Clarification on simple, incremental ZFS backup

[Paul Kraus]

On Jul 1, 2015, at 10:13, Brandon J. Wandersee <brandon.wandersee at gmail.com> wrote:

>  `zfs send -R <source_pool>@<snapshot> | zfs receive <backup_pool/backup_fs>`
> 
> There's a problem with this, though: when the filesystem is replicated
> all its properties are retained (naturally), including mountpoints. So
> the backup gets mounted over my local filesystem, making both
> useless. There's no way that I can see to unmount the backup filesystem
> (as that's now mounted to '/'), and importing the filesystem again after
> a reboot just mounts it over the local filesystem again. What am I doing
> wrong, here? Can anyone point me in the right direction? Does anyone
> here have a similar rsync-like backup scheme?

Use zfs recv -e (or -d) to put the replicated copy under a different mount point. I also set the mount point on my backup datasets to none to prevent accidental changes. Remember you can also snapshot and replicate the changes between snapshots to make the replication faster. We are backing up servers via ssh tunnels and incremental sends / recvs

--
Paul Kraus
paul at kraus-haus.org