Re: log file format

Unfortunately, the reason Log miner displays sql_redo columns like that is that the redo log contains exactly that information -

Oracle appears to update the row in situ - making it invalid for the partition, then deletes the row from the current partition and inserts it into the correct partition as a completely separate set of redo records. You would be hard pushed to demonstrate
that the three operations were a single partitioned row update.

You may already be aware of this, but if you want to play about with looking at the redo log stream in a 'semi-formatted' way, you can issue the command -

    alter system dump logfile '{fully qualified log file name}';

The file is dumped to a trace file. The log file can be any log file that your database can understand, including the current online logfile.

For experimental purposes, I tend to

    alter system switch logfile;
do a little SQL

    alter system dump logfile
so that the dump is small.

--
Jonathan Lewis
Yet another Oracle-related web site:  http://www.jlcomp.demon.co.uk

Practical Oracle 8i:  Building Efficient Databases
Publishers:  Addison-Wesley

Reviews at: http://www.jlcomp.demon.co.uk/book_rev.html



John wrote in message <8bc78dd8.0105230137.ae5a2c8_at_posting.google.com>...


>The problem with partitionned tables is as follow :

>

>suppose you make an update that cause a row to be moved from a

>partition to another with logminer you will get those sql_redo:

>

>* update ... where rowid = 'oldrowid'

>* delete ... where rowid = 'oldrowid'

>* insert ... where rowid = 'newrowid'

>

>So it seems to you there are three operations whereas there's only

>one.

>

>So if you have an automized process that is reacting to insert for

>example,it's problematic.

>