--
Robert Fazio
Senior Technical Analyst
dbabob_at_yahoo.com
"Shibu Mathew" <shibumathew_at_alliantfs.com> wrote in message
news:3004f220.0108131435.55b366ad_at_posting.google.com...
> Hello
>
> I am hoping very badly that someone would be able to help me on the
> following issue. We have a process that reads a large table into
> memory(I am using index by tables to hold them) and then run each line
> item from a loop, apply business logic, store the result in index by
> tables and at the end of the program write them back into a table.
It grows because index by tables are not physical tables, but nothing more
than memory arrays. If you are really dealing with very large tables, you
shouldn't be using index by tables. Consider at the very least using
temporary tables instead.
> When I run this program from unix, it sometimes fails with an ora-4030
> error(out of process memory). But sometimes it works. I have watched
> this process from glance, and I have noticed that when the
> RSS(Resident memory) reaches 1.03gb, it fails. Whenever it had worked,
> the RSS does not go beyond 1 gb. Does this have anything to do with
My guess is that your ulimit is 1G. If you are using MTS then you are
limited by the amount of free memory in the shared_pool or large_pool if you
have one defined.
> how the SGA is set up or how much shared memory is allocated for the
> instance. Since I am using index by tables, does the size of the PGA
> matter. What causes the RSS to grow? I would really appreciate if
> someone can give me any help on this issue.
The PGA is your limitation. And its your tables that are making the RSS
grow.
>
> Thank You
> Shibu Mathews
Received on Mon Aug 13 2001 - 18:05:14 CDT
