Re: ora-04030 out of process memory

nitin_kaul_at_rediffmail.com wrote:
> Joel Garry wrote:
> > >Patch to 9.2.0.4 or above.
> >
> > And see bug 3836164. When is 9.2.0.7 out, anyways?
> >
> > jg
> > --
> > @home.com is bogus.
> > "But I remember reading somewhere that New York has more rain than
> > Seattle. That seems to me quite possible. I was caught once in a
New
> > York rain and I practically drowned. Soaked to the skin I was." -
Don
> > Freeman
>
> hi thankx joel
>
> This could be a bug as the process is not releasing memory back.. i
> dont have oracle support ...also i couldt not access the bug
> 3836164..and the patch for it also...so where can i get patch
from..is
> it publicily accessible..
>
> regards
> nitin

How did you post to metalink Server-Enterprise and Standard Edition/DBA Administration Technical Forum fora if you don't have support? (Well, maybe there's something I don't know, since I always make sure and have support. One _used_ to have to have support.) That Oracle person there simply quoted the workarounds in the bug description. They don't work for me (on hp-ux 9206), so there is probably more going on. For your situation I would recommend looking at what is going on with the memory usage as described in the docs that talk about v$pgastat and such like Note:223730.1. And things like select sid,name,value
from v$statname n,v$sesstat s
where n.STATISTIC# = s.STATISTIC# and

   name like 'session%memory%'
order by 3 asc;
to find the worst offenders. On my system there are just a few offenders, but they don't want to disconnect their multiple sessions after each session expands, so it adds up. Of course, on my configuration the answer is a simple increase in swap size and take away SGA from another instance.

You might also cut your pga_aggregate_target by 2/3 (it is outside the SGA, so I think the estimates for pga_aggregate_target, given in the docs and everyone's papers, have achieved myth status, are incorrect, and everything is made worse by memory leaks, in different ways for each 9.2 patch set combined with each app). Cut sga_max_size and shared_pool_size in half (and what were those buffers again?). Did you set shared_pool_size that size to try to get rid of 4031 errors? Then start tuning all over again. And of course, maybe it's not pga at all, but some cursor issue. Nobody can really say until you fix the known issues by patching.

jg

--
@home.com is bogus?
New Opera product:
http://66.102.7.104/search?q=cache:u_mzTxdfTJYJ:www.opera.com/pressreleases/en/2005/04/01/+&hl=en