Re: How does Oracle manage the possiblity of memory leak?

Karsten Farrell <kfarrell_at_belgariad.com> wrote in message news:<dQiT9.1173$0m4.74378441_at_newssvr21.news.prodigy.com>...
> C Chang wrote:
> > if a trigger has to call another function to process data, how oracle
> > manages the multiple copies of the function when the trigger has been
> > called multiple times? Is there an automatic garbage collector within
> > PL/SQL process? my system is 8.1.6 on NT 4
> >
> > C Chang
> The Oracle application is written mostly in C (which is why it "runs" on
> any platform with a C compiler) ... and memory leaks are a problem in
> many C programs ... but the Oracle rdbms app is older than dirt and has
> closed up pretty much every memory leak.
>
> As for your question about multiple copies of stored procedures - as you
> know, code is read-only and only *one* copy needs to exist in memory. To
> make the call, I just place some stuff on my local stack and jump right
> into the beginning of that single segment of "shared" code. At the end
> of the code is a return that will clean the stack and return to me. As
> long as I have local stack space available, everything will work out ok.
> Oracle doesn't have to write its own garbage collection routines.
>
> Of course, bad things might happen if Oracle were to generate buggy
> executable code (but if your stored procedure compiles, then you can
> rest assured that the generated machine language code won't do anything
> strange). Just as in any other app that runs in an o/s, if the o/s
> hiccups while the app is running, the stack might not be cleaned up ...
> but if the o/s does something like that, you usually have to reboot anyhow.

I was under the understanding that Oracle performs major rewrites of the kernal with every major release.

Anyway, while I do not disagree with the responses to the actual question in the post my answer to the thread topic is, "it leaks".

This is from meatlink for today from one poster ORA-00600: internal error code, arguments: [729], [21688], [space leak], [], [], [], [], []

And the patch (8.1.7.4) for exp creating core files when it attemps to conventional export certain tables which can be select * from successfully and analyze validate cleanly mentions a space leak in its readme.

HTH -- Mark D Powell -- Received on Fri Jan 10 2003 - 09:39:37 CST