Re: strace dbwr to confirm O_DIRECT on Redhat Linux

You would not see the O_DIRECT flag unless you also saw the "open()" call. The key would be forcing DBWR to open another file-handle somehow, which is unlikely after instance startup if you have a small number of datafiles. Perhaps create a new tablespace in your SQL*Plus session while tracing DBWR, then ALTER SYSTEM CHECKPOINT?

Tim Gorman
consultant - Evergreen Database Technologies, Inc. P.O. Box 630791, Highlands Ranch CO 80163-0791

website   = http://www.EvDBT.com/
email     = Tim_at_EvDBT.com
mobile    = +1-303-885-4526
fax       = +1-303-484-3608
Yahoo IM  = tim_evdbt

hrishy wrote:
> Hi
>
> I am trying to strace the dbwr process to confirm that oracle is using direct IO on 10g R2 non RAC RHEL 4.0
>
> grep on the trace file does not show O_DIRECT this server uses LVM .
>
> Command used was
> strace -o /tmp/dbwr.out -p <<pid of dbwr>> &
>
> and then opened a sqlplus session to to a
> create test as
> select * from big_table
>
> i dont see any O_DIRECT in the dbwr.out file.
> filesystem options has been set to all in spfile.
>
> Is there anything special that needs to be done in the LVM mount this is using ext3 as the filesystem
>
> regards
> Hrishy
>
>
>
>
> --
> http://www.freelists.org/webpage/oracle-l
>
>
>
>
>

--
http://www.freelists.org/webpage/oracle-l