Re: Who Write Undo Information From The DbCache To The Physically UndoDatafile.
- From: Sheldon Quinny <sheldonquinny@xxxxxxxxx>
- To: Christian Antognini <Christian.Antognini@xxxxxxxxxxxx>
- Date: Fri, 13 May 2005 13:57:54 +0530
Hey , Got Confusion ,
I Came Across An Article At This Site "
http://distancelearning.ksi.edu/demo/333/cis333.htm";
Its States :
Rollback Segments When you issue a DML statement to change data, the server
process first saves the old values in a rollback segment.
Sever Process....This Happening From The Buffer To The Datafile Or Rollback
Segments.
How Can U Justifiy This..
Do U Think Any One Can Ask The OTN Guys About It.
Thank You.
Sheldon.
On 5/12/05, Christian Antognini <Christian.Antognini@xxxxxxxxxxxx> wrote:
>
> Hi Sheldon
>
> >I Would Like To Know , If Undo Block Is Brought In The Buffer Cache,
> >During An Update.. SO What Process Write The Undo Block [Old Image] To
> >The Physically Undo Datafile.
>
> Just an observation...
>
> Which processes opened the datafile of an undo tablespace?
>
> linux:oracle:A920 /u00/app/oracle> lsof /u00/oradata/A920/undotbs01.dbf
> COMMAND PID USER FD TYPE DEVICE SIZE NODE NAME
> oracle 18808 oracle 16uW REG 8,17 220209152 327696
> /u00/oradata/A920/undotbs01.dbf
> oracle 18810 oracle 17u REG 8,17 220209152 327696
> /u00/oradata/A920/undotbs01.dbf
> oracle 18812 oracle 16u REG 8,17 220209152 327696
> /u00/oradata/A920/undotbs01.dbf
> oracle 18814 oracle 13u REG 8,17 220209152 327696
> /u00/oradata/A920/undotbs01.dbf
> oracle 18820 oracle 14u REG 8,17 220209152 327696
> /u00/oradata/A920/undotbs01.dbf
> oracle 18837 oracle 16u REG 8,17 220209152 327696
> /u00/oradata/A920/undotbs01.dbf
> oracle 18839 oracle 16u REG 8,17 220209152 327696
> /u00/oradata/A920/undotbs01.dbf
> oracle 18841 oracle 16u REG 8,17 220209152 327696
> /u00/oradata/A920/undotbs01.dbf
> oracle 29676 oracle 15u REG 8,17 220209152 327696
> /u00/oradata/A920/undotbs01.dbf
> oracle 30232 oracle 14u REG 8,17 220209152 327696
> /u00/oradata/A920/undotbs01.dbf
>
> Ok, there are 10 processes... but if you carefully give a look to the
> output only one has a write lock on it (flag "W"), i.e. 18808.
>
> linux:oracle:A920 /u00/app/oracle> ps -fp 18808
> UID PID PPID C STIME TTY TIME CMD
> oracle 18808 1 0 May11 ? 00:00:08 ora_dbw0_A920
>
> HTH
> Chris
>
--
//www.freelists.org/webpage/oracle-l
Other related posts:
