Hey Ethan, > Just wondering where you are seeing this. Is it only your query or a system > initiated query which causes the problem to show up? Is there a query I can > run to test for the problem? How bad is the performance? The query I had issues with is listed in MOS 1464844.1: select p, NULL, NULL from (select count(*) p from v$rman_status where operation = 'BLOCK MEDIA RECOVERY') It's generated by an MMON slave. I could reproduce the issue from just the inner query. It ran last night by MMON for 15 minutes until TEMP filled. But without the 1652 event set, I don't know if this was the query that filled it or something else. With stats collected, it returns sub-second. Rich -- //www.freelists.org/webpage/oracle-l