Note
This feature is experimental at this time and it has not been fully production tested yet. These implementation instructions assume that ironic is being deployed as the sole hypervisor for the region.
Ironic is an OpenStack project which provisions bare metal (as opposed to virtual) machines by leveraging common technologies such as PXE boot and IPMI to cover a wide range of hardware, while supporting pluggable drivers to allow vendor-specific functionality to be added.
OpenStack’s ironic project makes physical servers as easy to provision as virtual machines in a cloud.
Modify the environment files and force nova-compute to run from within a container:
sed -i '/is_metal.*/d' /etc/openstack_deploy/env.d/nova.yml
In a general case, neutron networking can be a simple flat network. However, in a complex case, this can be whatever you need and want. Ensure you adjust the deployment accordingly. The following is an example:
neutron net-create cleaning-net --shared \
--provider:network_type flat \
--provider:physical_network ironic-net
neutron subnet-create ironic-net 172.19.0.0/22 --name ironic-subnet
--ip-version=4 \
--allocation-pool start=172.19.1.100,end=172.19.1.200 \
--enable-dhcp \
--dns-nameservers list=true 8.8.4.4 8.8.8.8
Images using the diskimage-builder must be built outside of a container. For this process, use one of the physical hosts within the environment.
Install the necessary packages:
apt-get install -y qemu uuid-runtime curl
Install the disk-imagebuilder package:
pip install diskimage-builder --isolated
Important
Only use the --isolated flag if you are building on a node deployed by OpenStack-Ansible, otherwise pip will not resolve the external package.
Optional: Force the ubuntu image-create process to use a modern kernel:
echo 'linux-image-generic-lts-xenial:' > \
/usr/local/share/diskimage-builder/elements/ubuntu/package-installs.yaml
Create Ubuntu initramfs:
disk-image-create ironic-agent ubuntu -o ${IMAGE_NAME}
Upload the created deploy images into the Image (glance) Service:
# Upload the deploy image kernel
glance image-create --name ${IMAGE_NAME}.kernel --visibility public \
--disk-format aki --container-format aki < ${IMAGE_NAME}.kernel
# Upload the user image initramfs
glance image-create --name ${IMAGE_NAME}.initramfs --visibility public \
--disk-format ari --container-format ari < ${IMAGE_NAME}.initramfs
Create Ubuntu user image:
disk-image-create ubuntu baremetal localboot local-config dhcp-all-interfaces grub2 -o ${IMAGE_NAME}
Upload the created user images into the Image (glance) Service:
# Upload the user image vmlinuz and store uuid
VMLINUZ_UUID="$(glance image-create --name ${IMAGE_NAME}.vmlinuz --visibility public --disk-format aki --container-format aki < ${IMAGE_NAME}.vmlinuz | awk '/\| id/ {print $4}')"
# Upload the user image initrd and store uuid
INITRD_UUID="$(glance image-create --name ${IMAGE_NAME}.initrd --visibility public --disk-format ari --container-format ari < ${IMAGE_NAME}.initrd | awk '/\| id/ {print $4}')"
# Create image
glance image-create --name ${IMAGE_NAME} --visibility public --disk-format qcow2 --container-format bare --property kernel_id=${VMLINUZ_UUID} --property ramdisk_id=${INITRD_UUID} < ${IMAGE_NAME}.qcow2
Create a new flavor called my-baremetal-flavor.
Note
The following example sets the CPU architecture for the newly created flavor to be x86_64.
nova flavor-create ${FLAVOR_NAME} ${FLAVOR_ID} ${FLAVOR_RAM} ${FLAVOR_DISK} ${FLAVOR_CPU}
nova flavor-key ${FLAVOR_NAME} set cpu_arch=x86_64
nova flavor-key ${FLAVOR_NAME} set capabilities:boot_option="local"
Note
Ensure the flavor and nodes match when enrolling into ironic. See the documentation on flavors for more information: http://docs.openstack.org/openstack-ops/content/flavors.html
After successfully deploying the ironic node on subsequent boots, the instance boots from your local disk as first preference. This speeds up the deployed node’s boot time. Alternatively, if this is not set, the ironic node PXE boots first and allows for operator-initiated image updates and other operations.
Note
The operational reasoning and building an environment to support this use case is not covered here.
From the utility container, enroll a new baremetal node by executing the following:
# Source credentials
. ~/openrc
# Create the node
NODE_HOSTNAME="myfirstnodename"
IPMI_ADDRESS="10.1.2.3"
IPMI_USER="my-ipmi-user"
IPMI_PASSWORD="my-ipmi-password"
KERNEL_IMAGE=$(glance image-list | awk "/${IMAGE_NAME}.kernel/ {print \$2}")
INITRAMFS_IMAGE=$(glance image-list | awk "/${IMAGE_NAME}.initramfs/ {print \$2}")
ironic node-create \
-d agent_ipmitool \
-i ipmi_address="${IPMI_ADDRESS}" \
-i ipmi_username="${IPMI_USER}" \
-i ipmi_password="${IPMI_PASSWORD}" \
-i deploy_ramdisk="${INITRAMFS_IMAGE}" \
-i deploy_kernel="${KERNEL_IMAGE}" \
-n ${NODE_HOSTNAME}
# Create a port for the node
NODE_MACADDRESS="aa:bb:cc:dd:ee:ff"
ironic port-create \
-n $(ironic node-list | awk "/${NODE_HOSTNAME}/ {print \$2}") \
-a ${NODE_MACADDRESS}
# Associate an image to the node
ROOT_DISK_SIZE_GB=40
ironic node-update $(ironic node-list | awk "/${IMAGE_NAME}/ {print \$2}") add \
driver_info/deploy_kernel=$KERNEL_IMAGE \
driver_info/deploy_ramdisk=$INITRAMFS_IMAGE \
instance_info/deploy_kernel=$KERNEL_IMAGE \
instance_info/deploy_ramdisk=$INITRAMFS_IMAGE \
instance_info/root_gb=${ROOT_DISK_SIZE_GB}
# Add node properties
# The property values used here should match the hardware used
ironic node-update $(ironic node-list | awk "/${NODE_HOSTNAME}/ {print \$2}") add \
properties/cpus=48 \
properties/memory_mb=254802 \
properties/local_gb=80 \
properties/size=3600 \
properties/cpu_arch=x86_64 \
properties/capabilities=memory_mb:254802,local_gb:80,cpu_arch:x86_64,cpus:48,boot_option:local
Important
You will not have access unless you have a key set within nova before your ironic deployment. If you do not have an ssh key readily available, set one up with ssh-keygen.
nova keypair-add --pub-key ~/.ssh/id_rsa.pub admin
Now boot a node:
nova boot --flavor ${FLAVOR_NAME} --image ${IMAGE_NAME} --key-name admin ${NODE_NAME}