How can I use 2GB of shared buffers on Windows? - Mailing list pgsql-patches

From Takayuki Tsunakawa
Subject How can I use 2GB of shared buffers on Windows?
Date
Msg-id 02cd01c74b7f$b4baf6b0$19527c0a@OPERAO
Whole thread Raw
Responses Re: How can I use 2GB of shared buffers on Windows?
Re: How can I use 2GB of shared buffers on Windows?
List pgsql-patches
Hello,

Could anyone tell me how to use 2GB of shared buffers on Windows?  I'm
sorry for attaching large text files and for sending this mail to this
ML.
When I try to start PostgreSQL 8.2.1 on Windows 2003 Server with
shared_buffers=1024MB, I get the following error messages in the Event
Log (with log_min_messages=debug5) and can't start PostgreSQL:

DEBUG:  mapped win32 error code 8 to 12

FATAL:  shmat(id=1880) failed: Not enough space


This means the Win32 API MapViewOfFile() failed with error code =
ERROR_NOT_ENOUGH_MEMORY.  However, the machine has 4GB of RAM and the
maximum size of paging file is 8GB.

But I could start PostgreSQL with shared_buffers=900MB.  Then, I
peeked the memory map of postgres.  The attached files are the memory
usage of postgres obtained by vadump. which is a tool included in
Microsoft Resource Kit (vadump is downloadable freely.)
(I'm using packaged PostgreSQL 8.2.1 available from
www.postgresql.org.)

--------------------------------------------------
Symbols loaded: 10000000 : 10107000  libeay32.dll
Symbols loaded: 1c000000 : 1c006000  comerr32.dll
Symbols loaded: 5ba20000 : 5ba77000  hnetcfg.dll
Symbols loaded: 61770000 : 61779000  LPK.DLL
--------------------------------------------------

These modules appear to be criminals.  They are spliting the address
space of postgres and preventing postgres from allocating a large
shared memory.  They seem to be the open source libraries (but what is
hnetcfg.dll?)
Why are they located on strange (evil) places?  What can I do?


Attachment

pgsql-patches by date:

Previous
From: Magnus Hagander
Date:
Subject: Re: Feature: POSIX Shared memory support
Next
From: Heikki Linnakangas
Date:
Subject: Re: [HACKERS] Dead code in _bt_split?