Ivan Tomica

Using Sanoid for snapshotting ZFS on Fedora

ZFS is really nice filesystem and I use it wherever I can. I could write essays on the topic of ZFS and how awesome it is, and that is just what might happen some time in the future, but for now I’ll just show you how snapshots are managed on my machines.

Why snapshots?

There are 10 types of people:

  • Those who do backups
  • And those who will

I’m not sure about you, but I’ve definitely deleted things I shouldn’t have only to realize few minutes (,or 3 days, or a month) later that I need the data that’s now gone. If I only had a point-in-time to which I could return and salvage my data…

To be clear, I don’t think having snapshots is enough, you should definitely have backups to some other machine, external drive array or something similar, preferably to an offsite location. And this can be accomplished in an automated fashion with ZFS send/receive functionality.

How to do them?

To do the snapshot you just issue:

zfs snapshot storage/your/dataset@name_of_the_snapshot

And this will do the filesystem snapshot that second you executed the command.

Well, you can imagine managing and remembering to do them in the first place can really be burden and might be neglected. If backups aren’t automated, they likely won’t be done.

There are many utilities that help managing backups, and I used some of them in the past. I liked ZFSnap very much but decided to go for Sanoid since it’s nicer on Linux, and by defining policies and templates in human readable format (shown below), I don’t have to mess with cronjobs. and thinking about them much. Just one config file to manage.

Installing Sanoid

Since I’m on Fedora I had to follow CentOS installation instructions. Instead of yum use dnf and you’re good.

Once you’re done just define your policies and templates and you’re done. Easy as that!

There are few notes I should mention regarding configuration though, and those are the ones I try to keep on my mind while defining them as well:

  • Policy name is the full dataset name used as a “source”
  • Defining some setting on policy level overrides just that particular setting. Other settings from the template are still valid
  • Template name is the thing you write after template_ defintion

At last, here’s an example of my config:

################################################################################
# Policies
################################################################################
[storage/home/ivan/Downloads]
	use_template = m1d3h4f3
	recursive = yes

[storage/home/ivan/Nextcloud]
	use_template = m3d30h72f8

[storage/home/ivan/Area52]
	use_template = m3d30h72f8

[storage/home/ivan/Documents]
	use_template = m3d30h72f8

################################################################################
# Templates
################################################################################
[template_m1d3h4f3]
	frequently = 3
	hourly = 4
	daily = 3
	monthly = 1
	yearly = 0
	autosnap = yes
	autoprune = yes

[template_m3d30h72f8]
	frequently = 8
	hourly = 72
	daily = 30
	monthly = 3
	yearly = 0
	autosnap = yes
	autoprune = yes

As you see, my template naming scheme is pretty imaginative.

Tagged in:, , , , ,
About 
Sysadmin on the everlasting journey of learning. Always in search for an opportunity to prove myself and to learn something new. My addiction is learning and my main goal is to excel in every aspect of Linux/Unix system administration.

0 Comments

Leave a Reply

Your email address will not be published. Required fields are marked *