Deploy sanoid for ZFS snapshot
management and Debian-based systems (Debian/Ubuntu/Proxmox). Also installs the
other tools in the repo: syncoid
(magical), findoid
, and sleepymutex
to
/usr/local/sbin
None.
# defaults
sanoid_version: "v2.2.0"
sanoid_run_each_x_minutes: 15
First variable is the version to pull from github
Second variable determies how regulary the systemd timer should run per hour. For example, a value of 4 means to run every 4th minute of each hour. This timer runs sanoid to check if snapshots need to be taken/pruned. When using high 'frequently' values you should adjust this value down so more snapshots can be taken.
Each of your snapshot templates are defined in the sanoid_templates
varialble. Below are a few example values but the
defaults/main.yaml
has all of the templates Jim Salter defines in the sample
sanoid.conf on
GitHub.
If you want to extend these values I recommend copying the default version of the variable and adding/editing your own templates as playbook variables.
sanoid_templates:
- name: docker
frequently: 3
hourly: 48
daily: 30
monthly: 6
- name: archive
daily: 5
weekly: 6
monthly: 12
yearly: 3
sanoid_backup_modules:
- name: zpoolname/docker
template: docker
recursive: yes
process_children_only: yes
- name: zpoolname/datasetORzvol
template: archive
daily: 3
recursive: yes
The 'sanoid_backup_modules' define which datasets/zvols are going to be managed by sanoid and which template to use as a snapshot policy. You can define multiple template by comma separating them. For example, 'docker,archive' would process those templates in order and apply the end result.
You can also override any given value by passing as a key value pair. Above the 'zpoolname/datasetORzvol' dataset will have 3 daily snapshots.
Learn more about the configuration values on sanoid github. Not all of the available values have been added to the template. If you need anything not here add it to the role and send a pull request. I may revisit this in the future but for now the role does all I need.
None.
- name: Setup sanoid to auto-snapshot your datasets
hosts: all
become: true
roles:
- role: gikeymarcia.sanoid
vars:
sanoid_backup_modules:
- name: tank/vms
template: vm_live
recursive: 'yes'
process_children_only: 'yes'
- name: tank/media
template: media
recursive: 'yes'
sanoid_templates:
- name: vm_live
hourly: 48
daily: 5
autosnap: "yes"
autoprune: "yes"
- name: media
hourly: 6
daily: 3
weekly: 2
monthly: 3
yearly: 1
autosnap: "yes"
autoprune: "yes"
Above I define two different templates and apply them to two different datasets.After running this on a node the sanoid systemd unit will be enabled and configured to check every 15 minutes for new snapshots to take or prune snapshots (based on the applied template).
You can dive deeper but this is 99% of what most need to take proper care of our ZFS snapshots.
GPLv3
Find me on GitHub @ https://github.com/gikeymarcia