I'm sort of pulling at straws here, but I'm reading the manpage for
sa-learn and it says that sa-learn will try to expire bayes tokens
according to this:
- the number of tokens in the DB is> 100,000
- the number of tokens in the DB is> bayes_expiry_max_db_size
- there is at least a 12 hr difference between the oldest and
newest token atimes
I haven't changed bayes_expiry_max_db_size and I run sa-learn
--force-expire every night via cron and I have bayes_auto_expire set to 0.
That said, my bayes_token database is huge:
| Name | Engine | Version | Row_format | Rows |
Avg_row_length | Data_length | Max_data_length | Index_length |
Data_free | Auto_increment | Create_time | Update_time |
Check_time | Collation | Checksum | Create_options |
Comment |
+-------------------+--------+---------+------------+-----------+----------------+-------------+-----------------+--------------+-----------+----------------+---------------------+-------------+------------+-------------------+----------+----------------+--------------------------+
| bayes_expire | InnoDB | 9 | Fixed | 1
| 16384 | 16384 | NULL | 16384
| 0 | NULL | 2006-07-06 11:25:28 | NULL |
NULL | latin1_swedish_ci | NULL | | InnoDB
free: 29522944 kB |
| bayes_global_vars | InnoDB | 9 | Dynamic | 1
| 16384 | 16384 | NULL | 0
| 0 | NULL | 2006-07-06 11:25:28 | NULL |
NULL | latin1_swedish_ci | NULL | | InnoDB
free: 29522944 kB |
| bayes_seen | InnoDB | 9 | Dynamic | 90902320
| 175 | 15980298240 | NULL | 0
| 0 | NULL | 2006-07-06 11:25:28 | NULL |
NULL | latin1_swedish_ci | NULL | | InnoDB
free: 29522944 kB |
| bayes_token | InnoDB | 9 | Fixed | 596422823
| 83 | 49507483648 | NULL | 40946384896
| 0 | NULL | 2006-07-06 11:25:28 | NULL |
NULL | latin1_swedish_ci | NULL | | InnoDB
free: 29522944 kB |
particularly bayes_token which is almost 50GB and has WAY more then
150,000 rows.
Is this sane?
