Re: Setting db_32k_cache_size breaks 11.1.0.7

From: Greg Rahn <greg_at_structureddata.org>
Date: Sun, 23 Nov 2008 14:46:32 -0800
Message-ID: <a9c093440811231446u724ba48egf6898f360d9295e2@mail.gmail.com>


There are lots of parameters that exist that only work on certain platforms but are visible on all.

You are correct, it shouldnt crash - file a bug with support on that.

If you are using MSMM and not ASMM (sga_target/memory_target), and you try and start up your database with db_32k_cache_size >0, it mentions that 16k is the max supported size:

SQL> ORA-00382: 32768 not a valid block size, valid range [2048..16384]

Personally I dont use ASMM on Linux because it doesnt support the use of hugepages, only old school manually segment management does currently.

On Sun, Nov 23, 2008 at 12:19 AM, Andre van Winssen <dreveewee_at_gmail.com> wrote:
> Hi Greg,
> But what about below
>
> SQL> show parameter k_cache_size
>
> NAME TYPE VALUE
> ------------------------------------ ----------- -----
> db_16k_cache_size big integer 0
> db_2k_cache_size big integer 0
> db_32k_cache_size big integer 0 <<
> db_4k_cache_size big integer 0
> db_8k_cache_size big integer 0
>
> appearantly db_32k_cache_size is an allowed value.
>
> more importantly: oracle database should never crash because of this request
> to add a 32K buffer cache. It should say: sorry pal, I can't support this
> request for a 32k buffer cache. No, instead it crashes and we have to pick
> up the pieces.

-- 
Regards,
Greg Rahn
http://structureddata.org
--
http://www.freelists.org/webpage/oracle-l
Received on Sun Nov 23 2008 - 16:46:32 CST

Original text of this message