Redo log switches and waits database freezes
- From: "Jack van Zanen" <jack@xxxxxxxxxxxx>
- To: <oracle-l@xxxxxxxxxxxxx>
- Date: Mon, 20 Nov 2006 22:52:48 +0100
Hi All,
OS: windows 2003 server
Database : 10.2.0.2 STANDARD EDITION
I have a database that is running in noarchivelog mode and the log files are
not multiplexed. I have only 2 drives available for oracle so I have put my
redo log files separate from datafiles
This database is for conversion purposes and needs to be as fast as the
hardware allows me.
I am using SQL*loader to load files into this database after which I can do
transformations followed by and extraction to the final system.
I need enabled constraints (mainly check) so direct load is out of the
question I guess.
Now when I load my data After a while all 4 250MB logfiles have status
active and checkpointing freezes the database.
But even if it just loads 1 file, leaving just 1 redo log file active and 1
current, and I manually checkpoint the database this takes a long time
during which the database is not responding.
I have searched metalink and google but have not found anything usefull
relating to this.
I have some of the wait events relating to the logfiles listed here and the
top wait events for this loader session was actually log file sync
WAIT_CLASS#|EVENT|TIME_WAITED|AVERAGE_WAIT
9|log file sequential read|15|0.37
9|log file single write|68|1.67
9|log file parallel write|897575|0.75
2|log file switch completion|742|33.74
5|log file sync|896616|0.75
If you need more information please let me know specific which
Brgds Jack
Other related posts:
