autolearn=yes but sa-learn dump magic shows no new spam - SpamAssassin

This is a discussion on autolearn=yes but sa-learn dump magic shows no new spam - SpamAssassin ; Hi, I've been googling quite a bit today to find the answer to what I'm seeing that is happening on my mail server. However, I just can't seem to find a definitive answer. When looking at my mail logs I ...

+ Reply to Thread
Results 1 to 5 of 5

Thread: autolearn=yes but sa-learn dump magic shows no new spam

  1. autolearn=yes but sa-learn dump magic shows no new spam

    Hi,

    I've been googling quite a bit today to find the answer to what I'm
    seeing that is happening on my mail server. However, I just can't seem
    to find a definitive answer. When looking at my mail logs I see a number
    of autolearn=spam, however when I run "sa-learn --dump magic" nspam does
    not increment. If I run sa-learn manually, nspam increments. Is this
    normal or should each autolearn=spam indicate that nspam should
    increment by one.

    Here is my local.cf file as it pertains to bayes and autolearn:

    use_bayes 1
    bayes_auto_learn 1
    bayes_ignore_header X-Bogosity
    bayes_ignore_header X-Spam-Flag
    bayes_ignore_header X-Spam-Status
    bayes_auto_learn_threshold_nonspam -0.1
    bayes_auto_learn_threshold_spam 5.0
    use_auto_whitelist 1
    bayes_use_hapaxes 1
    bayes_min_ham_num 150
    bayes_min_spam_num 150
    score BAYES_00 -3
    score BAYES_05 -1
    score BAYES_95 6
    score BAYES_99 9
    score BAYES_20 -0.8
    score BAYES_40 0
    score BAYES_50 1.567
    score BAYES_60 3.515
    score BAYES_80 3.608

    Thanks,

    Brett


  2. Re: autolearn=yes but sa-learn dump magic shows no new spam

    On Fri, 2008-08-01 at 15:28 -0600, Brett Millett wrote:
    > Hi,
    >
    > I've been googling quite a bit today to find the answer to what I'm
    > seeing that is happening on my mail server. However, I just can't seem
    > to find a definitive answer. When looking at my mail logs I see a number
    > of autolearn=spam, however when I run "sa-learn --dump magic" nspam does
    > not increment. If I run sa-learn manually, nspam increments. Is this
    > normal or should each autolearn=spam indicate that nspam should
    > increment by one.


    The site-wide spamassassin user or the user spamassassin has been called
    on behalf is not the user you are running sa-learn --dump magic as?

    Just a guess.

    guenther


    --
    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 (c=*++x); c&128 && (s+=h); if (!(h>>=1)||!t[s+h]){ putchar(t[s]);h=m;s=0; }}}


  3. Re: autolearn=yes but sa-learn dump magic shows no new spam

    On Fri, 1 Aug 2008, Brett Millett wrote:

    > I've been googling quite a bit today to find the answer to what I'm
    > seeing that is happening on my mail server. However, I just can't seem
    > to find a definitive answer. When looking at my mail logs I see a number
    > of autolearn=spam, however when I run "sa-learn --dump magic" nspam does
    > not increment. If I run sa-learn manually, nspam increments. Is this
    > normal or should each autolearn=spam indicate that nspam should
    > increment by one.


    I can only speculate this would have something to do with the user
    autolearn is running against. I'm going to assume you are using MySQL as
    you did not state. Have you tried going into MySQL and:

    select count(*) from bayes_vars;

    to see how many usernames are in the table?

    Perhaps you can post the startup parameters spamd is using. Also how spamc
    is being called (if that is the case). That may shed more light on why you
    are getting the results you are.

    > Here is my local.cf file as it pertains to bayes and autolearn:
    >
    > use_bayes 1
    > bayes_auto_learn 1
    > bayes_ignore_header X-Bogosity
    > bayes_ignore_header X-Spam-Flag
    > bayes_ignore_header X-Spam-Status


    You do not need to include the X-Spam-* header fields as they are
    stripped before learning.

    > bayes_auto_learn_threshold_nonspam -0.1
    > bayes_auto_learn_threshold_spam 5.0
    > use_auto_whitelist 1
    > bayes_use_hapaxes 1
    > bayes_min_ham_num 150
    > bayes_min_spam_num 150
    > score BAYES_00 -3
    > score BAYES_05 -1
    > score BAYES_95 6
    > score BAYES_99 9
    > score BAYES_20 -0.8
    > score BAYES_40 0
    > score BAYES_50 1.567
    > score BAYES_60 3.515
    > score BAYES_80 3.608


    -d


  4. RE: autolearn=yes but sa-learn dump magic shows no new spam

    Great guess! I was running as root before (sudo.) Here are the results when I run the command as the site-wide user.

    sa-learn --dump magic

    0.000 0 3 0 non-token data: bayes db version
    0.000 0 329 0 non-token data: nspam
    0.000 0 42903 0 non-token data: nham
    0.000 0 158973 0 non-token data: ntokens
    0.000 0 1204946608 0 non-token data: oldest atime
    0.000 0 1205337646 0 non-token data: newest atime
    0.000 0 1217631298 0 non-token data: last journal sync atime
    0.000 0 1205335826 0 non-token data: last expiry atime
    0.000 0 417421 0 non-token data: last expire atime delta
    0.000 0 0 0 non-token data: last expire reduction count

    Thanks!

