"Saikat Chakraborty" <saikatchak_at_hotmail.com> wrote in message news:<87f1c0d1e3771f0394b8c44591996ec1.16981_at_mygate.mailgate.org>...
> Hi,
> Database version and OS information would be helpful. But in sort:
> DB_WRITER_IO_SLAVES simulates asynch IO. They do not perform
> asynch IO. They are helpful only when the operating system
> donot support asynch IO. If your OS supports Asynch IO, then
> you should use multiple DBWR, and set disk_asynch_io=TRUE
>
> One problem of this is, if the I/O volume is very high, so that
> the OS could not handle it, In that case there is a chance that asynch
> I/O
> will time out. This will lead to crash of the database.
>
> Only in that case you should use slaves with setting
> disk_asynch_io=FALSE
>
> You can not use DBWR_IO_SLAVES and multiple DBWR process at the same
> time.
>
> Hope this helps,
> Thanks,
> Saikat Chakraborty
>
> "yls177" <yls177_at_hotmail.com> wrote in message
> news:c06e4d68.0311242117.6491e1f1_at_posting.google.com
>
> > i have multiple "Cannot start multiple dbwrs when using I/O slaves"
> > entries in my alert file. upon checking the parameters, i have this
> > database writer and database I/O slaves set at the same time.
> >
> > upon doing some reading, realise that i cant have these two at the
> > same time unless i have DB_BLOCK_LRU_LATCHES too, am i right?
> >
> >
> > however, what my colleague managed to start oracle, is not to
> > configure any of these parameters but instead 2 other which are the
> > number of processes and lock_sga. afterwhich, the multiple dbwrs
> > messages are gone.
> >
> > comments?
hi.. my os is aix 4.3 and oracle is 8.0.7. sorry for my ignorance..
but your posts doesnt really say much about what i am getting... and
the strange thing is that my colleague changed 2 parameters, processes
and lock_sga afterwhich the oracle could be started up.
Received on Sat Nov 29 2003 - 03:21:11 CST
