Re: RAC in NAS

From: Nuno Souto <dbvision_at_iinet.net.au>
Date: Sat, 29 Jul 2006 23:38:17 +1000
Message-ID: <44CB64C9.1090904@iinet.net.au>

Frits Hoogland wrote,on my timestamp of 29/07/2006 10:27 PM:

> doesn't do a strace on the dbwr and lgwr do what you want to know?
> it is simple, and it tells you all you want!
> the calls of synchronous and asynchronous IO are different!

not really. At least not in Linux RH3 and O9i: the calls show up as the same no matter if aio is being used or not. And the open flags show up the same as well. I'm assuming aio on file systems - aio on raw might be a different kettle, I have not tried strace there. In fact I have not tried raw with Linux, period. And I believe with 2.6 kernel raw becomes "deprecated", so things are getting harder on that front!

The only way I've found to confirm that *something* is using aio kernel code is have a gander at slabinfo and then use stats in Oracle instance to see if the aio counters are ticking over.

Direct IO however will show up nicely as an open flag with strace, no matter raw or cooked io!

-- 
Cheers
Nuno Souto
in sunny Sydney, Australia
dbvision_at_iinet.net.au
--
http://www.freelists.org/webpage/oracle-l

Received on Sat Jul 29 2006 - 08:38:17 CDT