This is an opinionated fork of TrilliumIT/docker-zfs-plugin that philosophically disagrees with the parent project about the use of fully-qualified dataset names. In particular, this fork does not require (or even accept) them. Instead, when launching the plugin, a parent dataset is passed in. All further requests to the plugin are scoped underneath that dataset.
Portability. If you have multiple nodes, they may very well use distinct
tank names. (In fact, this is recommended practice according to ZFS devs). With
the parent project's version, you would have to have node-specific configuration
for each container/service, which kind of defeats the purpose of using the
plugin altogether. If you were going to be editing your docker-compose file
for each node, why not just do a local bind mount for its mount point?
Just run multiple instances of the plugin, registered with Docker under different names.
Download the latest binary from Github releases and place it in
/usr/local/bin/.
If using a systemd based distribution, copy
docker-zfs-plugin.service to /etc/systemd/system.
Then enable and start the service with systemctl daemon-reload && systemctl enable docker-zfs-plugin.service && systemctl start docker-zfs-plugin.service.
After the plugin is running, you can interact with it through normal docker volume commands.
Recently, support was added for passing in ZFS attributes from the docker volume create command:
docker volume create -d zfs -o compression=lz4 -o dedup=on
--name=tank/docker-volumes/data