Mailu
Spamd Training
In the mailu-dovecot container run e.g.
JUNKFOLDERS="/mail/mail@redimp.de/.Junk/cur/ /mail/rebecca@hillenkoetter.de/.Junk/cur/ /mail/christian@hillenkoetter.de/.Junk/cur/" # fuzzy add rspamc -h mailu-rspamd:11334 -P mailu -f 11 fuzzy_add $JUNKFOLDERS # learn spam rspamc -h mailu-rspamd:11334 -P mailu learn_spam $JUNKFOLDERS
Note: learn error: all learn conditions denied learning ham in default classifier is miss worded, with debug log on it reads:
rspamd_stat_classifier_is_skipped: learn condition for classifier bayes returned: already in class spam; probability 100.00%; skip classifier.
fuzzy flags -f
Defined in mailu-rspamd:/etc/rspamd/local.d/fuzzy_check.conf
... fuzzy_map = { LOCAL_FUZZY_DENIED { # Local threshold max_score = 20.0; # Flag to match flag = 11; } LOCAL_FUZZY_PROB { max_score = 10.0; flag = 12; } LOCAL_FUZZY_WHITE { max_score = 2.0; flag = 13; } }
Fulltext Search indices
# check the cache sizes find /mail -type d -name xapian-indexes -exec du -sh {} \; | sort -h
https://doc.dovecot.org/main/core/man/doveadm-fts.1.html
# Removes mails from the index that have already been expunged doveadm fts rescan -u mail@nachtling.com