CSI driver for provisioning Local PVs backed by ZFS and more.
With the release of version 1.2.0, the ZFS-LocalPV is promoted to GA. It is ready for the production, see our adopters.
See roadmap, e2e-wiki and e2e-test.
Before installing ZFS driver please make sure your Kubernetes Cluster must meet the following prerequisites:
- all the nodes must have zfs utils installed
- ZPOOL has been setup for provisioning the volume
- You have access to install RBAC components into kube-system namespace. The OpenEBS ZFS driver components are installed in kube-system namespace to allow them to be flagged as system critical components.
K8S : 1.17+
OS : Ubuntu, CentOS
ZFS : 0.7, 0.8
Check the features supported for each k8s version.
All the node should have zfsutils-linux installed. We should go to the each node of the cluster and install zfs utils
$ apt-get install zfsutils-linux
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.
If you have the disk(say /dev/sdb) then you can use the below command to create a striped pool:
zpool create zfspv-pool /dev/sdb
You can also create mirror or raidz pool as per your need. Check https://github.com/openzfs/zfs for more information.
If you don't have the disk, then you can create the zpool on the loopback device which is backed by a sparse file. Use this for testing purpose only.
truncate -s 100G /tmp/disk.img
zpool create zfspv-pool `sudo losetup -f /tmp/disk.img --show`
Once the ZFS Pool is created, verify the pool via zpool status command, you should see something like this :
$ sudo zpool status
pool: zfspv-pool
state: ONLINE
scan: none requested
config:
NAME STATE READ WRITE CKSUM
zfspv-pool ONLINE 0 0 0
sdb ONLINE 0 0 0
errors: No known data errors
Configure the custom topology keys (if needed). This can be used for many purposes like if we want to create the PV on nodes in a particuler zone or building. We can label the nodes accordingly and use that key in the storageclass for taking the scheduling decesion:
https://github.com/openebs/zfs-localpv/blob/master/docs/faq.md#6-how-to-add-custom-topology-key
We can install the latest release of OpenEBS ZFS driver by running the following command.
$ kubectl apply -f https://openebs.github.io/charts/zfs-operator.yaml
We can also install it via kustomize using kubectl apply -k deploy/yamls, check the kustomize yaml.
For microk8s, we need to change the kubelet directory to /var/snap/microk8s/common/var/lib/kubelet/, we need to replace /var/lib/kubelet/ with /var/snap/microk8s/common/var/lib/kubelet/ at all the places in the operator yaml and then we can apply it on microk8s.
Verify that the ZFS driver Components are installed and running using below command :
$ kubectl get pods -n kube-system -l role=openebs-zfs
Depending on number of nodes, you will see one zfs-controller pod and zfs-node daemonset running on the nodes.
NAME READY STATUS RESTARTS AGE
openebs-zfs-controller-0 5/5 Running 0 5h28m
openebs-zfs-node-4d94n 2/2 Running 0 5h28m
openebs-zfs-node-gssh8 2/2 Running 0 5h28m
openebs-zfs-node-twmx8 2/2 Running 0 5h28m
Once ZFS driver is installed we can provision a volume.
$ 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
The storage class contains the volume parameters like recordsize(should be power of 2), compression, dedup and fstype. You can select what are all parameters you want. In case, zfs properties paramenters are not provided, the volume will inherit the properties from the ZFS Pool. The poolname is the must argument. It should be noted that poolname can either be the root dataset or a child dataset e.g.
poolname: "zfspv-pool"
poolname: "zfspv-pool/child"
Also the dataset provided under poolname must exist on all the nodes with the name given in the storage class. Check the doc on storageclasses to know all the supported parameters for ZFS-LocalPV
If we provide fstype as ext2/3/4 or xfs or btrfs, the driver will create a ZVOL, which is a blockdevice carved out of ZFS Pool. This blockdevice will again formatted as corresponding filesystem(ext2/3/4 or xfs). In this way applications will get desired filesystem. Here, in this case there will be a filesystem layer on top of ZFS filesystem, and applications may not get the optimal performance. The sample storage class for ext4 fstype is provided below :-
$ cat sc.yaml
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
name: openebs-zfspv
parameters:
volblocksize: "4k"
compression: "off"
dedup: "off"
fstype: "ext4"
poolname: "zfspv-pool"
provisioner: zfs.csi.openebs.io
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. Here, please note that for ZFS, volblocksize should be power of 2.
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 recordsize has to be power of 2 for ZFS datasets.
If ZFS pool is available on certain nodes only, then make use of topology to tell the list of nodes where we have the ZFS pool available. As shown in the below storage class, we can use allowedTopologies to describe ZFS pool availability on nodes.
