[Spambayes] Reproducible DB corruption

[Christian Robottom Reis]

Hello there,

    I have a .hammie.db file that works fine until I train it with this
week's spam folder; at that point, I get this traceback:

kiko at anthem:~$ /usr/bin/sb_mboxtrain.py -s ~/mail/spam
Training spam (/home/kiko/mail/spam):
  Reading as Unix mbox
   870Traceback (most recent call last):
  File "/usr/bin/sb_mboxtrain.py", line 350, in <module>
    main()
  File "/usr/bin/sb_mboxtrain.py", line 341, in main
    train(h, s, True, force, trainnew, removetrained)
  File "/usr/bin/sb_mboxtrain.py", line 265, in train
    mbox_train(h, path, is_spam, force)
  File "/usr/bin/sb_mboxtrain.py", line 188, in mbox_train
    if msg_train(h, msg, is_spam, force):
  File "/usr/bin/sb_mboxtrain.py", line 106, in msg_train
    h.train(msg, is_spam)
  File "/usr/lib/python2.5/site-packages/spambayes/hammie.py", line 162, in train
    self.bayes.learn(tokenize(msg), is_spam)
  File "/usr/lib/python2.5/site-packages/spambayes/classifier.py", line 272, in learn
    self._add_msg(wordstream, is_spam)
  File "/usr/lib/python2.5/site-packages/spambayes/classifier.py", line 374, in _add_msg
    record = self._wordinfoget(word)
  File "/usr/lib/python2.5/site-packages/spambayes/storage.py", line 284, in _wordinfoget
    r = self.db.get(word)
  File "/usr/lib/python2.5/shelve.py", line 104, in get
    if self.dict.has_key(key):
  File "/usr/lib/python2.5/site-packages/PIL/__init__.py", line 256, in has_key
    
  File "bsddb/dbutils.py", line 62, in DeadlockWrap
bsddb.db.DBRunRecoveryError: (-30975, 'DB_RUNRECOVERY: Fatal error, run database recovery -- /home/kiko/.hammie.db: pgin failed for page 23238')
Exception bsddb.db.DBRunRecoveryError: DBRunRecoveryError(-30975, 'DB_RUNRECOVERY: Fatal error, run database recovery -- PANIC: fatal region error detected; run recovery') in  ignored
Exception bsddb.db.DBRunRecoveryError: DBRunRecoveryError(-30975, 'DB_RUNRECOVERY: Fatal error, run database recovery -- PANIC: fatal region error detected; run recovery') in  ignored

I can train it with other folders and it works well, but this specific folder
doesn't. I have daily backups of my .hammie.db and the spam folder going
for about 2 weeks so I can probably trace back to see where the problem
started.

Does this sound like it is something worth investigating?
-- 
Christian Robottom Reis | http://async.com.br/~kiko/ | [+55 16] 3376 0125