Two Linux servers - both SA 3.1.9 - different results - SpamAssassin

This is a discussion on Two Linux servers - both SA 3.1.9 - different results - SpamAssassin ; Trying to set up a backup MX that will be able to take over if my primary fails. Both are set up identically yet a message sent to one yields a score of 2.5 while the other (primary) reports 6.9. ...

+ Reply to Thread
Results 1 to 4 of 4

Thread: Two Linux servers - both SA 3.1.9 - different results

  1. Two Linux servers - both SA 3.1.9 - different results

    Trying to set up a backup MX that will be able to take over if my primary
    fails.
    Both are set up identically yet a message sent to one yields a score of 2.5
    while the other (primary) reports 6.9.

    X-Spam-Status: No hits=2.5 required=4.0 tests=ALL_TRUSTED,BAYES_40,
    BODY_ENHANCEMENT2,DNS_FROM_RFC_BOGUSMX,NO_REAL_NAM E,UNDISC_RECIPS
    autolearn=no version=3.1.9

    X-Spam-Status: Yes hits=6.9 required=4.0 tests=BAYES_60,BODY_ENHANCEMENT2,
    DNS_FROM_RFC_BOGUSMX,MSGID_FROM_MTA_ID,NO_REAL_NAM E,UNDISC_RECIPS
    autolearn=no version=3.1.9

    Both setups invoke spamc/spamd via procmail local delivery and deliver
    filtered email to internal server.
    Both postfix 2.3.5 - same kernel.
    The local.cf files are identical.

    This brings to mind two questions.
    1. what could be causing the discrepancy?
    2. what is the proper practice in setting up two parallel SA servers? My
    particular setup does not employ manual spam/ham learning since the
    recipients reside on the internal email system. I have copied the bayes
    files for each user from the primary to the backup. Again what is the best
    practice to keep the two SA servers in sync given most of the traffic
    travels through the higher priority primary mail server

    This is my first post - sorry if I left out something.
    Thanks for any help.


  2. Re: Two Linux servers - both SA 3.1.9 - different results

    James M wrote:
    > Trying to set up a backup MX that will be able to take over if my
    > primary fails.
    > Both are set up identically yet a message sent to one yields a score
    > of 2.5 while the other (primary) reports 6.9.
    >
    > X-Spam-Status: No hits=2.5 required=4.0 tests=ALL_TRUSTED,BAYES_40,
    > BODY_ENHANCEMENT2,DNS_FROM_RFC_BOGUSMX,NO_REAL_NAM E,UNDISC_RECIPS
    > autolearn=no version=3.1.9
    >
    > X-Spam-Status: Yes hits=6.9 required=4.0 tests=BAYES_60,BODY_ENHANCEMENT2,
    > DNS_FROM_RFC_BOGUSMX,MSGID_FROM_MTA_ID,NO_REAL_NAM E,UNDISC_RECIPS
    > autolearn=no version=3.1.9

    Look at your bayes scores, you have more training in your primary.
    BAYES_60 vis BAYES_40

    -Matt


    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.4.9 (GNU/Linux)
    Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

    iEYEARECAAYFAkg7Kg0ACgkQyxB1nuZfLEZyiwCgqU86UjC2nj LXuqFodF67FF8E
    UnQAoJ91FSWEFmOhUwL5/r658uV7QgV6
    =hCgb
    -----END PGP SIGNATURE-----


  3. Re: Two Linux servers - both SA 3.1.9 - different results

    Hi James,

    Am 2008-05-26 12:29:06, schrieb James M:
    > Both are set up identically yet a message sent to one yields a score of 2.5
    > while the other (primary) reports 6.9.
    >
    > X-Spam-Status: No hits=2.5 required=4.0 tests=ALL_TRUSTED,BAYES_40,
    > BODY_ENHANCEMENT2,DNS_FROM_RFC_BOGUSMX,NO_REAL_NAM E,UNDISC_RECIPS
    > autolearn=no version=3.1.9
    >
    > X-Spam-Status: Yes hits=6.9 required=4.0 tests=BAYES_60,BODY_ENHANCEMENT2,
    > DNS_FROM_RFC_BOGUSMX,MSGID_FROM_MTA_ID,NO_REAL_NAM E,UNDISC_RECIPS
    > autolearn=no version=3.1.9


    I have the same effect here...

    Two machines where the installation on the first one was TARed up and
    burned with a bootable rescue system on CD. The I have installed the
    Rescue CD on a second Mailserver and gotten nearly the same results like
    you...

    One time it was arround 1.7 and the second 6.8 (and hit the score)

    > This brings to mind two questions.
    > 1. what could be causing the discrepancy?
    > 2. what is the proper practice in setting up two parallel SA servers? My


    Thanks, Greetings and nice Day
    Michelle Konzack
    Systemadministrator
    24V Electronic Engineer
    Tamay Dogan Network
    Debian GNU/Linux Consultant


    --
    Linux-User #280138 with the Linux Counter, http://counter.li.org/
    ##################### Debian GNU/Linux Consultant #####################
    Michelle Konzack Apt. 917 ICQ #328449886
    +49/177/9351947 50, rue de Soultz MSN LinuxMichi
    +33/6/61925193 67100 Strasbourg/France IRC #Debian (irc.icq.com)

    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v1.4.1 (GNU/Linux)

    iD8DBQFIOzJlC0FPBMSS+BIRAkXJAJsFaJJzwEuqmg4N3ou+Gt L2f6C4zACeNSUn
    NY8thf0KMoGpDkrl4cQkVBg=
    =tVHJ
    -----END PGP SIGNATURE-----


  4. Re: Two Linux servers - both SA 3.1.9 - different results

    Matt Rude wrote:
    > James M wrote:
    >> Trying to set up a backup MX that will be able to take over if my
    >> primary fails.
    >> Both are set up identically yet a message sent to one yields a score
    >> of 2.5 while the other (primary) reports 6.9.
    >>
    >> X-Spam-Status: No hits=2.5 required=4.0 tests=ALL_TRUSTED,BAYES_40,
    >>
    >> BODY_ENHANCEMENT2,DNS_FROM_RFC_BOGUSMX,NO_REAL_NAM E,UNDISC_RECIPS
    >> autolearn=no version=3.1.9
    >>
    >> X-Spam-Status: Yes hits=6.9 required=4.0
    >> tests=BAYES_60,BODY_ENHANCEMENT2,
    >>
    >> DNS_FROM_RFC_BOGUSMX,MSGID_FROM_MTA_ID,NO_REAL_NAM E,UNDISC_RECIPS
    >> autolearn=no version=3.1.9

    > Look at your bayes scores, you have more training in your primary.
    > BAYES_60 vis BAYES_40

    Also the first one matched ALL_TRUSTED, which suggests James has a
    broken trust path (unless the spam really did originate within your
    network).

    Usually this happens when your MX is static-NATed and is quickly fixed
    by manually declaring a trusted_networks setting.

    See:

    http://wiki.apache.org/spamassassin/TrustPath/


+ Reply to Thread