Add CHILD_ID or something of the kind for V$SQL. It is only true for V$SQLAREA, which happesn to be a GROUP BY. SF Mladen Gogala wrote: >I thought that hash value and address uniquely identify SQL statement? > From the following SQL, >it's obviously not the case. Is there any other way of uniquely >identifying them? It's 9.2.0.5 on Solaris, RAC. > > 1 select * from ( > 2 select address,hash_value,count(*) cnt from v$sql > 3 group by address,hash_value > 4 having count(*)>1 > 5 order by 3 ) > 6* where rownum<10 >SQL> / > >ADDRESS HASH_VALUE CNT >-------- ---------- ---------- >722070F0 1382559059 2 >72213B2C 3035575473 2 >7221C894 1680401187 2 >722252D8 2857128568 2 >72229E6C 1420320603 2 >72230078 313534591 2 >7233DBC0 173091138 2 >72381604 2956793749 2 >72386BDC 3767700815 2 > >9 rows selected. > > > -- //www.freelists.org/webpage/oracle-l