| Hi Tom, Alvaro. Thanks so much for your replies. On 22-Jun-06, at 12:56 PM, Alvaro Herrera wrote:
We had to restart the server to upgrade to 8.1.4 when it was released, but since then we have not restarted our server. We haven't changed our stats gathering recently, although we've made some changes to auto-vacuum to do it more frequently (outlined in my first e-mail). Unfortunately, I do not have any data to supply previous to the upgrade as we have just recently changed our logging to be more verbose.
Here's the pg_stat_user_tables entry for this table: -[ RECORD 1 ]-+---------- relid | ##### schemaname | public relname | tablename seq_scan | 345 seq_tup_read | 46699390 idx_scan | 33731562 idx_tup_fetch | 171166937 n_tup_ins | 3359 n_tup_upd | 8205786 n_tup_del | 0 As you can see, the number of deleted tuples is 0, so I don't believe that it could be an issue with DELETEs on the table because there aren't any! It has to be something that occurs when the VACCUM threshold is hit, because it resets both the VACUUM and ANALYZE counters, and it decreases Yesterday we tried running the ANALYZE command and it did reset the counter to a non-negative value. However, the next time that the VACUUM threshold was reached, the ANALYZE counter was decreased to a negative number. Thanks for looking into this, please let me know if there's any information I can provide! -- Dylan Hansen Enterprise Systems Developer |
