3

Trying kdump on CentOS 6.3.

On VMWare everything is fine and on oops crash kernel is loaded and vmcore created.

On phisical machine I see oops message, but crash kernel is not started.

I used kexec to load kernel, kdump loads. I used "kexec --load" instead of "kexec --load-panic" as kdump does. After that I run "kexec -e" to load into new kernel. On console I see message "Starting new kernel" forever.

Any ideas?

2

If you only want to configure kdump the redhat way, then you don't need to use kexec at all. Everything is integrated into kdump deamon itself.

Simple steps.

1) Download and install corresponding kernel-debuginfo and kernel-debuginfo-common

2) yum install kexec-tools crash

3) edit the /etc/kdump.conf file . eg if you want to push the vmcore file to remote machine then add

net root@10.2.2.222
core_collector makedumpfile -d 31 -c

4) Add

crashkernel=256M@16M

in /etc/grub.conf in kernel command line option

5)reboot the machine/vm

6) Start the kdump server

service kdump start 

It will create the dump kernel image to load when crash happens.

7) Now test it

echo c > /proc/sysrq-trigger
0

I had similar issue when reserved area for crash kernel is small (I've used crashkernel=64M). Try to increase crashkernel value. It should be specified as kernel boot parameter in your favourite bootloader.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.