Re: Oracle import allocate memory until AIX crash ?!

Thanks for Your answer Pradhan.

Ran the commands below and they told me 128MB physical memory and 128MB+256MB virtual memory was present.

The question has been sent to Oracle support for now. And I still insist on that we should being able to do the import without crashing the machine by eating up all memory.

Observation: When doing kill on the import process. Nothing happend to the allocated memory. We had to restart Oracle to release it. Don't know if that makes it any easier to figure it out thou.

I'll wait for Oracle and see what they say. And if it is of any interest and solves this problem I'll post it to the group.

/JEr

In article <R2Nn3.189$js3.420_at_newsfeed.slurp.net>,   "Pradhan" <empradhan_at_dplus.net> wrote:
> Hi
> looking into the paging space into the AIX box will be a good idea.
> If your disks have enough space then increasing the paging space
> will solve the problem. As an thumb rule two times the size of real
> memory should be the paging space and this is not necessarily
> a RULE it just a thumb rule. If you need the larger paging space
> and your hard drive can fit it then if I were U i will increase the
> paging space little more..
>
> lsps -a
> will show the size of paging space in the system
>
> lsattr -El mem0 -a size
> will show the size of real memory
>
> HTH
> pradhan
>
> Goin'n bold wrote in message <7nmr7d$oon$1_at_nnrp1.deja.com>...
> >Hi !
> >
> >When running imp on one of our large (tables containing more then
> >1.000.000 rows) databases in Oracle 8 runing on a AIX 4.x machine the
> >machine crashes.
> >
> >We started a monitoring program on the AIX server and found that it's
> >eating up all virtual memory during import. Starting at ca: 128 MB of
> >free memory it just start to allocate memory until it reaches 0 MB
and
> >crashes.
> >
> >Has anyone had any experience with this ? is it working as designed (
> >and we need to buy more memory ) or is it a bug ?
> >
> >If it's working as designed where can we restrict memory usage so the
> >import can continue ( a bit slower due to less memory) without taking
> >down the machine.
> >
> >Looking at the imp parameters we have played around with BUFFER and
> >COMMIT but the result is the same...
> >
> >/TIA, Jer

--
Gandhi: An eye for an eye only ends up making the whole world blind.

Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't. Received on Thu Jul 29 1999 - 07:10:25 CDT