[Novalug] Red Hat Ent 4 kernel upgrades lose working memory ?
John A. Heidingsfelder Jr.
john at heidingsfelder.org
Tue Nov 7 23:32:44 EST 2006
If they are identical machines with identical motherboards/BIOS levels
then it should have come up if you are running identical kernels. The
ram shown/released is determined by the motherboard and bios. How much
the system grabs at boot and the like. If you have some HBA cards or
other differing peripherals, it may grab more system memory for them.
And you would therefore see less via the OS/kernel. At 16 or 32 GB's of
ram, it's not a kernel limit issue unless you aren't using the BIGMEM
kernel- which is what you use at that size.
On Tue, 2006-11-07 at 19:40 -0500, Kevin Dwyer wrote:
> On Tue, Nov 07, 2006 at 04:42:58PM -0500, moliver at mmoliver.org wrote:
> > We recently upgraded the kernel from 2.6.9-11 to 2.6.9-42.0.3 and 3
> > of the 5 did not recognize the entire 16GB after reboot.
> > Has anyone seen this type of behavior before? I've been looking and
> > doing some searching and haven't had much luck.
> > Any clues?
> Kernels can/do have compiled-in limits on maximum main memory. It's
> highly unusual for two identical machines to come up differently with
> identical kernels though.
> Have you combed through dmesg? Probably even diffing the output of
> dmesg for each kernel will be helpful. You're most interested in the
> earlier parts.
> Novalug mailing list
> Novalug at calypso.tux.org
More information about the Novalug