping expire
    > [18595] dbg: bayes: expiry completed


    How to expire this?

    Kai

    --
    Kai Schätzl, Berlin, Germany
    Get your web at Conactive Internet Services: http://www.conactive.com


+ Reply to Thread