[lug] Strange memory problem ...

rm at mamma.varadinet.de rm at mamma.varadinet.de
Fri Feb 9 13:43:45 MST 2001


On Fri, Feb 09, 2001 at 01:21:04PM -0700, Calvin Dodge wrote:
> On Fri, Feb 09, 2001 at 09:14:36PM +0100, rm at mamma.varadinet.de wrote:
> > Hello,
> > 
> > i just found something really strange: i observed some funny
> > vm warnings opn a server that went online recently. 
> > Looking at the bootup mesages i realized that the box seems
> > tro come up with only 64Mb of memory as opposed to the 512 
> > it should have. Since the server was shipped by the hardware
> > manufacturer after i installed the OS and they did a bios
> > upgrade prior to shipping it i wasn't shure whether i messed
> > something up or they did or the memory went bad. So i dug out 
> > the tarball i made before shipping and had a look at the log
> > files (see attached excerpts). Now, has anyone on this list
> > seen similar erratic behaviour?
> 
> Yes, on an Abit BP-6 after upgrading the BIOS.
> 
> IIRC, this stems from changes in the way the BIOS reports how much memory is installed.  The BIOS method got out of sync with the method Linux uses to query the BIOS - this has been fixed in the 2.4 kernel.

This was my first guess, but would it change every time the box boots
(as i said, it's down to 64Mb now :-( Of course this is on a really high
traffic webserver.

> You need to tell the kernel how much RAM is installed.  The simplest way is to put
> 
> append="mem=512M"
> 
> in your /etc/lilo.conf file, then rerun lilo.

Hmm, i know, but the box is on the other side of the country and
i just tested on my local box what happens if the kernel _can't_ 
find all this memory ... :-(


 Ralf



More information about the LUG mailing list