Does Oracle8i work?

Does Oracle8i work?

Post by Fabio A. Arnabold » Sun, 31 Dec 1899 09:00:00



Hi there,
    We have developed an application for P&U (Biostatistics and Data
Management, Italy), based upon Microsoft Windows NT 4.0 and Oracle8i
Enterprise Edition Release 8.1.5.0.0. We have detected a bug in the
software cited, that causes the operating system handles number and/or
resources allocated to increase and never decrease until the application

crashes. The bug has been identified by Oracle as 935119 with severity
level 1. This should mean that Oracle development works on it until it's
fixed. They
didn't tell us when this will be :(. At the moment we cannot use our
system (NT 4.0 + Oracle8i + "java stored procedures") as production
server.

Can someone try to verify if the problem arises on another
platform?

The following java class, stored in Oracle8i as a java stored procedure
and called in an infinite loop, allows the problem to be detected.

The java class:

public class Test {
    public static String request(String string) throws Exception {
        return string;
    }

Quote:}

The pl/sql package to publish the java class in the dbms:

create or replace package Test as
    function request(stringa IN varchar2) return VARCHAR2;
end;
/
create or replace package body Test as
    function request(stringa IN varchar2) return VARCHAR2 as
    language java name 'Test.request(java.lang.String) return
java.lang.String';
end;
/

The infinite loop:

declare
    dummy varchar2(100);
begin
    while TRUE LOOP
        select Test.request('Hello') into dummy from dual;
    end loop;
end;

Thank you for any help.

--
Biostatistics & Data Management
Pharmacia & Upjohn - Milano - Italy

 
 
 

Does Oracle8i work?

Post by Thomas Ky » Sun, 31 Dec 1899 09:00:00


I tested this in solaris without issue, let it run for 1/2 million iterations
and it worked OK.

In reading the bug, it appears the patch is ready for NT...


(if that email address didn't require changing)


>Hi there,
>    We have developed an application for P&U (Biostatistics and Data
>Management, Italy), based upon Microsoft Windows NT 4.0 and Oracle8i
>Enterprise Edition Release 8.1.5.0.0. We have detected a bug in the
>software cited, that causes the operating system handles number and/or
>resources allocated to increase and never decrease until the application

>crashes. The bug has been identified by Oracle as 935119 with severity
>level 1. This should mean that Oracle development works on it until it's
>fixed. They
>didn't tell us when this will be :(. At the moment we cannot use our
>system (NT 4.0 + Oracle8i + "java stored procedures") as production
>server.

>Can someone try to verify if the problem arises on another
>platform?

>The following java class, stored in Oracle8i as a java stored procedure
>and called in an infinite loop, allows the problem to be detected.

>The java class:

>public class Test {
>    public static String request(String string) throws Exception {
>        return string;
>    }
>}

>The pl/sql package to publish the java class in the dbms:

>create or replace package Test as
>    function request(stringa IN varchar2) return VARCHAR2;
>end;
>/
>create or replace package body Test as
>    function request(stringa IN varchar2) return VARCHAR2 as
>    language java name 'Test.request(java.lang.String) return
>java.lang.String';
>end;
>/

>The infinite loop:

>declare
>    dummy varchar2(100);
>begin
>    while TRUE LOOP
>        select Test.request('Hello') into dummy from dual;
>    end loop;
>end;

>Thank you for any help.

--
See http://govt.us.oracle.com/~tkyte/ for my columns 'Digging-in to Oracle8i'...
Current article is "Part I of V, Autonomous Transactions" updated June 21'st


Oracle Service Industries     Reston, VA   USA

Opinions are mine and do not necessarily reflect those of Oracle Corporation

 
 
 

Does Oracle8i work?

Post by Fabio A. Arnabold » Sun, 31 Dec 1899 09:00:00


Thank you very much for your prompt reply.
We executed this code on a machine with 300MB of RAM and the maximum number of
iterations reached, before the crash, was about 2 millions. For each iteartion one
NT handle was allocated and not released.
The problem is that I don't  know if the two test are comparable. Pheraps a more
significative test is to let it run al least for a day while monitoring some OS
resources.


> I tested this in solaris without issue, let it run for 1/2 million iterations
> and it worked OK.

> In reading the bug, it appears the patch is ready for NT...

> [...]

--
Biostatistics & Data Management
Pharmacia & Upjohn - Milano - Italy
 
 
 

Does Oracle8i work?

Post by Thomas Ky » Sun, 31 Dec 1899 09:00:00



(if that email address didn't require changing)


>Thank you very much for your prompt reply.
>We executed this code on a machine with 300MB of RAM and the maximum number of
>iterations reached, before the crash, was about 2 millions. For each iteartion one
>NT handle was allocated and not released.
>The problem is that I don't  know if the two test are comparable. Pheraps a more
>significative test is to let it run al least for a day while monitoring some OS
>resources.

A day is a little too long for me.

I let it roll for an hour and did about 10,000,000 iterations.

No errors.  Just chewed up the cpu for a while :)


>> I tested this in solaris without issue, let it run for 1/2 million iterations
>> and it worked OK.

>> In reading the bug, it appears the patch is ready for NT...

>> [...]

--
See http://govt.us.oracle.com/~tkyte/ for my columns 'Digging-in to Oracle8i'...
Current article is "Part I of V, Autonomous Transactions" updated June 21'st


Oracle Service Industries     Reston, VA   USA

Opinions are mine and do not necessarily reflect those of Oracle Corporation