This role allows users to configure local storage with minimal input.
As of now, the role supports managing file systems and mount entries on
- unpartitioned disks
- lvm (unpartitioned whole-disk physical volumes only)
See meta/requirements.yml
for the requirements. You must install the
requirements before using this role:
ansible-galaxy collection install -vv -r meta/requirements.yml
See https://docs.ansible.com/ansible/latest/galaxy/user_guide.html#using-meta-requirements-yml for more information.
NOTE: Beginning with version 1.3.0, unspecified parameters are interpreted
differently for existing and non-existing pools/volumes. For new/non-existent
pools and volumes, any omitted parameters will use the default value as
described in defaults/main.yml
. For existing pools and volumes, omitted
parameters will inherit whatever setting the pool or volume already has.
This means that to change/override role defaults in an existing pool or volume,
you must explicitly specify the new values/settings in the role variables.
The storage_pools
variable is a list of pools to manage. Each pool contains a
nested list of volume
dicts as described below, as well as the following
keys:
This specifies the name of the pool to manage/create as a string. (One example of a pool is an LVM volume group.)
This specifies the type of pool to manage.
Valid values for type
: lvm
.
A list which specifies the set of disks to use as backing storage for the pool.
Supported identifiers include: device node (like /dev/sda
or /dev/mapper/mpathb
),
device node basename (like sda
or mpathb
), /dev/disk/ symlink
(like /dev/disk/by-id/wwn-0x5000c5005bc37f3f
).
When used with type: lvm
it manages a volume group with a mdraid array of given level
on it. Input disks
are in this case used as RAID members.
Accepted values are: linear
, striped
, raid0
, raid1
, raid4
, raid5
, raid6
, raid10
This is a list of volumes that belong to the current pool. It follows the
same pattern as the storage_volumes
variable, explained below.
This specifies whether the pool will be encrypted using LUKS. WARNING: Toggling encryption for a pool is a destructive operation, meaning the pool itself will be removed as part of the process of adding/removing the encryption layer.
This string specifies a password or passphrase used to unlock/open the LUKS volume(s).
This string specifies the full path to the key file on the managed nodes used to unlock the LUKS volume(s). It is the responsibility of the user of this role to securely copy this file to the managed nodes, or otherwise ensure that the file is on the managed nodes.
This string specifies a non-default cipher to be used by LUKS.
This integer specifies the LUKS key size (in bytes).
This integer specifies the LUKS version to use.
The storage_volumes
variable is a list of volumes to manage. Each volume has the following
variables:
This specifies the name of the volume.
This specifies the type of volume on which the file system will reside.
Valid values for type
: lvm
, disk
or raid
.
The default is determined according to the OS and release (currently lvm
).
This specifies the set of disks to use as backing storage for the file system.
This is currently only relevant for volumes of type disk
, where the list
must contain only a single item.
The size
specifies the size of the file system. The format for this is intended to
be human-readable, e.g.: "10g", "50 GiB". The size of LVM volumes can be specified as a
percentage of the pool/VG size, eg: "50%" as of v1.4.2.
When using compression
or deduplication
, size
can be set higher than actual available space,
e.g.: 3 times the size of the volume, based on duplicity and/or compressibility of stored data.
NOTE: The requested volume size may be reduced as necessary so the volume can fit in the available pool space, but only if the required reduction is not more than 2% of the requested volume size.
This indicates the desired file system type to use, e.g.: "xfs", "ext4", "swap".
The default is determined according to the OS and release
(currently xfs
for all the supported systems).
The fs_label
is a string to be used for a file system label.
The fs_create_options
specifies custom arguments to mkfs
as a string.
The mount_point
specifies the directory on which the file system will be mounted.
The mount_options
specifies custom mount options as a string, e.g.: 'ro'.
Specifies RAID level. LVM RAID can be created as well.
"Regular" RAID volume requires type to be raid
.
LVM RAID needs that volume has storage_pools
parent with type lvm
,
raid_disks
need to be specified as well.
Accepted values are: linear
(N/A for LVM RAID), striped
, raid0
, raid1
, raid4
, raid5
, raid6
, raid10
WARNING: Changing raid_level
for a volume is a destructive operation, meaning
all data on that volume will be lost as part of the process of
removing old and adding new RAID. RAID reshaping is currently not
supported.
When type is raid
specifies number of active RAID devices.
When type is raid
specifies number of spare RAID devices.
When type is raid
specifies RAID metadata version as a string, e.g.: '1.0'.
When type is raid
specifies RAID chunk size as a string, e.g.: '512 KiB'.
Chunk size has to be multiple of 4 KiB.
Specifies which disks should be used for LVM RAID volume.
raid_level
needs to be specified and volume has to have storage_pools
parent with type lvm
.
Accepts sublist of disks
of parent storage_pools
.
In case multiple LVM RAID volumes within the same storage pool, the same disk can be used
in multiple raid_disks
.
This specifies whether the volume will be encrypted using LUKS. WARNING: Toggling encryption for a volume is a destructive operation, meaning all data on that volume will be removed as part of the process of adding/removing the encryption layer.
This string specifies a password or passphrase used to unlock/open the LUKS volume.
This string specifies the full path to the key file on the managed nodes used to unlock the LUKS volume(s). It is the responsibility of the user of this role to securely copy this file to the managed nodes, or otherwise ensure that the file is on the managed nodes.
This string specifies a non-default cipher to be used by LUKS.
This integer specifies the LUKS key size (in bits).
This integer specifies the LUKS version to use.
This specifies whether the Virtual Data Optimizer (VDO) will be used.
When set, duplicate data stored on storage volume will be
deduplicated resulting in more storage capacity.
Can be used together with compression
and vdo_pool_size
.
Volume has to be part of the LVM storage_pool
.
Limit one VDO storage_volume
per storage_pool
.
Underlying volume has to be at least 9 GB (bare minimum is around 5 GiB).
This specifies whether the Virtual Data Optimizer (VDO) will be used.
When set, data stored on storage volume will be
compressed resulting in more storage capacity.
Volume has to be part of the LVM storage_pool
.
Can be used together with deduplication
and vdo_pool_size
.
Limit one VDO storage_volume
per storage_pool
.
When Virtual Data Optimizer (VDO) is used, this specifies the actual size the volume
will take on the device. Virtual size of VDO volume is set by size
parameter.
vdo_pool_size
format is intended to be human-readable,
e.g.: "30g", "50GiB".
Default value is equal to the size of the volume.
This specifies whether the volume should be cached or not. This is currently supported only for LVM volumes where dm-cache is used. NOTE: Only creating new cached volumes and removing cache from an existing volume is currently supported. Enabling cache for an existing volume is not yet supported.
Size of the cache. cache_size
format is intended to be human-readable,
e.g.: "30g", "50GiB".
Mode for the cache. Supported values include writethrough
(default) and writeback
.
List of devices that will be used for the cache. These should be either physical volumes or drives these physical volumes are allocated on. Generally you want to select fast devices like SSD or NVMe drives for cache.
When true (the default), an error will occur instead of automatically removing existing devices and/or formatting.
When true (the default is false), the role will use udevadm trigger to cause udev changes to take effect immediately. This may help on some platforms with "buggy" udev.
- hosts: all
roles:
- name: linux-system-roles.storage
storage_pools:
- name: app
disks:
- sdb
- sdc
volumes:
- name: shared
size: "100 GiB"
mount_point: "/mnt/app/shared"
#fs_type: xfs
state: present
- name: users
size: "400g"
fs_type: ext4
mount_point: "/mnt/app/users"
storage_volumes:
- name: images
type: disk
disks: ["mpathc"]
mount_point: /opt/images
fs_label: images
MIT