ceph_wipe_disk.yaml

# Instructions:
# 1. save this file somewhere, for example /home/stack/ceph_wipe_disk.yaml
# 2. create an empty ceph_wipe_env.yaml with the following contents:
#
# resource_registry:
#   OS::TripleO::NodeUserData: /home/stack/ceph_wipe_disk.yaml
# parameter_defaults:
#   ceph_disks: "/dev/sdb /dev/sdc"
#
# 3. when deploying, add to the CLI command the following argument:
#
# -e /path/to/ceph_wipe_env.yaml

heat_template_version: 2014-10-16

parameters:
  ceph_disks:
    type: string
    description: A list of block devices to be cleared with an empty GPT label.
    default: "/dev/sdb /dev/sdc"

description: >
  This is an example showing how you can clear the cephstorage nodes disk with an empty GPT label on firstboot.

resources:
  userdata:
    type: OS::Heat::MultipartMime
    properties:
      parts:
      - config: {get_resource: wipe_disk}

  wipe_disk:
    type: OS::Heat::SoftwareConfig
    properties:
      config:
        str_replace: 
          template: |
            #!/bin/bash
            case "$(hostname)" in
              *cephstorage*)
                for d in $disks; do
                  if [ -b ${d} ]; then
                    echo "Wiping disk ${d} ..."
                    sgdisk -Z ${d}
                    sgdisk -o ${d}
                  fi
                done
              ;;
            esac
          params:
            $disks: {get_param: ceph_disks}

outputs:
  OS::stack_id:
    value: {get_resource: userdata}

Note that this will only work if the user hasn't renamed their ceph nodes. If someone sets their ceph hostname format to my-super-awesome-ceph-node-%index% this isn't going to work properly.

Also, is there any way we could avoid duplicating the disk list? That information should already be available in the ceph metadata, shouldn't it?

hi, I think best option to make sure this is applied only on the relevant nodes is to add role-specific NodeUserData in the parent templates, what do you think?