Maas deploy is used to configure and deploy bare-metal machines using Canonical's MAAS. It uses in input a yaml file containing the description of the machine we want to deploy. Maas-deploy uses an OS env variable with your API key. The variable is named MAAS_API_KEY.

maas-deploy uses python 3 to run. And the only dependency it has is with MAAS Client Library & CLI (python-libmaas). You can intall it with pip. Just run:

pip install python-libmaas

For more about libmaas, visit: https://maas.github.io/python-libmaas/

export MAAS_API_URL=<Maas url>
export MAAS_API_KEY=<Private API key>

deploy.py my_fancy_machine.yaml

Each machine description starts always with the hostname of the machine. And everything else is child os that root item. The machines are deplared under machines.

Then, various items can be defined:

• os
• os_raid1
• os_raid6
• os_partitions
• kernel
• admin_net
• net_bonding
• user_data
• unused_disks

OS we want to deploy on the node. The options here are the same as using the maas API. For example, use xenial for ubuntu 16.04 or bionic for ubuntu 18.04. Default value if not specifed is the MAAS default.

os: bionic

Only raid 1 and raid 6 for os disks is implemented yet. If the option is set, the entry disks will contain the list of disks to use on the raid cluster. If this item is not specified, the script will try to guess the 2 disks to use as system disks, with raid 1. We can use LVM on top of the raid array as on the example below. name and enable are mandatory under use_lvm. os_partitions need to be set if you want to use lvm.

os_raid1:
    disks:
        - sda
        - sdc
    use_lvm:
        enable: True
        name: lxc

You can define a partition for the os_disks by declaring the mount point. Then, you define the size of the partition and the filesystem. Yet the size and filesystem is mandatory for each mount point, but the filesystem could be optional on future releases.

os_partitions:
    /var:
        size: 20G
        filesystem: ext4
    /tmp:
        size: 20G
        filesystem: btrfs
    /boot:
        size: 1G
        filesystem: ext4
    /:
        size: 59G
        filesystem: xfs
    /home:
        size: 19G
        filesystem: ext4

If you need to use a diferent kernel as the one used by default, for example the hwe one, you can use this option to set it.

kernel: hwe-16.04

You can configure the subnet for the admin_net. This subnet will be configured on the interface where the machine boot with PXE. MaaS will configure the interface with DHCP. If you want not to use, you can set None. Make sure that the machine can reach the MaaS server with other interfaces if you omit admin_net config or set it to None.

admin_net: rack1:admin

Net bounding has 3 parameters, two mandatory and one optional. Mandatory ones are slaves, which contains the network interface for the bond, and name, which will be the bond interface name. You can attach vlans to the bound. The vlans numbers specified need to exist on MaaS. For each vlan, you need to tell the name of the vlan as it exists on MaaS. All the others options are optional. But subnet need to exist on MaaS, and the IP need to exist on the particular subnet. default_gateway will ensure that the configuration attach to it will be the first to be configured. Bond parameters are hard coded yet to bond_mode="802.3ad", bond_lacp_rate="fast", bond_xmit_hash_policy="layer3+4".

net_bonding:
    slaves:
        - enp4s0f0
        - enp4s0f1
    name: bond0
    vlans:
        mgmt:
            vlan: mgmt
            subnet: rack2:mgmt
            ip: 192.0.2.50
            default_gateway: true
            mtu: 9050
        vxlan:
            vlan: vxlan

If you want to personalized you installation, you can use cloud-init configuration. Juste add cloud-init yaml under the user_data field. Example adding some repo:

user_data:
    package_upgrade: true
    apt:
        sources:
            saltstack.list: >
                deb
                https://repo.saltstack.com/apt/ubuntu/18.04/amd64/latest/
                $RELEASE
                main
            keyid: 754A1A7AE731F165D5E6D4BD0E08A149DE57BFBE
    packages:
        - salt-minion

You can configure the non-os disks in two ways. As jbod_disks or as an disk_array. For jbod_disks you need to provide the device, filesystem and mointpoint. For disk_array you need to provide the array for cloud-init command, for example creating a raid. You can add a step2 to it if you want to create a volume group on the raid. For raid array that need a mount point, you can configure it with raid_array. Only Raid6 is supported yet.

Note that jbod_disks and raid_array will always be evaluated befor disk_array, and disk_array will apply on all disks that have neither be used on jbod_disks, raid_array nor os_raid1.

unused_disks:
    jbod_disks:
        - device: /dev/sdc
          fs: ext4
          mountpoint: /data/01
        - device: /dev/sdd
          fs: ext4
          mountpoint: /data/02
    raid_array:
        disks:
            - sdc
            - sdd
            - sde
            - sdf
        fs: ext4
        mountpoint: /media/raid6_data
    disk_array:
         - cloud-init-per
         - once
         - softraid
         - mdadm
         - --create
         - /dev/md1
         - --level=6
         - --raid-devices=4
    step2:
        - cloud-init-per
        - once
        - b_volume-group
        - vgcreate
        - lxc
        - /dev/md1