    Also, I'm just using the flat files (non-sql.)

    I'm still confused though...a "$ps aux | grep spam" reveals

    /usr/sbin/spamd --create-prefs --max-children 5 --helper-home-dir -x --virtual-config-dir=/etc/mail/spamassassin --username spamassassin -d --pidfile=/var/run/spamd.pid

    Of course my manual sa-learns were altering the bayes files under /etc/mail/spamassassin whereas the bayes files spam was writing to were under /home/spamassassin/.spamassassin. So I changed the home directory for spamassassin to /etc/spamassassin. That created a hidden directory ..spamassassin under /etc/spamassassin and when running "sa-learn --dump magic" I got errors (because spamd is still writing to bayes files just one directory up.) Thus, I made symlinks to the bayes files under the ..spamassassin pointing to the same files one dir up and everything seems to be working.

    My question is: Is that the best way to do that. Did I miss something?

    Thanks for all who have helped.

    Brett

    -----Original Message-----
    From: Karsten Bräckelmann [mailto:guenther@rudersport.de]
    Sent: Friday, August 01, 2008 3:59 PM
    To: users@spamassassin.apache.org
    Subject: Re: autolearn=yes but sa-learn dump magic shows no new spam

    On Fri, 2008-08-01 at 15:28 -0600, Brett Millett wrote:
    > Hi,
    >
    > I've been googling quite a bit today to find the answer to what I'm
    > seeing that is happening on my mail server. However, I just can't seem
    > to find a definitive answer. When looking at my mail logs I see a number
    > of autolearn=spam, however when I run "sa-learn --dump magic" nspam does
    > not increment. If I run sa-learn manually, nspam increments. Is this
    > normal or should each autolearn=spam indicate that nspam should
    > increment by one.


    The site-wide spamassassin user or the user spamassassin has been called
    on behalf is not the user you are running sa-learn --dump magic as?

    Just a guess.

    guenther


    --
    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 (c=*++x); c&128 && (s+=h); if (!(h>>=1)||!t[s+h]){ putchar(t[s]);h=m;s=0; }}}


  5. RE: autolearn=yes but sa-learn dump magic shows no new spam

    As you can see from the response I just posted, I'm not using MySQL for
    bayes (albeit, maybe I should be, that seems very convenient.)

    >You do not need to include the X-Spam-* header fields as they are
    >stripped before learning.


    Thanks. I'll pull those out.

    -----Original Message-----
    From: Duane Hill [mailto:d.hill@yournetplus.com]
    Sent: Friday, August 01, 2008 4:27 PM
    To: users@spamassassin.apache.org
    Subject: Re: autolearn=yes but sa-learn dump magic shows no new spam

    On Fri, 1 Aug 2008, Brett Millett wrote:

    > I've been googling quite a bit today to find the answer to what I'm
    > seeing that is happening on my mail server. However, I just can't seem
    > to find a definitive answer. When looking at my mail logs I see a

    number
    > of autolearn=spam, however when I run "sa-learn --dump magic" nspam

    does
    > not increment. If I run sa-learn manually, nspam increments. Is this
    > normal or should each autolearn=spam indicate that nspam should
    > increment by one.


    I can only speculate this would have something to do with the user
    autolearn is running against. I'm going to assume you are using MySQL as

    you did not state. Have you tried going into MySQL and:

    select count(*) from bayes_vars;

    to see how many usernames are in the table?

    Perhaps you can post the startup parameters spamd is using. Also how
    spamc
    is being called (if that is the case). That may shed more light on why
    you
    are getting the results you are.

    > Here is my local.cf file as it pertains to bayes and autolearn:
    >
    > use_bayes 1
    > bayes_auto_learn 1
    > bayes_ignore_header X-Bogosity
    > bayes_ignore_header X-Spam-Flag
    > bayes_ignore_header X-Spam-Status


    You do not need to include the X-Spam-* header fields as they are
    stripped before learning.

    > bayes_auto_learn_threshold_nonspam -0.1
    > bayes_auto_learn_threshold_spam 5.0
    > use_auto_whitelist 1
    > bayes_use_hapaxes 1
    > bayes_min_ham_num 150
    > bayes_min_spam_num 150
    > score BAYES_00 -3
    > score BAYES_05 -1
    > score BAYES_95 6
    > score BAYES_99 9
    > score BAYES_20 -0.8
    > score BAYES_40 0
    > score BAYES_50 1.567
    > score BAYES_60 3.515
    > score BAYES_80 3.608


    -d


+ Reply to Thread