ac on 30 Jul 2016 22:20:51 -0700

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: [PLUG] Spamassassin Bayes Question - Where is the Database? Is it Working?

On Sat, 30 Jul 2016 19:27:22 -0400
Casey Bralla <> wrote:
> I've been using spamassassin on my postfix/amavis system for some
> time.   I have it catching spam system-wide and sending all the spam
> to a spam user eMail address.  It seems to work fine, but I'm not
> sure the bayes learning is working.
The senders of spam are becoming much more sophisticated and what is
spam is also ham and is also spam as well as ham... 

there are some of our mail servers that no longer uses sa or bayes, for
ipv4 we created weight rules based on listing, if two listings - drop -
single listing - marked as spam ( spam folder) - seems to work well -
but it is still a "reactive" tech, so it depends on the quality of the
list / contributions / size of spam reporting network - speed of
listings - speed/balance of de-listings (to reduce fp) etc etc etc.

ipv6 works on white list, so is actually super clean - no spam from ipv6
as the listing also includes the abuse contact email address

we tried running an ipv4 white list - also with the abuse email ady in
the TXT, but there was so much push back (and no support) - the problem
with spam is not that we cannot stop it - it is that people/society does
not want to stop it :)



> I've got a cron job to run sa-learn every night on my ham and spam
> Maildir folders.  (I have manually sorted any misidentified mail to
> the correct user Maildir folders, so these directories are "pure".).
> I run this cron job as root every night, but I suspect that when I
> run it, it is not creating the same database that is used by the
> postfix/amavis system when it examines each incoming email message.
> I've found 3 versions of the bayes_toks file on the disk.  They have
> "similar" timestamps, but radically different sizes:
> -rw------- 1 root root 167919616 Jul 30
> 18:59 /root/.spamassassin/bayes_toks -rw------- 1 amavis amavis
> 5640192 Jul 30 18:23 /var/lib/amavis/.spamassassin/bayes_toks
> -rw------- 1 debian-spamd debian-spamd 667648 Jul 30
> 17:35 /var/lib/spamassassin/.spamassassin/bayes_toks
> I cannot find anywhere in my debian configuration files where the
> path is specified for bayes_toks, so I cannot set them to be the same
> file.
> So, why do I have 3 different versions of this file with -almost- the
> same timestamp but different sizes?
> Is the bayes data being used by postfix/amavis?
> Anybody have a suggestion of how to make them work properly?  (or
> maybe they are all working properly).
> I welcome thoughts and suggestions!

Philadelphia Linux Users Group         --
Announcements -
General Discussion  --