Ansible Role: Kernel Crash Dump

This is an Ansible role that sets up kdump.

Warning

This role will change the kdump settings on the managed host. Any previous settings will be removed, even if they are not mentioned in the role variables. It will at least change this configuration file:

• /etc/kdump.conf

Requirements

Please see below.

Collection Requirements

This role only needs external collections for managing rpm-ostree nodes. To install them, run this command if you need to manage rpm-ostree nodes:

ansible-galaxy collection install -vv -r meta/collection-requirements.yml

Role Variables

kdump_target: Specifies where to save vmcore. If type is raw or a filesystem type, the location refers to a partition (by device node name, label, or uuid). For example:

kdump_target:
  type: raw
  location: /dev/sda1

Or for an ext4 filesystem:

kdump_target:
  type: ext4
  location: "12e3e25f-534e-4007-a40c-e7e080a933ad"

If type is ssh, the location refers to a server, for example:

  type: ssh
  location: [email protected]

For nfs, it will look like this:

  type: nfs
  location: nfs.example.com

Only the ssh type is considered stable; the others are experimental.

kdump_path: The path to where vmcore will be written. If kdump_target is set, the path is relative to that target. Otherwise, it must be an absolute path in the root filesystem.

kdump_core_collector: This command is used to copy the vmcore. If left empty, it will use makedumpfile with options depending on the kdump_target.type.

kdump_system_action: This specifies what to do if dumping the core file fails. Options are reboot, halt, poweroff, or shell.

kdump_auto_reset_crashkernel: This decides whether to reset the crashkernel to a new default value when kexec-tools updates it, affecting existing kernels that use the old value.

kdump_dracut_args: Add extra dracut options when rebuilding the kdump initrd.

kdump_reboot_ok: If the managed node doesn't have memory reserved for the crash kernel (checked by /sys/kernel/kexec_crash_size which shows 0), a reboot may be necessary to set up kdump.

By default, the role does not reboot the node. If a reboot is needed, the role sets the kdump_reboot_required fact and fails, allowing the user to reboot the node. To have the role reboot the system if needed, set this variable to true. You won't need to run the role again after rebooting.

Default: false

Ansible Facts Returned by the Role

kdump_reboot_required: This fact is set if the managed node needs a reboot to finish the kdump setup. Run the role again after rebooting to make sure kdump is working.

rpm-ostree

Check README-ostree.md for more information.

License

MIT