Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> RE: v$session_longops hash join operation oddity

RE: v$session_longops hash join operation oddity

From: Alvaro Jose Fernandez <alvaro.fernandez_at_sivsa.com>
Date: Thu, 13 Sep 2007 13:56:26 +0200
Message-ID: <607D6181D9919041BE792D70EF2AEC48242D4C@LIMENS.sivsa.int>


May be a bug, or bad constructed query which fires a bug.  

Or simply a bad written query - or very bad filters. Check the current plan for this hash value, now , if you can .  

alvaro      

Hi all

This is oracle 10.1.0.4 on Sun8.

We're running a pretty intensive data mart query, that is at the moment doing a hash join.

v$session_longops shows the following:

SID SERIAL# OPNAME TARGET TARGET_DESC SOFAR TOTALWORK UNITS START_TIME LAST_UPDATE_TIME TIME_REMAINING 255.00 1'988.00 Table Scan MIS_DM.MD**_RISK NULL 80'162.00 80'162.00 Blocks 13.09.2007 13.09.2007 0.00 255.00 1'988.00 Hash Join NULL NULL 22'064.00 22'064.00 Blocks 13.09.2007 13.09.2007 0.00 255.00 1'988.00 Table Scan MIS_DM.MF**_LIMIT NULL 15'883.00 15'883.00 Blocks 13.09.2007 13.09.2007 0.00

<<... rest of partitions omitted for clarity -- all in all they sum up to about 1mio blocks ...>>

255.00 1'988.00 Table Scan MIS_DM.MF**_LIMIT NULL 29'183.00 29'183.00 Blocks 13.09.2007 13.09.2007 0.00 255.00 1'988.00 Hash Join NULL NULL 12'603'570.00 2'294'422' 320.00 Blocks 13.09.2007 13.09.2007 1'444'380.00

There is no way we can get up to 2 billion blocks. Does anyone know what exactly this is supposed to mean ? Are we hitting a bug ?

Stefan

--
http://www.freelists.org/webpage/oracle-l
Received on Thu Sep 13 2007 - 06:56:26 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US