storage.md

lxd::storage

This Puppet defined-type handles lxd storage create/update/delete. The type is a wrapper for the lxd_storage type.

Exact parameters and what they mean can be found in REFERENCE.md under lxd::storage section.

Examples for single node LXD server

Create an LXD dir storage on a single node

lxd::storage { 'dir-storage':
  ensure      => present,
  driver      => 'dir',
  description => 'Managed by Puppet'
}

Create an LXD dir storage on a single node, with some configration values.

lxd::storage { 'dir-storage':
  ensure      => present,
  driver      => 'dir',
  description => 'Managed by Puppet',
  config      => {
    'volume.size' => '5GB'
  }
}

Create an LXD lvm storage on a single node.

lxd::storage { 'lvm-storage':
  ensure      => present,
  driver      => 'lvm',
  description => 'Managed by Puppet',
  source      => '/dev/sdb'
}

Examples for an LXD cluster member

When setting up storage in an LXD cluster, things become more tricky. The manual process goes as this:

1. Create the storage-pool on each LXD member via --target flag.
2. Commit the storage-pool as Created by running the create command without the --target flag.

This process also accepts some configuration variables during Step 1 and others on Step 2.

As such, it is expected that when running all the examples below you will first bring each member's storage-pool into Pending state by running the puppet agent on each node.

Afterwards, on any of the members nodes, you need to run the puppet agent to bring the storage-pool into a Created state.

Create an LXD dir storage-pool in a cluster

lxd::storage { 'dir-storage':
  ensure      => present,
  driver      => 'dir',
  description => 'Managed by Puppet'
}

Create an LXD lvm storage-pool in a cluster

lxd::storage { 'lvm-storage':
  ensure      => present,
  driver      => 'lvm',
  description => 'Managed by Puppet',
  source      => '/dev/sdb'
}

Known issues & limitations

Pool not defined on nodes: $node-name

Example error message: returned 1: Error: Pool not defined on nodes: node02

This error will occur when you are setting up a storage-pool in a clustered environment. You have ran the storage creation on node01 and then again ran puppet on the same node, before you created the storage-pool on the other nodes.

To tackle this, run the storage-pool create on all cluster members, before running the agent again to put the pool in Created stage.

Removing a storage config value

If you have defined a config value that you want to remove, just excluding it from the config has will not actually remove it from the server OR reset it to the default value.

Example:

# original storage
lxd::storage { 'default':
  ensure => present,
  driver => 'lvm',
  config => {
    'lvm.thinpool_name' => 'example'
  }
}

$ lxc storage get lvm.thinpool_name
example

# change storage
lxd::storage { 'default':
  ensure => present,
  driver => 'lvm',
  config => {}
}

$ lxc storage get lvm.thinpool_name
example

While this is inconvenient in some cases, it's implemented like this to avoid Puppet messing up the storage-pool.