AFAIK, amavis adds spam header only if score is above 2.0. I think
it's tunable, though.
The suggested spamassassin threshold is 5.0 per "maybe spam" and
should not be lowered too much.
I rely heavily on bayesian filters to catch spam. I usually pick a
"good user" and use his mail archive to train the filters.
Something like (being "filippo" the good user):
cd /var/lib/nethserver/vmail/filippo/Maildir/.Sent/cur
sa-learn -L --ham --dbpath /var/spool/amavisd/.spamassassin $(ls -t | head -300)
cd /var/lib/nethserver/vmail/filippo/Maildir/.junkmail/cur/
sa-learn -L --spam --dbpath /var/spool/amavisd/.spamassassin $(ls -t |
head -300)
Then I check that SpamAssassin has enough (200) mails to start using bayes:
su amavis -s /bin/sh -c 'sa-learn --dump magic'
nham and nspam should be above 200.
To train bayes, a user needs to move spam to the junkmail folder:
http://docs.nethserver.org/en/latest/mail.html#anti-spam
How did you enable the Zen DNSBL?
Could you send send me some unrecognized spam samples in eml/mailbox
format in a pass protected zip file?
> --
> You received this message because you are subscribed to the Google Groups
> "NethServer" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to
nethserver+...@googlegroups.com.
> To post to this group, send email to
neths...@googlegroups.com.
> To view this discussion on the web, visit
>
https://groups.google.com/d/msgid/nethserver/01beb45f-8d84-4060-8a96-c5ed10a81f66%40googlegroups.com.
> For more options, visit
https://groups.google.com/d/optout.
--
Ciao,
Filippo