This is the multi-page printable view of this section. Click here to print.

Return to the regular view of this page.

Installation

1: Installation Using Helm

1.1: Installing Huawei CSI on Kubernetes, OpenShift, and Tanzu
1.2: Installing Huawei CSI on the CCE or CCE Agile Platform
1.3: Parameters in the values.yaml File of Helm

2: Manual Installation

This section describes how to install Huawei CSI.

In the current version, resource requests and limits are added to Huawei CSI. For details, see Huawei CSI Resource Management .

Prerequisites

Operations described in Installation Preparations have been completed.
All worker nodes of the cluster communicate properly with the service network of the storage device to be connected. In iSCSI scenarios, the ping command can be used to verify the connectivity.
Software clients required by the corresponding protocol, such as iSCSI and NFS clients, have been installed on all worker nodes of the cluster.

1 - Installation Using Helm

This section describes how to install and deploy Huawei CSI using Helm. The Helm software version must be 3.0 or later.

Helm Installation Description

Huawei CSI can be installed as the root user or a non-root user. When installing Huawei CSI as a non-root user, ensure that the current user can access the API Server of the Kubernetes cluster. For details about how to configure access to the Kubernetes cluster as a non-root user, see Configuring Access to the Kubernetes Cluster as a Non-root User .
Huawei CSI must be run as the root user.

Helm is a software package management tool in the Kubernetes ecosystem. Similar to Ubuntu APT, CentOS YUM, or Python pip, Helm manages Kubernetes application resources.

You can use Helm to package, distribute, install, upgrade, and roll back Kubernetes applications in a unified manner.

For details about how to obtain and install Helm, see https://helm.sh/docs/intro/install/ .
For details about the mapping between Helm and Kubernetes versions, see https://helm.sh/docs/topics/version_skew/ .

When installing huawei-csi-controller, Helm deploys the following components in the workloads of the Deployment type in the specified namespace:

huawei-csi-driver: Huawei CSI driver.
storage-backend-controller: Huawei backend management controller, used to manage storageBackendClaim resources.
storage-backend-sidecar: used to manage storageBackendContent resources.
Kubernetes External Provisioner: used to provide or delete volumes.
Kubernetes External Attacher: used to attach or detach volumes.
Kubernetes External Resizer: used to expand the capacity of volumes.
Kubernetes External liveness-probe: used to determine the health status of a Pod.
(Optional) huawei-csi-extender: used to support changing persistent volumes.
(Optional) Kubernetes External Snapshotter: used to provide snapshot support (installed as CRD).
(Optional) Kubernetes External Snapshot Controller: used to control volume snapshots.
(Optional) Kubernetes External Health Monitor Controller: used to monitor and control volume health.

When installing huawei-csi-node, Helm deploys the following components in the workloads of the DaemonSet type in the specified namespace:

huawei-csi-driver: Huawei CSI driver.
Kubernetes Node Registrar: used to process driver registration.
liveness-probe: used to determine the health status of a Pod.

1.1 - Installing Huawei CSI on Kubernetes, OpenShift, and Tanzu

Installation Procedure

Use a remote access tool, such as PuTTY, to log in to any master node in the cluster through the management IP address.
Copy the helm directory in the Kubernetes CSI component package to any directory on the master node. For details about the Helm tool path, see Table 1 .
Go to the helm/esdk working directory.
```
cd helm/esdk
```
Prepare the values.yaml file. Huawei CSI provides the values.yaml template file in the helm/esdk directory of the software package. You can also modify parameters according to Parameters in the values.yaml File of Helm to customize Huawei CSI.
Perform the following configuration before the installation:
- If the container platform is Kubernetes, skip this step.
- If the container platform is OpenShift, perform the configuration in Installation and Configuration on the OpenShift Platform .
- If the container platform is Tanzu, perform the configuration in Installation and Configuration on the Tanzu Platform .
Run the following command to update the storage backend CRD.
```
kubectl apply -f ./crds/backend/
```
(Optional) Check snapshot-dependent components by following the instructions provided in Checking Volume Snapshot-Dependent Components . After confirming that the components are correct, run the following command to update the snapshot CRD. If controller.snapshot.enabled is set to false or the Kubernetes version is earlier than v1.20, you can skip this step. For details, see Table 2 .
```
kubectl apply -f ./crds/snapshot-crds/ --validate=false
```
Run the following command to install Huawei CSI. In the preceding command, helm-huawei-csi indicates the custom Helm chart name, ./ indicates that the Helm project in the current directory is used, and huawei-csi indicates the custom Helm chart namespace.
```
helm install helm-huawei-csi ./ -n huawei-csi --create-namespace
```
The following is an example of the command output.
```
NAME: helm-huawei-csi
LAST DEPLOYED: Wed Jun  8 11:50:28 2022
NAMESPACE: huawei-csi
STATUS: deployed
REVISION: 1
TEST SUITE: None
```

