UnixPedia : HPUX / LINUX / SOLARIS: 2019

Monday, January 14, 2019

Configuration of kdump on VMs is the same as on physical servers.

.  Full details about configuring and testing kdump on servers (both physical and virtual) are available here:

   How to troubleshoot kernel crashes, hangs, or reboots with kdump
   on Red Hat Enterprise Linux

Red Hat also has a simple tool to help configure kdump on a system
here:

   Kdump Helper

The configuration of kdump on VMs is the same as on physical servers.
You essentially need:
- Reservation of memory for the crashkernel (GRUB "crashkernel="
   setting),
- Kdump configuration for location and filtering (/etc/kdump.conf) -
   make sure enough free disk space is available in the location
   specified.
- Configuration parameters may be desired to enable forced crashes on
   hung systems:
      kernel.sysrq=1
      kernel.unknown_nmi_panic=1
      kernel.panic_on_unrecovered_nmi=1
      kernel.panic_on_io_nmi=1
- If forced crashes will be performed, the process for crashing hung
   VMware VMs is here:
      How to Force a Crash on a Hung Linux Virtual Machine Running
      on VMware ESXi Server

I'll note that it is very important that kdump be tested to ensure it works correctly and can save a vmcore file.