I've got bayes running on a per-user database. When I run: sa-learn -D -u SOMEUSER --force-expire
It goes through and tries to expire. It decides it wants to remove 34399 tokens, but it can't do it due to an invalid delta atime. I'm confused. This particular user has the following values in the bayes_vars MySQL table: spam_count: 1440 ham_count: 652 token_count: 146899 Since it's over the threshold of 112500 I figured that it would expire the extra tokens. INSTEAD, it complains about delta not having a good atime. I'm watching this MySQL DB getting bigger and bigger, and expecting it's eventually going to grow exponentially large unless the expire functions work. I'm currently at 227 megs for the .MYD file and 428 megs for the .MYI file. The last time I had an issue, the database was in excess of 4 gigs. Here's the debug output. If someone can give me a hint on how to give a proper delta atime, I'd be really thankful. May 25 20:55:12.219 [8792] dbg: bayes: using username: someu...@example.com May 25 20:55:12.219 [8792] dbg: bayes: learner_new: got store=Mail::SpamAssassin::BayesStore::MySQL=HASH(0x8055549c0) May 25 20:55:12.220 [8792] dbg: plugin: Mail::SpamAssassin::Plugin::Bayes=HASH(0x80488a7b0) implements 'learner_is_scan_available', priority 0 May 25 20:55:12.400 [8792] dbg: bayes: database connection established May 25 20:55:12.401 [8792] dbg: bayes: found bayes db version 3 May 25 20:55:12.403 [8792] dbg: bayes: Using userid: 80 May 25 20:55:12.405 [8792] dbg: config: score set 3 chosen. May 25 20:55:12.406 [8792] dbg: sa-learn: spamtest initialized May 25 20:55:12.406 [8792] dbg: learn: initializing learner May 25 20:55:12.407 [8792] dbg: plugin: Mail::SpamAssassin::Plugin::Bayes=HASH(0x80488a7b0) implements 'learner_sync', priority 0 May 25 20:55:12.408 [8792] dbg: bayes: bayes journal sync starting May 25 20:55:12.408 [8792] dbg: bayes: bayes journal sync completed May 25 20:55:12.409 [8792] dbg: plugin: Mail::SpamAssassin::Plugin::Bayes=HASH(0x80488a7b0) implements 'learner_expire_old_training', priority 0 May 25 20:55:12.410 [8792] dbg: bayes: expiry starting May 25 20:55:12.432 [8792] dbg: bayes: expiry check keep size, 0.75 * max: 112500 May 25 20:55:12.433 [8792] dbg: bayes: token count: 146899, final goal reduction size: 34399 May 25 20:55:12.433 [8792] dbg: bayes: first pass? current: 1338004512, Last: 1337967033, atime: 0, count: 0, newdelta: 0, ratio: 0, period: 43200 May 25 20:55:12.434 [8792] dbg: bayes: can't use estimation method for expiry, unexpected result, calculating optimal atime delta (first pass) May 25 20:55:12.434 [8792] dbg: bayes: expiry max exponent: 9 May 25 20:55:13.466 [8792] dbg: bayes: atime token reduction May 25 20:55:13.466 [8792] dbg: bayes: ======== =============== May 25 20:55:13.467 [8792] dbg: bayes: 43200 127066 May 25 20:55:13.467 [8792] dbg: bayes: 86400 119617 May 25 20:55:13.467 [8792] dbg: bayes: 172800 99251 May 25 20:55:13.468 [8792] dbg: bayes: 345600 71996 May 25 20:55:13.468 [8792] dbg: bayes: 691200 37284 May 25 20:55:13.468 [8792] dbg: bayes: 1382400 0 May 25 20:55:13.468 [8792] dbg: bayes: 2764800 0 May 25 20:55:13.469 [8792] dbg: bayes: 5529600 0 May 25 20:55:13.469 [8792] dbg: bayes: 11059200 0 May 25 20:55:13.469 [8792] dbg: bayes: 22118400 0 May 25 20:55:13.470 [8792] dbg: bayes: couldn't find a good delta atime, need more token difference, skipping expire May 25 20:55:13.472 [8792] dbg: bayes: expiry completed May 25 20:55:13.473 [8792] dbg: plugin: Mail::SpamAssassin::Plugin::Bayes=HASH(0x80488a7b0) implements 'learner_close', priority 0 :confused: -- View this message in context: http://old.nabble.com/sa-learn-expire-not-working.-Keeps-complaining-about-delta-atime-tp33911519p33911519.html Sent from the SpamAssassin - Users mailing list archive at Nabble.com.