After the huawei-csi service is deployed, run the following command to check whether the service is started.

kubectl get pod -n huawei-csi

The following is an example of the command output. If the Pod status is Running, the installation is successful.

NAME                                     READY   STATUS    RESTARTS   AGE
huawei-csi-controller-6dfcc4b79f-9vjtq   9/9     Running   0          24m
huawei-csi-controller-6dfcc4b79f-csphc   9/9     Running   0          24m
huawei-csi-node-g6f4k                    3/3     Running   0          20m
huawei-csi-node-tqs87                    3/3     Running   0          20m

Installation and Configuration on the OpenShift Platform

For the OpenShift platform, perform the following steps to create the SecurityContextConstraints resource.

Run the following command to edit the helm_scc.yaml file.
```
vi helm_scc.yaml
```

Modify the helm_scc.yaml file. In the following commands, huawei-csi indicates the created namespace. Replace it based on the actual situation.

apiVersion: security.openshift.io/v1
kind: SecurityContextConstraints
metadata:
  name: helm-scc
allowHostDirVolumePlugin: true
allowHostIPC: true
allowHostNetwork: true
allowHostPID: true
allowHostPorts: true
allowPrivilegeEscalation: true
allowPrivilegedContainer: true

defaultAddCapabilities:
- SYS_ADMIN
runAsUser:
  type: RunAsAny
seLinuxContext:
  type: RunAsAny
fsGroup:
  type: RunAsAny
users:
- system:serviceaccount:huawei-csi:huawei-csi-controller
- system:serviceaccount:huawei-csi:huawei-csi-node

Run the following command to create a SecurityContextConstraints file.
```
oc create -f helm_scc.yaml
```

Installation and Configuration on the Tanzu Platform

On the Tanzu platform, run the following command to configure the kubelet installation directory.

Go to the helm/esdk directory in the installation package, and run the following command to open the configuration file. For details about the installation package directory, see Table 1 .
```
vi values.yaml
```

Modify the kubeletConfigDir parameter as follows and save the settings.

# Specify kubelet config dir path.
# kubernetes and openshift is usually /var/lib/kubelet
# Tanzu is usually /var/vcap/data/kubelet
# CCE is usually /mnt/paas/kubernetes/kubelet
kubeletConfigDir: /var/vcap/data/kubelet

For TKGI 1.16 or earlier of the Tanzu platform, run the following commands to configure the RBAC permission.

Run the following command to create a file named rbac.yaml.
```
vi rbac.yaml
```

Copy the following content to the rbac.yaml file, save the file, and exit.

apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: huawei-csi-psp-role
rules:
- apiGroups: ['policy']
  resources: ['podsecuritypolicies']
  verbs: ['use']
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: huawei-csi-psp-role-cfg
roleRef:
  kind: ClusterRole
  name: huawei-csi-psp-role
  apiGroup: rbac.authorization.k8s.io
subjects:
- kind: Group
  apiGroup: rbac.authorization.k8s.io
  name: system:serviceaccounts:huawei-csi
- kind: Group
  apiGroup: rbac.authorization.k8s.io
  name: system:serviceaccounts:default

Run the following command to create the RBAC permission.
```
kubectl create -f rbac.yaml
```

1.2 - Installing Huawei CSI on the CCE or CCE Agile Platform

This section describes how to install Huawei CSI on the CCE or CCE Agile platform.

