-----Original Message-----
From: Weiss, Rick [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, January 28, 2003 10:09 AM
To: Multiple recipients of list ORACLE-L
Subject: RE: Case of the Missing RowsDan:Thanks, I think I got through it, here's what I've found (explanations would be greatly appreciated)1) Created table as before from Designer/2000 scripts with indexes and constraints2) SQLLDR to create initial data list3) SELECT COUNT(*) FROM TB - yields 88640 rows4) RENAME TB TO TB_HOLD (takes indexes along)5) SELECT COUNT(*) FROM TB_HOLD - yields only 87257 rows (ahhh nuts)6) TRUNCATE TABLE TB_HOLD7) CREATE TB AS SELECT * FROM TB_HOLD - two empty tables8) SQLLDR into TB9) SELECT COUNT(*) FROM TB - yields 88640 rows10) INSERT INTO TB_HOLD (SELECT * FROM TB) - inserts 88640 rows11) ANALYZE both tables COMPUTE STATISTICS12) SELECT COUNT(*) FROM TB - yields 88640 rows13) SELECT COUNT(*) FROM TB - yields 88640 rows14) EXP - both tables now export 88640 rowsWhat in the world is going on??Great puzzler for the group to mull overNo jobs scheduled except for RMAN level 0 on Friday mornings@ 5:30 and other daily level 1 cumulatives and a weekly full DB export on Fridays @ 2:30Rick WeissOracle DBA-----Original Message-----
From: Fink, Dan [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, January 28, 2003 7:45 AM
To: Multiple recipients of list ORACLE-L
Subject: RE: Case of the Missing RowsRick,Try the following to see which rows are missing. It the same rows are missing each time, perhaps there is a common thread. If not, well....After Step 2, do a create table as select or sql*plus copy. This will create a backup version. Do a count(*) from each to make sure the numbers agree.After Step 4, select * from table1 minus select * from backup_copy to locate the missing rows.Dan Fink-----Original Message-----
From: Weiss, Rick [mailto:[EMAIL PROTECTED]]
Sent: Monday, January 27, 2003 4:54 PM
To: Multiple recipients of list ORACLE-L
Subject: Case of the Missing RowsI have a recurring, repeatable problem I was wondering about its cause.
Oracle 9.2.0.1 on W2K Professional (SP2) Dell Optiplex workstation Pentium 4
Step 1 - I do an SQLLDR process that loads 88640 rows to a table
Step 2 - SQL*Plus session - SELEC COUNT(*) from the table returns 88640 rows
Step 3 - Do an EXP on the table (to allow fall back to this point) - only exports 87257 rows
Step 4 - SQL*Plus session again - SELEC COUNT(*) from the table returns 87257 rowsNo one else has access to the database. There are no unusual entries in the alert log.
There is nothing I have found in the UDUMP or BDUMP directories that would help.Has anyone else experienced this??
Thanks
Rick Weiss
Title: Message
I'm
thinking it is related to the execution plan. When you compute stats, you don't
seem to have a problem. Run the same tests and check the execution plans in
autotrace.
- Re: Case of the Missing Rows JApplewhite
- RE: Case of the Missing Rows Fink, Dan
- Re: Case of the Missing Rows Ruth Gramolini
- RE: Case of the Missing Rows Fink, Dan
- RE: Case of the Missing Rows Weiss, Rick
- RE: Case of the Missing Rows Weiss, Rick
- RE: Re: Case of the Missing Rows Stephane Faroult
- RE: Case of the Missing Rows Jared . Still
- Re: Case of the Missing Rows Jonathan Lewis
- RE: Case of the Missing Rows Rachel Carmichael
- RE: Case of the Missing Rows Fink, Dan
- RE: Case of the Missing Rows Weiss, Rick
