On 08/15/2005 03:21:12 AM, fairlie rego wrote: > Hi Mladen, > > Hmmm I do not have the same problem... > > > SQL> select * from v$sga_target_advice; > > SGA_SIZE SGA_SIZE_FACTOR ESTD_DB_TIME ESTD_DB_TIME_FACTOR > ESTD_PHYSICAL_READS > ---------- --------------- ------------ ------------------- > ------------------- > 424 1 65 1 > 7565 > 106 .25 65 1 > 7630 > 212 .5 65 1 > 7565 > 848 2 65 1 > 7565 > 530 1.25 65 1 > 7565 > 636 1.5 65 1 > 7565 > 742 1.75 65 1 > 7565 > 318 .75 65 1 > 7565 > 8 rows selected. > > So generate an errorstack and send me the file and we'll see which ratio is > the problem... > > SQL> alter session set events '1476 trace name errorstack level 3'; > Session altered. > > Reissue the statement > > Regards, > Fairlie Here is the heart of the problem: 1* select base_estd_dbtime,base_sgasz,base_estd_phy_reads from x$kmgsbsadv SQL> / BASE_ESTD_DBTIME BASE_SGASZ BASE_ESTD_PHY_READS ---------------- ---------- ------------------- 35 352 0 35 352 0 35 352 0 35 352 0 35 352 0 35 352 0 35 352 0 35 352 0 As you can see, estimated physical reads are 0. The view definition is: select A.inst_id, A.sgasz, round((A.sgasz / A.base_sgasz), 4), round(A.base_dbtime * round((A.dbtime / A.base_estd_dbtime), 4), 0), round((A.dbtime / A.base_estd_dbtime), 4), round(A.base_phy_reads * round((A.estd_physical_reads / A.base_estd_phy_reads), 4), 0) from x$kmgsbsadv A order by A.inst_id The last column causes division by zero. This is an interface to kernel management routines and one of them should prevent this number from ever becoming zero. -- Mladen Gogala http://www.mgogala.com -- //www.freelists.org/webpage/oracle-l