Creating a Helm Installation Package

The CCE or CCE Agile platform cannot directly install Huawei CSI using Helm. You need to manually create a Helm installation package and upload it to the chart list on the platform for installation.

Use a remote access tool, such as PuTTY, to log in to any node where Helm is deployed through the management IP address.
Copy the helm directory in the Huawei CSI component package to any directory on the node. For details about the Helm tool path, see Table 1 .
Go to the helm working directory.
```
cd helm/
```

Modify the kubeletConfigDir and csiDriver.driverName parameters in the helm/esdk/values.yaml file.

vi ./esdk/values.yaml

Modify the following parameters:

# Specify kubelet config dir path.
# kubernetes and openshift is usually /var/lib/kubelet
# Tanzu is usually /var/vcap/data/kubelet
# CCE is usually /mnt/paas/kubernetes/kubelet
kubeletConfigDir: /mnt/paas/kubernetes/kubelet

# The CSI driver parameter configuration
csiDriver:
  # Driver name, it is strongly recommended not to modify this parameter
  # The CCE platform needs to modify this parameter, e.g. csi.oceanstor.com
  driverName: csi.oceanstor.com

Run the following command to create a Helm installation package. This command will generate the installation package to the current path.
```
helm package ./esdk/ -d ./
```

Installing Huawei CSI

Use a remote access tool, such as PuTTY, to log in to any master node where the CCE Agile platform is deployed through the management IP address.
Run the following command to create a namespace for deploying Huawei CSI. huawei-csi indicates the custom namespace.
```
kubectl create namespace huawei-csi
```
Export the Helm installation package. For details, see Creating a Helm Installation Package .
On the home page, choose Charts > My Charts > Upload Chart. The Upload Chart dialog box is displayed. Import the exported Helm installation package to the CCE Agile platform.
After the installation package is uploaded, choose Charts > My Charts. On the My Charts page that is displayed, choose Install > Submit. The chart release name can be customized.
On the home page, choose Charts > Releases and select the project specified during installation (for example, default in the following figure). After the installation is successful, Installed is displayed in the Status column.

1.3 - Parameters in the values.yaml File of Helm

When using Helm to install CSI, you need to prepare the values.yaml file of the Helm project based on the features required during deployment. Huawei CSI provides the values.yaml template file in the helm/esdk directory of the software package.

This section describes the configuration items in the values.yaml file and backend configuration examples in typical scenarios.

images Parameters

The images parameters in the values.yaml file are used to configure the component image information on which Huawei CSI depends during running. Set the following parameters:

Table 1 images parameters

Parameter	Description	Mandatory	Default Value
images.huaweiCSIService	huawei-csi image.	Yes	huawei-csi:4.12.0
images.storageBackendSidecar	Huawei back-end management sidecar image.	Yes	storage-backend-sidecar:4.12.0
images.storageBackendController	Huawei back-end management controller image.	Yes	storage-backend-controller:4.12.0
images.huaweiCSIExtender	huawei-csi-extender image.	No	huawei-csi-extender:4.12.0
images.sidecar.livenessProbe	livenessprobe sidecar image.	Yes	registry.k8s.io/sig-storage/livenessprobe:v2.15.0
images.sidecar.provisioner	csi-provisioner sidecar image.	Yes	registry.k8s.io/sig-storage/csi-provisioner:v5.2.0
images.sidecar.attacher	csi-attacher sidecar image.	Yes	registry.k8s.io/sig-storage/csi-attacher:v4.8.0
images.sidecar.resizer	csi-resizer sidecar image.	Yes	registry.k8s.io/sig-storage/csi-resizer:v1.13.2
images.sidecar.snapshotter	csi-snapshotter sidecar image.	Yes	registry.k8s.io/sig-storage/csi-snapshotter:v8.2.0
images.sidecar.snapshotController	snapshot-controller sidecar image.	Yes	registry.k8s.io/sig-storage/snapshot-controller:v8.2.0
images.sidecar.registrar	csi-node-driver-registrar sidecar image.	Yes	registry.k8s.io/sig-storage/csi-node-driver-registrar:v2.13.0
images.sidecar.healthMonitorController	health-monitor-controller sidecar image.	No	registry.k8s.io/sig-storage/csi-external-health-monitor-controller:v0.17.0

For details about the values of huaweiCSIService, storageBackendSidecar, storageBackendController, and huaweiCSIExtender, see Uploading a Huawei CSI Image . Use the name and version of the finally generated image.
For details about other sidecar image parameters, see Checking the Images on Which CSI Depends . Use the name and version of the finally uploaded image.

controller Parameters

The controller parameters are used to configure the huawei-csi-controller component.

