memory hungry postgres process

Hi,

I am using postgresql as the database server for our demo web site. I
have a Sun Ultra Enterprise 2 machine with 2 gigs of RAM. The problem
is that the postgres process keeps growing in memory without bound.
In a few days it grows to more than half a gig and I have to restart
everything. Is there any way to limit the core memory allocated by
the postgres process? Or is this a memory leak bug? I am using
version 6.3.2. Following is a more detailed bug report I submitted
but got no response. Please cc to my email address if you have a
suggestion.

best,

Deniz Yuret

Postdoctoral Research Associate
Artificial Intelligence Laboratory
Massachusetts Institute of Technology
545 Technology Square, NE43-815,
Cambridge, MA 02139, USA
tel: 1-617-253-6247
fax: 1-617-253-5060
mailto:deniz(at)ai(dot)mit(dot)edu
http://www.ai.mit.edu/people/deniz

============================================================================
POSTGRESQL BUG REPORT TEMPLATE
============================================================================

Your name : Deniz Yuret
Your email address : deniz(at)ai(dot)mit(dot)edu

Category : runtime: back-end
Severity : serious

Summary: Size of postgres process in main memory keeps increasing.

System Configuration
--------------------
Operating System : SunOS 5.5.1

PostgreSQL version : 6.3.2

Compiler used : gcc 2.8.1

Hardware:
---------
Sun Ultra Enterprise 2

Versions of other tools:
------------------------
GNU Make version 3.74

--------------------------------------------------------------------------

Problem Description:
--------------------
I start the postmaster using "postmaster -i". My program
opens a connection to the backend using libpq. The size
of the resulting postgres process seems to increase without
bound.

--------------------------------------------------------------------------

Test Case:
----------
Here is a typical example:
% date
Sun Mar 14 11:40:02 EST 1999
% postmaster -i
%
% ps -auxww | grep postgres
USER PID %CPU %MEM SZ RSS TT S START TIME COMMAND
start 1298 0.1 0.1 4536 2288 pts/0 S 11:43:00 0:00 /projects/start/bin/postgres -p -Q -P5 -v 65536 omnibase

;;; after 1 day of operation:

% date
Mon Mar 15 11:09:13 EST 1999
11:08:30/worldbook% ps -auxww
USER PID %CPU %MEM SZ RSS TT S START TIME COMMAND
start 1298 0.5 4.3 90024 88672 pts/0 S 11:43:00 13:49 /projects/start/bin/postgres -p -Q -P5 -v 65536 omnibase

;;; after 2 days of operation

% date
Tue Mar 16 10:03:25 EST 1999
% ps -auxww
USER PID %CPU %MEM SZ RSS TT S START TIME COMMAND
start 1298 0.0 7.8 163040 161688 pts/0 S Mar 14 38:34 /projects/start/bin/postgres -p -Q -P5 -v 65536 omnibase

;;; after 4 days of operation

% date
Thu Mar 18 12:13:08 EST 1999
% ps -auxww
USER PID %CPU %MEM SZ RSS TT S START TIME COMMAND
start 1298 9.4 18.8 390008 388664 pts/0 S Mar 14 189:40 /projects/start/bin/postgres -p -Q -P5 -v 65536 omnibase

;;; after 8 days of operation

% date
Mon Mar 22 10:56:38 EST 1999
% ps -auxww | grep omnibase
start 1298 27.7 24.1 756504 499256 pts/0 S Mar 14 668:58 /projects/start/bin/postgres -p -Q -P5 -v 65536 omnibase

--------------------------------------------------------------------------

Solution:
---------

--------------------------------------------------------------------------