You can use the below instructions to configure kdump on a Bright managed cluster. These instructions should work on RHEL/CentOS 7 and 8.
1. Install required packages
Install the kexec-tools in the software image:# yum install kexec-tools --installroot=/cm/images/<image-name>
2. Modify the software image
Configure the software image to allow crashkernel memory and update the exclude list to avoid wiping out the crash path:# cmsh
% softwareimage use <image-name>
% append kernelparameters " crashkernel=128M" # please make sure that the append starts with a blank space
% commit
% category use <category-name>
% set excludelistsyncinstall
(add the following lines to exclude; save changes and exit)
- /var/crash/*
no-new-files: - /var/crash/*
% commit
3. Enable the kdump service in the software image:
# chroot /cm/images/<image name>
# chkconfig kdump on
4. Confirm kdump is configured correctly
Make sure that the kdump service is active when the node comes up after reboot.[root@node001 ~]# cat /proc/cmdline
BOOT_IMAGE=images/default-image/vmlinuz
initrd=images/default-image/initrd crashkernel=128M
ip=10.141.0.1:10.141.255.254:10.141.255.254:255.255.0.0
BOOTIF=01-08-00-27-0c-4c-ca
[root@node001 ~]# service kdump status
Kdump is operational
5. Generated a kernel panic to test.
You can then try to generate a kernel panic with "echo c > /proc/sysrq-trigger" and see if the crash report will be generated successfully "WARNING this will crash the node and cause a kernel panic":