Ironic Boot-from-Volume with DevStack

This guide shows how to setup DevStack for enabling boot-from-volume feature, which has been supported from the Pike release.

This scenario shows how to setup DevStack to enable nodes to boot from volumes managed by cinder with VMs as baremetal servers.

DevStack Configuration

The following is local.conf that will setup DevStack with 3 VMs that are registered in ironic. A volume connector with IQN is created for each node. These connectors can be used to connect volumes created by cinder. The detailed description for DevStack is at Deploying Ironic with DevStack.

[[local|localrc]]

enable_plugin ironic https://opendev.org/openstack/ironic

IRONIC_STORAGE_INTERFACE=cinder

# Credentials
ADMIN_PASSWORD=password
DATABASE_PASSWORD=password
RABBIT_PASSWORD=password
SERVICE_PASSWORD=password
SERVICE_TOKEN=password
SWIFT_HASH=password
SWIFT_TEMPURL_KEY=password

# Set glance's default limit to be baremetal image friendly
GLANCE_LIMIT_IMAGE_SIZE_TOTAL=5000

# Enable Neutron which is required by Ironic and disable nova-network.
disable_service n-net
disable_service n-novnc
enable_service q-svc
enable_service q-agt
enable_service q-dhcp
enable_service q-l3
enable_service q-meta
enable_service neutron

# Enable Swift for the direct deploy interface.
enable_service s-proxy
enable_service s-object
enable_service s-container
enable_service s-account

# Disable Horizon
disable_service horizon

# Disable Heat
disable_service heat h-api h-api-cfn h-api-cw h-eng

# Swift temp URL's are required for the direct deploy interface.
SWIFT_ENABLE_TEMPURLS=True

# Create 3 virtual machines to pose as Ironic's baremetal nodes.
IRONIC_VM_COUNT=3
IRONIC_BAREMETAL_BASIC_OPS=True
DEFAULT_INSTANCE_TYPE=baremetal

# Enable additional hardware types, if needed.
#IRONIC_ENABLED_HARDWARE_TYPES=ipmi,fake-hardware
# Don't forget that many hardware types require enabling of additional
# interfaces, most often power and management:
#IRONIC_ENABLED_MANAGEMENT_INTERFACES=ipmitool,fake
#IRONIC_ENABLED_POWER_INTERFACES=ipmitool,fake
#IRONIC_DEFAULT_DEPLOY_INTERFACE=direct

# Change this to alter the default driver for nodes created by devstack.
# This driver should be in the enabled list above.
IRONIC_DEPLOY_DRIVER=ipmi

# The parameters below represent the minimum possible values to create
# functional nodes.
IRONIC_VM_SPECS_RAM=1280
IRONIC_VM_SPECS_DISK=10

# Size of the ephemeral partition in GB. Use 0 for no ephemeral partition.
IRONIC_VM_EPHEMERAL_DISK=0

# To build your own IPA ramdisk from source, set this to True
IRONIC_BUILD_DEPLOY_RAMDISK=False

VIRT_DRIVER=ironic

# By default, DevStack creates a 10.0.0.0/24 network for instances.
# If this overlaps with the hosts network, you may adjust with the
# following.
NETWORK_GATEWAY=10.1.0.1
FIXED_RANGE=10.1.0.0/24
FIXED_NETWORK_SIZE=256

# Log all output to files
LOGFILE=$HOME/devstack.log
LOGDIR=$HOME/logs
IRONIC_VM_LOG_DIR=$HOME/ironic-bm-logs

After the environment is built, you can create a volume with cinder and request an instance with the volume to nova:

# set up the user for authentication purposes
# note that all users can be seen in /etc/openstack/clouds.yaml
export OS_CLOUD=devstack-admin-demo

# query the image id of the default cirros image
image=$(openstack image show $DEFAULT_IMAGE_NAME -f value -c id)

# create keypair
ssh-keygen
openstack keypair create --public-key ~/.ssh/id_rsa.pub default

# create volume
volume=$(openstack volume create --image $image --size 1 my-volume -f value -c id)

# spawn instance
openstack server create --flavor baremetal --volume $volume --key-name default testing

You can also run an integration test that an instance is booted from a remote volume with tempest in the environment:

cd /opt/stack/tempest
tox -e venv-tempest -- pip install (path to the ironic-tempest-plugin directory)
tox -e all -- ironic_tempest_plugin.tests.scenario.test_baremetal_boot_from_volume

Please note that the storage interface will only indicate errors based upon the state of the node and the configuration present. As such a node does not exclusively have to boot via a remote volume, and as such validate actions upon nodes may be slightly misleading. If an appropriate volume target is defined, no error should be returned for the boot interface.