hi, i am using SA 2.63 together with amvisd. (debian linux) Autolearning is enabled. Due to this fact it seams that something strange is happening with the bayes db. In /var/lib/amavis/.spamassassin i am seeing about 1700 of these files bayes_toks.expire$$$$$ which consume about 3.3 GB of space in the filesystem.
I have added bayes_journal_max_size 209715200 to local.cf and wanted to force an expiery with sa-learn --force-expire but notthing seems to happen. Having a look to the debugging output i noticed the following lines. debug: bayes: expiry check keep size, 75% of max: 112500 debug: bayes: token count: 138956, final goal reduction size: 26456 debug: bayes: First pass? Current: 1080247256, Last: 1080247251, atime: 86400, count: 62142, newdelta: 202943, ratio: 2.34888116117327 debug: bayes: Can't use estimation method for expiry, something fishy, calculating optimal atime delta (first pass) debug: bayes: atime token reduction debug: bayes: ======== =============== debug: bayes: 43200 37103 debug: bayes: 86400 12 debug: bayes: 172800 0 debug: bayes: 345600 0 debug: bayes: 691200 0 debug: bayes: 1382400 0 debug: bayes: 2764800 0 debug: bayes: 5529600 0 debug: bayes: 11059200 0 debug: bayes: 22118400 0 debug: bayes: couldn't find a good delta atime, need more token difference, skipping expire. debug: Syncing complete. I am learning about 1000+ spams every dav (autolearn) for a minimum of 3 Month now. How can it bee that the maximum atime is only 86400? Is SA unable to find a good delta beacause of this fact? Is someone able to help me to reduce the bayes db size without having to loose the spam collected over the time? relevant local.cf statements. auto_learn 1 use_bayes 1 # bayes_expiry_scan_count 200000 # 10MB Database # wrong option # right would be #bayes_expiry_min_db_size 200000 bayes_journal_max_size 209715200 #auto_learn_threshold_nonspam 1.0 report_safe 0 auto_learn_threshold_spam 8.0 bayes_path /var/lib/amavis/.spamassassin/bayes many thanks -- Timo, whos filessystem has only 700MB free space left.
