Re: ORA-4030

From: Paul Drake <discgolfdba_at_yahoo.com>
Date: Fri, 6 Feb 2004 15:05:39 -0800 (PST)
Message-ID: <20040206230539.2226.qmail@web20415.mail.yahoo.com>

Jared.Still_at_radisys.com wrote:
> Hi Lisa,
> This sounds very similar to a recent discussion on
> this list.
>
> I believe that the temporary solution was to go back
> to using
> sort_area_size, though I could have it wrong.
>
> Try searching the archives as fatcity.com and
> freelists.org.
>
> Jared

bingo.
it all depends upon how many concurrent users you have to support.
we have a warehouse server
quad xeon MP
w2k adv svr (w/large memory support enabled 3GT, PAE) 8 GB physical memory
dell | emc cx200 storage unit

only 30 max concurrent users

using:

sort_area_size=24M
hash_area_size=48M
pga_aggregate_target=0

with a 1.5 GB buffer cache.
(did not yet kick the tires on indirect_buffers)

large memory support does work in 9.2.0.4, as you have already found out.

can you move your buffer cache up to indirect address space, and use the memory below 3 GB for the PGA?

hth.

> "Koivu, Lisa" <Lisa.Koivu_at_Cendant-TRG.com>
> Sent by: oracle-l-bounce_at_freelists.org
> 02/06/2004 01:38 PM
> Please respond to oracle-l
>
>
> To: <oracle-l_at_freelists.org>
> cc:
> Subject: ORA-4030
>
>
> Hi everyone,=20
>
> 9204, Windows 2003 Enterprise
>
> ORA-04030: out of process memory when trying to
> allocate 49676 bytes
> (hash-join subh,kllcqas:kllsltba)
>
> What fun. Yes, I know this means it is out of
> memory. I know that hash
> joins are memory-intensive. I want to allow the db
> to utilize more
> memory. Using hash joins speeds up the data loads.
>
> I have pga_aggregate_target set to 900m (due to a
> bug). I decreased the
> amount of memory allocated to the sga and the db
> block buffers as
> indicated in note 116076.1. Yes I know it's about
> NT, but I'm grasping
> here.=20
>
> I watched the memory allocation ramp up on the
> server while Informatica
> was running. It gets to around 2.1, 2.2 gb and
> doesn't go any farther.
> Jobs begin failing.=20
>
> Our sysadmin says there's no problem with w2003E
> addressing more than
> 2GB contiguous memory, like the note says. However,
> the behavior was
> pretty suspect. =20
>
> Any windows people that can point me in the right
> direction? =20
>
> This was brought on by setting the optimizer_goal
> for this user to
> choose. We set it back to first_rows and this error
> does not occur.
> This makes sense to me, however I really want to see
> the db use
> available resources instead of barf.
>
>
> Lisa Koivu
> Senior DBA, yea right
> Orlando, FL, USA

Do you Yahoo!?
Yahoo! Finance: Get your refund fast by filing online. http://taxes.yahoo.com/filing.html

Please see the official ORACLE-L FAQ: http://www.orafaq.com

To unsubscribe send email to: oracle-l-request_at_freelists.org put 'unsubscribe' in the subject line.

--
Archives are at http://www.freelists.org/archives/oracle-l/
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------

Received on Fri Feb 06 2004 - 17:05:39 CST