http://www.dedoimedo.com/computers/crash-analyze.html

http://fedoraproject.org/wiki/How_to_use_kdump_to_debug_kernel_crashes

http://www.ibm.com/developerworks/cn/linux/l-task-killable/index.html

http://www.2cto.com/os/201204/127771.html


kernel.panic:

the seconds  before server rebooting on a panic

但是经过测试 ,panic设置成0,或则 非0,没区别。

echo "c" > /proc/sysrq-trigger

该cmd会trigger一个crash,此时server会hung住(同时产生vmcore文件在 /var/crash),然后reboot,从测试的时间来看,kernel.panic=0,non -0,都是 一样的效果。server在vmcore产生完毕后就自动reboot,根本不会考虑panic的时间,个人感觉也没有必要。


sysrq-trigger is a indirect to set kernel parameter kernel.sysrq


crash need pacakge kernel-debuginfo,  but redhat does not provide the kernel-debuginfo package, you need manually add centos-debuginfo repository

http://debuginfo.centos.org/6/x86_64/

because there are more than one version for kernel-debuginfo in repository above, you need show all versions using:

yum --showduplicates list | grep kernel-debuginfo

then choose the same version as your machine:


yum install kernel-debuginfo-2.6.32-431.el6.x86_64