Table 2 controller parameters

Parameter	Description	Mandatory	Default Value	Remarks
controller.controllerCount	Number of huawei-csi-controller component copies.	Yes	1	If the Kubernetes version is earlier than v1.17, the huawei-csi-controller component can be deployed only in single-copy mode because the csi-provisioner sidecar image provided by the Kubernetes community does not support the --leader-election parameter. Therefore, if the Kubernetes version is earlier than v1.17, this parameter can only be set to 1.
controller.volumeNamePrefix	PV name prefix. The default value is pvc, that is, the name of a created PV is pvc-<uuid>. The prefix must comply with the naming rules of DNS Subdomain Names, and the total length of the PV name cannot exceed 253 characters.	No	pvc	The corresponding provisioner parameter name is --volume-name-prefix. It is recommended that the prefix contain no more than 20 characters. For details, see Configuring the PV Name Prefix. If the connected backend is OceanStor V5 SAN storage, it is recommended that the prefix contain a maximum of 5 characters. If the connected backend is OceanStor V5 NAS storage, the prefix can contain only lowercase letters, hyphens (-), and digits. If the connected backend is OceanStor Dorado, OceanStor, OceanDisk, or OceanStor A series storage, the prefix can contain only lowercase letters, hyphens (-), and digits. If the connected backend is OceanStor Pacific series storage, the prefix can contain a maximum of 58 characters, including only letters, digits, underscores (_), hyphens (-), and periods (.). If the connected backend is FusionStorage Block, the prefix can contain a maximum of 58 characters, including only letters, digits, underscores (_), and hyphens (-).
controller.webhookPort	Port used by the webhook service.	Yes	4433	If a port conflict occurs, change the port number to an idle one.
controller.snapshot.enabled	Whether to enable the snapshot feature.	Yes	true	If you want to use snapshot-related functions, enable this feature. The Kubernetes version must be later than v1.20.
controller.resizer.enabled	Whether to enable the capacity expansion feature.	Yes	true	The Kubernetes version must be later than v1.16.
controller.resizer.kubeApiQps	Number of requests that can be sent by the client to kube-apiserver per second.	No	5
controller.resizer.kubeApiBurst	Maximum number of requests that can be sent by the client in a short period.	No	10
controller.nodeSelector	Node selector of huawei-csi-controller. After this parameter is set, huawei-csi-controller will be scheduled only to a node with the label.	No	-	For details about the node selector, see Assign Pods to Nodes.
controller.tolerations	Taint toleration of huawei-csi-controller. After this parameter is set, huawei-csi-controller can tolerate taints on a node.	No	-	For details about taints and tolerations, see Taints and Tolerations.
controller.affinity	Node affinity of huawei-csi-controller. After this parameter is set, huawei-csi-controller will be preferentially scheduled to a node with the label.	No	-	For details about node affinity, see Assigning Pods to Nodes.
controller.livenessProbePort	Liveness probe port of huawei-csi-controller, used for health check.	Yes	9808	If a port conflict occurs, change the port number to an idle one.
controller.csiExtender.volumeModify.enabled	Whether to enable the PVC change feature.	No	false	If you want to use PVC change-related functions, enable this feature.
controller.csiExtender.volumeModify.retryBaseDelay	Minimum retry interval when a PVC change fails to be created.	No	5s	The default value is recommended.
controller.csiExtender.volumeModify.retryMaxDelay	Maximum retry interval when a PVC change fails to be created.	No	5m	The default value is recommended.
controller.csiExtender.volumeModify.reconcileDelay	Interval for reconciling VolumeModifyClaim objects.	No	1s	The default value is recommended.
controller.exportCsiService.enabled	Whether to run CSI services on the Service of the Kubernetes cluster.	No	false	After this function is enabled, other services in the Kubernetes cluster can access CSI services through gRPC.
controller.exportCsiService.port	Port used when CSI services run on the Service of the Kubernetes cluster.	No	9090	If a port conflict occurs, change the port number to an idle one.
controller.healthMonitor.enabled	Whether to enable the volume monitoring service.	No	false	After this function is enabled, volume monitoring features can be used.
controller.healthMonitor.monitorInterval	Monitoring polling interval.	No	60s	Set the volume monitoring polling interval.
controller.healthMonitor.workThreads	Number of monitoring PV worker threads.	No	10	Set the number of monitoring PV worker threads.
controller.provisioner.kubeApiQps	Number of requests that can be sent by the client to kube-apiserver per second.	No	5
controller.provisioner.kubeApiBurst	Maximum number of requests that can be sent by the client in a short period.	No	10
controller.attacher.kubeApiQps	Number of requests that can be sent by the client to kube-apiserver per second.	No	5
controller.attacher.kubeApiBurst	Maximum number of requests that can be sent by the client in a short period.	No	10
controller.snapshotter.kubeApiQps	Number of requests that can be sent by the client to kube-apiserver per second.	No	5
controller.snapshotter.kubeApiBurst	Maximum number of requests that can be sent by the client in a short period.	No	10
controller.storageBackendController.kubeApiQps	Number of requests that can be sent by the client to kube-apiserver per second.	No	5
controller.storageBackendController.kubeApiBurst	Maximum number of requests that can be sent by the client in a short period.	No	10
controller.storageBackendSidecar.kubeApiQps	Number of requests that can be sent by the client to kube-apiserver per second.	No	5
controller.storageBackendSidecar.kubeApiBurst	Maximum number of requests that can be sent by the client in a short period.	No	10
controller.huaweiCsiExtender.kubeApiQps	Number of requests that can be sent by the client to kube-apiserver per second.	No	5
controller.huaweiCsiExtender.kubeApiBurst	Maximum number of requests that can be sent by the client in a short period.	No	10
controller.huaweiCsiDriver.kubeApiQps	Number of requests that can be sent by the client to kube-apiserver per second.	No	5
controller.huaweiCsiDriver.kubeApiBurst	Maximum number of requests that can be sent by the client in a short period.	No	10

