Parent dataset snapshots - what are good practices

OpenZFS
1

I use some datasets such as vm to group for example VMs together. These parents don’t contain data. I use Sanoid to only take snapshots of the child datasets which actually contain data. This has left me with parent datasets without any snapshots. I feel I’ve had epiphany as I now realise that a dataset with a snapshot cannot be replicated…

How can I ensure parent datasets are included in a replication?

  • create a single snapshot for any parents just for replication purposes
  • include parents in the standard Sanoid policy and create lots of seemingly pointless snapshots.

Is there a way of ensuring no data gets written to parent datasets?

2

Just treat the empty parent datasets the same way you treat the children. The empty snapshots aren’t hurting anything.

3

Ok, thanks that likely simplifies my sanoid.conf quite a bit. I guess I overthought this initially!

1 Like
4

Hey, I had to learn that lesson the hard way myself, and I wrote the damn program. :slight_smile:

1 Like