RHEL3 U5's kernel is unstable for 8-way SMP server?

jOe smartjoe at gmail.com
Tue May 16 20:23:56 UTC 2006


Hello all,

I met a "system crash" probelm yesterday after processor replacement.
The server is a HP DL760g2 8way server ,originally equipped with
4x2.7GHzXeon CPU and replaced by 8xXEON
3.0GHz  yesterday.
The upgrading service are provided by HP team and this server passed the
status checking and it's system ROM has been updated to latest version as
well.

We use this server with 4 old CPUs for long time, the system is very stable,
it  has RHEL AS 3.0 U5 installed (the kernel is 2.4.21-32.ELsmp).
The crash happend 1 hours later after finished processor upgrading. We've
also checked the System ROM setting : the HT & Full Table APIC are enabled
by default.

I googled but got few information by searching  it's crash information, just
wonder how to search the redhat's bugzilla database so that i can have
further information to confirm if it is not U5's kernel bring us the
trouble.

The following is the crash inoformation that appear on console.

=======================cut=====================================
Red Hat Enterprise Linux As release 3 ( U5)
Kernel  2.4.21-32.ELsmp on an i686

jydbserver1 login: CPU 4: Machine Check Exception:000000000000000000004
<<>CP>#CPU 7: Machine Check Exception:000000000000000000<4>04
<U0>Ke0>Kerconnue
nic: U<4>nablea<0Exe  1: Machine Check ExceptionionExce00ion:<40>Kleon
ItanieU
ablani : U<<4>t4>n40>Kleon ItanieU
ablani : U<<4>t4>>ncont<0>blieernel<<4<>y#4>Yevb a huam#rodbwaril #pr hhyove
a
Radhw##############################################################
###################################################################
####################################################################
#######################

====================end cut==========================================

It is a kernel bug in U5 cause the system crash?
Or something we should set in system ROM or recompile the current kernel ?
btw: The hardware(including server and it's processor/memory, hardisk,
controller ...) has been checked again & again.

Appreciate if any opinion.

-- 
Joe Yu
Senior Linux Solution Architect
LinuxTea Professional Group



More information about the redhat-list mailing list