zfs-storage

This role installs ZFS on Linux and configures filesystems and ZVOLs. nfs-kernel-server will be installed if a configured filesystem makes use of the sharenfs attribute. NFS exports will be permanently configured in /etc/exports.d/zfs-storage-ansible.exports if zfs_save_nfs_exports is set accordingly. The ZFS PPAs from JonathanF are used to install the latest ZFS release.

Requirements

One of the following distribution releases:

• Debian stretch or buster
• Ubuntu xenial, bionic, cosmic or devel

systemd is required in order to scrub pools using timers instead of distribution-specific cron jobs. python-jmespath is needed on the machine executing this role to correctly determine the need for NFS.

Role Variables

Name	Default / Mandatory	Description
zfs_parent_fs		Existing parent ZFS filesystem for all filesystems and zvols that is optionally prepended to all configurations
zfs_storage_defaults	{}	Dict containing ZFS attributes that will be applied to all configured ZFS filesystems/zvols
zfs_filesystems	[]	List of zfs_filesystems defined by a name and a dict of attributes (mandatory for each entry)
zfs_zpools	[]	List of dicts having a name (mandatory) and scrub (optional) key. scrub is a boolean and can disable scrubbing for this pool.
zvols	[]	List of zvols defined by a name and a dict of attributes (mandatory for each entry)
zfs_default_enable_scrub	True	Whether to enable zpool scrubbing by default
zfs_scrub_frequency	weekly	Frequency of zfs scrubbing - already running scrubs will be aborted. See systemd.time for allowed values.
zfs_kernel_module_parameters	{}	Dict containing ZFS kernel module options to be set in /etc/modprobe.d/zfs.conf.
zfs_storage_drive_wwids	[]	List of WWID strings of blockdevices in the pool. UDEV rules will be written to set the IO scheduler to noop for those devices. This is needed for partitioned or encrypted devices as ZFS only sets the scheduler to noop if it gets ahold of the whole drive.
zfs_save_nfs_exports	True	Whether to save and delete (if not needed) a NFS export file for consistent exports across reboots.

Note: There are some ZFS attributes that can only be set at creation (see man zfs). These are utf8only, normalization and casesensitivity for filesystems and volsize and volblocksize for ZVOLs. The respective task will fail if you try to change those. Also, volsize is a mandatory attribute for ZVOLs.

If an attribute is not defined, the ZFS default will be configured (even if the attribute is already set to something else), except for these four attributes:

• acltype=posixacl
• compression=on
• relatime=on
• xattr=sa

All default values for ZFS attributes can be seen in the defaults.

In addition to the actual ZFS attributes described in the man page, this role sets attributes to control automatic snapshotting using tools such as zfs-auto-snapshot. This is achieved using the following variables, which are all set to False by default:

ZFS attribute	Variable name in this role	Description
com.sun:auto-snapshot	com_sun_auto_snapshot	Enable automatic snapshotting and cleaning (destroying) of snapshots using all available intervals for the current dataset (basically acts as if all the following attributes are set to True).
com.sun:auto-snapshot:frequent	com_sun_auto_snapshot_frequent	Enable snapshotting and cleaning in 15min intervals.
com.sun:auto-snapshot:hourly	com_sun_auto_snapshot_hourly	Enable snapshotting and cleaning in hourly intervals.
com.sun:auto-snapshot:daily	com_sun_auto_snapshot_daily	Enable snapshotting and cleaning in daily intervals.
com.sun:auto-snapshot:weekly	com_sun_auto_snapshot_weekly	Enable snapshotting and cleaning in weekly intervals.
com.sun:auto-snapshot:monthly	com_sun_auto_snapshot_monthly	Enable snapshotting and cleaning in monthly intervals.

Example Playbook

- hosts: zfsstorage
  roles:
    - role: zfs-storage
      zfs_parent_fs: tank
      zfs_zpools:
        - name: rpool
          scrub: False
        - name: tank
          scrub: False
      zfs_scrub_frequency: monthly
      zfs_kernel_module_parameters:
        zfs_arc_max: 30064771072 # Allow ARC to grow to 30GB
      zfs_storage_defaults:
        acltype: posixacl
        volsize: 50G
        quota: 50G
      zfs_filesystems:
        - name: testing
          attributes:
            quota: 200G
        - name: testing/wiki
          attributes:
            sharenfs: [email protected]
            compression: off
      zvols:
        - name: testing/dns01
          attributes:
            volsize: 100G
        - name: testing/ldap01
          attributes:

Result

Name	Type	Attributes
tank/testing	filesystem	acltype=posixacl, quota=200G
tank/testing/wiki	filesystem	acltype=posixacl, sharenfs=[email protected], compression=off, quota=50G
tank/testing/dns01	zvol	volsize=100G
tank/testing/ldap01	zvol	volsize=50G

rpool won't be scrubbed, tank will be scrubbed monthly.

License

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Author Information

• Michel Weitbrecht (SlothOfAnarchy) [email protected]