From 0 to ZFS replication in 5m with syncoid
The ZFS filesystem has many features that once you try them you can never go back. One of the lesser known is probably the support for replicating a zfs filesystem by sending the changes over the network with zfs send/receive.
Technically the filesystem changes don't even need to be sent over a network: you could as well dump them on a removable disk, then receive from the same removable disk.
I suppose the reason send/receive is not so instantly popular among Linux users starting with ZFS is because they had access to rsync for such a long time and it works so well that they just don't feel the need for another replication tool.
The way ZFS send/receive works is by selecting just the changed blocks between two snapshot. It does not have to walk the filesystem tree and compute, exchange and compare hashes, sizes, timestamps and so on like rsync does which means it is extremely efficient. Also, excluding compression, it is more bandwidth efficient than rsync because the involved parties do not have to exchange file lists.
TIP: when rsync'ing to ZFS use --in-place to improve performance
The way ZFS send/receive works is by selecting just the changed blocks between two snapshot. It does not have to walk the filesystem tree and compute, exchange and compare hashes, sizes, timestamps and so on like rsync does which means it is extremely efficient. Also, excluding compression, it is more bandwidth efficient than rsync because the involved parties do not have to exchange file lists.
While you could script zfs send/receive manually, the fastest way to replicate one pool or filesystem to another is by using one of the many tools building upon it. Probably the best tool for this job is syncoid (which is a part of sanoid).
Caveat: syncoid requires that target filesystem does not exist: it will complain and refuse to work if it does. It is best to start with an empty target or replicate one filesystem at the time.
While not necessary it is a good idea to install the following extra packages to run syncoid at its full potential: mbuffer, lzop, pv, git. Ubuntu provides binaries for all of them, other distributions might not.
To use syncoid first clone the sanoid github repository:
git clone https://github.com/jimsalterjrs/sanoid.git
then cd into sanoid or copy syncoid to a directory in your path. Also I have found that syncoid attempts to use an SSH cypher that might not always be available. If that happens edit line 28 and leave the sshcipher option empty (there are probably ciphers more efficient than others for this purpose, but I'm not an expert. If you knwo of them let me know in the comments).
After that set up ssh for key-based autentication and then go ahead and try your first replication:
syncoid --recursive tank root@remotehost:tank
try adding --debug if you want to look under the hood.
Syncoid uses mbuffer to read large chunks of data into a memory buffer, compress it with lzop and transfer it over the network with ssh. The same happens on the receiving side, only in reversed order.
To replicate again re-run syncoid with the same options.
To replicate again re-run syncoid with the same options.