How can I 1:1 replicate a pool without datasets (same name, mount on target)?

[Catfriend1]

Hi,

I've successfully used zfs-autobackup to create snapshots at the source pool and replicate them to the target pool.

My source looks like this:

zfs get all | grep mountpoint

pool0                      mountpoint            /pool0                default

My target looks like this:

backup                                    mountpoint            /backup/vm-fs01/pool0          received
backup/vm-fs01                            mountpoint            /backup/vm-fs01/pool0/vm-fs01  inherited from backup
backup/vm-fs01/pool0                      mountpoint            /backup/vm-fs01/pool0          local

I'd like to replicate the current pool0 state from the source to the target, then remove all snapshots, datasets so I'll end up with the current state as /pool0 at the target.

pool0                      mountpoint            /pool0                default

How could I go about this using zfs-autobackup?

Is there an alternative way to convert the existing, replicated last snapshot into the "root /pool0", removing the rest from the target?

Target snapshot state:

zfs list -t snapshot

NAME                                       USED  AVAIL  REFER  MOUNTPOINT
backup/vm-fs01/pool0@fs02-20210704153546   328K      -   813G  -

Thank you.

[psy0rz]

I think this is related to #41 ?

Keep in mind that the main pool0 cant be replicated, since you have to create it manually on the target in the first place.

[psy0rz]

Try using --strip-path 1

[Catfriend1]

@psy0rz Thanks, will try it.

Btw. zfs send (snapshot) | ssh ip zfs recv -F pool0 can replicate the whole pool "as is" to the target pool root.

[psy0rz]

You're right. So in that case the strip-path might be not enough for you, since it doesnt receive the root-dataset of a pool, only the sub-datasets.

[psy0rz]

Since your name is Catfriend1, check out this project of mine: https://github.com/psy0rz/meowton/wiki :)

[Catfriend1]

@psy0rz Pretty cool cat gadget :-)! 🐱

[psy0rz]

@psy0rz Thanks, will try it.

Btw. zfs send (snapshot) | ssh ip zfs recv -F pool0 can replicate the whole pool "as is" to the target pool root.

Since this is an issue for multiple people, i now improved stripping and added the --force (-F) option.

Also updated the documentation to explain how 1:1 replication works:

https://github.com/psy0rz/zfs_autobackup/wiki#making-source-and-target-paths-look-the-same

[psy0rz]

pip install zfs-autobackup==v3.1.2-rc2