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
name: openebs-zfspv
allowVolumeExpansion: true
parameters:
recordsize: "4k"
compression: "off"
dedup: "off"
fstype: "zfs"
poolname: "zfspv-pool"
provisioner: zfs.csi.openebs.io
allowedTopologies:
- matchLabelExpressions:
- key: kubernetes.io/hostname
values:
- zfspv-node1
- zfspv-node2
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.
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
name: openebs-zfspv
allowVolumeExpansion: true
parameters:
recordsize: "4k"
compression: "off"
dedup: "off"
fstype: "zfs"
poolname: "zfspv-pool"
provisioner: zfs.csi.openebs.io
volumeBindingMode: WaitForFirstConsumer
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.
$ cat pvc.yaml
kind: PersistentVolumeClaim
apiVersion: v1
metadata:
name: csi-zfspv
spec:
storageClassName: openebs-zfspv
accessModes:
- ReadWriteOnce
resources:
requests:
storage: 4Gi
Create a PVC using the storage class created for the ZFS driver. Here, the allocated volume size will be rounded off to the nearest Mi or Gi notation, check the faq section for more details.
If we are using the immediate binding in the storageclass then we can check the kubernetes resource for the corresponding zfs volume, other wise in late binding case, we can check the same after pod has been scheduled.
$ kubectl get zv -n openebs
NAME ZPOOL NODE SIZE STATUS FILESYSTEM AGE
pvc-34133838-0d0d-11ea-96e3-42010a800114 zfspv-pool zfspv-node1 4294967296 Ready zfs 4s
$ kubectl describe zv pvc-34133838-0d0d-11ea-96e3-42010a800114 -n openebs
Name: pvc-34133838-0d0d-11ea-96e3-42010a800114
Namespace: openebs
Labels: kubernetes.io/nodename=zfspv-node1
Annotations: <none>
API Version: openebs.io/v1alpha1
Kind: ZFSVolume
Metadata:
Creation Timestamp: 2019-11-22T09:49:29Z
Finalizers:
zfs.openebs.io/finalizer
Generation: 1
Resource Version: 2881
Self Link: /apis/openebs.io/v1alpha1/namespaces/openebs/zfsvolumes/pvc-34133838-0d0d-11ea-96e3-42010a800114
UID: 60bc4df2-0d0d-11ea-96e3-42010a800114
Spec:
Capacity: 4294967296
Compression: off
Dedup: off
Fs Type: zfs
Owner Node ID: zfspv-node1
Pool Name: zfspv-pool
Recordsize: 4k
Volume Type: DATASET
Status:
State: Ready
Events: <none>
The ZFS driver will create a ZFS dataset(or zvol as per fstype in the storageclass) on the node zfspv-node1 for the mentioned ZFS pool and the dataset name will same as PV name. Go to the node zfspv-node1 and check the volume :-
$ zfs list
NAME USED AVAIL REFER MOUNTPOINT
zfspv-pool 444K 362G 96K /zfspv-pool
zfspv-pool/pvc-34133838-0d0d-11ea-96e3-42010a800114 96K 4.00G 96K legacy
Create the deployment yaml using the pvc backed by ZFS-LocalPV storage.
$ cat fio.yaml
apiVersion: v1
kind: Pod
metadata:
name: fio
spec:
restartPolicy: Never
containers:
- name: perfrunner
image: openebs/tests-fio
command: ["/bin/bash"]
args: ["-c", "while true ;do sleep 50; done"]
volumeMounts:
- mountPath: /datadir
name: fio-vol
tty: true
volumes:
- name: fio-vol
persistentVolumeClaim:
claimName: csi-zfspv
After the deployment of the application, we can go to the node and see that the zfs volume is being used by the application for reading/writting the data and space is consumed from the ZFS pool.
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 :
kubectl edit zv pvc-34133838-0d0d-11ea-96e3-42010a800114 -n openebs
You can edit the relevant property like make compression on or make dedup on and save it. This property will be applied to the corresponding volume and can be verified using below command on the node:
zfs get all zfspv-pool/pvc-34133838-0d0d-11ea-96e3-42010a800114
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.
$ kubectl delete -f fio.yaml
pod "fio" deleted
$ kubectl delete -f pvc.yaml
persistentvolumeclaim "csi-zfspv" deleted
- Access Modes
- ReadWriteOnce
ReadOnlyManyReadWriteMany
- Volume modes
-
Filesystemmode -
Blockmode
-
- Supports fsTypes:
ext4,btrfs,xfs,zfs - Volume metrics
- Snapshot
- Clone
- Volume Resize
- Raw Block Volume
- Backup/Restore
- Ephemeral inline volume
