[](https://bestpractices.coreinfrastructure.org/en/projects/3523)
The current implementation only supports provisioning and de-provisioning of ZFS Volumes. Also, few properties like compression, dedup and recordsize can be provided while provisioning the volumes and can also be changed after provisioning is done.
Go to each node and create the ZFS Pool, which will be used for provisioning the volumes. You can create the Pool of your choice, it can be striped, mirrored or raidz pool.
Here please note that we are providing `volblocksize` instead of `recordsize` since we will create a ZVOL, for which we can choose the blocksize with which we want to create the block device.
##### zfs as FsType
In case if we provide "zfs" as the fstype, the zfs driver will create ZFS DATASET in the ZFS Pool, which is the zfs filesystem.
Here, there will not be any extra layer between application and storage, and applications can get the optimal performance.
The sample storage class for zfs fstype is provided below :-
```
$ cat sc.yaml
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
name: openebs-zfspv
parameters:
recordsize: "4k"
compression: "off"
dedup: "off"
fstype: "zfs"
poolname: "zfspv-pool"
provisioner: zfs.csi.openebs.io
```
Here please note that we are providing `recordsize` which will be used to create the ZFS datasets, which specifies the maximum block size for files in the zfs file system.
The above storage class tells that ZFS pool "zfspv-pool" is available on nodes zfspv-node1 and zfspv-node2 only. The ZFS driver will create volumes on those nodes only.
Please note that the provisioner name for ZFS driver is "zfs.csi.openebs.io", we have to use this while creating the storage class so that the volume provisioning/deprovisioning request can come to ZFS driver.
The ZFS driver has a scheduler which will try to distribute the PV across the nodes so that one node should not be loaded with all the volumes. Currently the driver has
VolumeWeighted scheduling algorithm, in which it will try to find a ZFS pool which has less number of volumes provisioned in it from all the nodes where the ZFS pools are available.
Once it is able to find the node, it will create a PV for that node and also create a ZFSVolume custom resource for the volume with the NODE information. The watcher for this ZFSVolume
CR will get all the information for this object and creates a ZFS dataset(zvol) with the given ZFS property on the mentioned node.
The scheduling algorithm currently only accounts for the number of ZFS volumes and does not account for other factors like available cpu or memory while making scheduling decisions.
So if you want to use node selector/affinity rules on the application pod, or have cpu/memory constraints, kubernetes scheduler should be used.
To make use of kubernetes scheduler, you can set the `volumeBindingMode` as `WaitForFirstConsumer` in the storage class.
This will cause a delayed binding, i.e kubernetes scheduler will schedule the application pod first and then it will ask the ZFS driver to create the PV.
The driver will then create the PV on the node where the pod is scheduled.
Please note that once a PV is created for a node, application using that PV will always get scheduled to that particular node only, as PV will be sticky to that node.
The scheduling algorithm by ZFS driver or kubernetes will come into picture only during the deployment time. Once the PV is created,
the application can not move anywhere as the data is there on the node where the PV is.
ZFS Volume Property can be changed like compression on/off can be done by just simply editing the kubernetes resource for the corresponding zfs volume by using below command :
for deprovisioning the volume we can delete the application which is using the volume and then we can go ahead and delete the pv, as part of deletion of pv this volume will also be deleted from the ZFS pool and data will be freed.
