Red Hat Bugzilla – Bug 240010
Kexec/Kdump fails with RHEL5 on ES7000 ia64 EL Servers.
Last modified: 2009-09-09 01:12:26 EDT
Description of problem:
Kexec/Kdump fails with RHEL5 and more steps are described in the HOW TO
Version-Release number of selected component (if applicable):
OS LEVEL :RHEL5
Kexec Tool Used :Server/kexec-tools-1.101-164.el5.ia64.rpm
Steps to Reproduce:
1.crash kernel memory was allocated with 512M@256M at install time
2.Rebooted the partition and execute
kexec -l /boot/vmlinuz-`uname -r` --initrd=/boot/initrd-`uname -r`.img --
3.Tried to crash the kernel using
echo 1 > /proc/sys/kernel/sysrq
echo c > /proc/sysrq-trigger
The crash Kernel image is not seen after reboot in /var/crash.
The crash Kernel image has to be found in the /var/crash
Does base kernel work or not?
Is the kdump service enabled? You need to 'chkconfig kdump on' and 'service
kdump start' to make sure the kdump environment actually gets set up. A manual
kexec -l only loads a kernel for rebooting, not for panic dump.
(In reply to comment #1)
> Does base kernel work or not?
By the steps to reproduce, we used the base kernel to kexec and it didn't work.
(In reply to comment #2)
> Is the kdump service enabled? You need to 'chkconfig kdump on' and 'service
> kdump start' to make sure the kdump environment actually gets set up. A manual
> kexec -l only loads a kernel for rebooting, not for panic dump.
We didn't check the kdump part when we performed the steps in the bugzilla. We
want to get the second kernel to boot when the first kernel panics. After we
get to reboot, we can try the dump feature.
Is kdump necessary to get kexec working?
Just a 'kexec -l' won't work, that only specifies a kernel to reboot into via a
clean kexec reboot. You need to load a kernel with 'kexec -p' (-p = panic
kernel) for it to be used for an automatic reboot following a panic (this is
part of what the kdump initscript does). You also must have a crashkernel memory
region reserved for the panic kernel to use. Please see the HOWTO included in
our kexec-tools rpm.
Today we tried the kexec/kdump tests on RHEL5. All the tests were run in
runlevel 3(Non-GUI mode)
1. RHEL5 Single Processor Results
i. Kexec worked fine with kexec -l and kexec -e commands.
ii.kdump gave a kernel panic with kexec -p command. Also, I executed a crash
without the kexec -p but there were PROCNODE errors.
2. RHEL5 Multi Processor Results
i. Kexec failed with Procnode error with kexec -l and kexec -e without kdump
ii.Kdump failed with Procnode error when a crash was initiated.
Neil, any of this seem familiar to you at all? This is a new one to me...
No, Can we please get:
1) /etc/kdump.conf from this system
2) the initramfs that kdump produces which gets loaded on kdump service start
3) serial console log of kexec kernels boot process
ping. Any update here?
ping. Is it safe to close?
(In reply to comment #10)
> ping. Is it safe to close?
Ya you can close this issue.