If controller.snapshot.enabled is set to true, you need to install the volume snapshot CRD resource in the helm/crd/snapshot-crds directory.

node Parameters

The node parameters are used to configure the huawei-csi-node component.

Table 3 node parameters

Parameter	Description	Mandatory	Default Value	Remarks
node.maxVolumesPerNode	Maximum number of volumes provisioned by Huawei CSI that can be used by a node. If this parameter is not specified or is set to 0, the number is unlimited. If nodeName is specified during Pod creation, this configuration will be ignored.	No	100	For details, see Volume Limits.
node.nodeSelector	Node selector of huawei-csi-node. After this parameter is set, huawei-csi-node will be scheduled only to a node with the label.	No	-	For details about the node selector, see Assign Pods to Nodes.
node.tolerations	Taint toleration of huawei-csi-node. After this parameter is set, huawei-csi-node can tolerate taints on a node.	No	- key: "node.kubernetes.io/memory-pressure" operator: "Exists" effect: "NoExecute" - key: "node.kubernetes.io/disk-pressure" operator: "Exists" effect: "NoExecute" - key: "node.kubernetes.io/network-unavailable" operator: "Exists" effect: "NoExecute"	For details about taints and tolerations, see Taints and Tolerations.
node.affinity	Node affinity of huawei-csi-node. After this parameter is set, huawei-csi-node will be preferentially scheduled to a node with the label.	No	-	For details about node affinity, see Assigning Pods to Nodes.
node.livenessProbePort	Liveness probe port of huawei-csi-node, used for health check.	Yes	9800	If a port conflict occurs, change the port number to an idle one.
node.kubeletVolumeDevicesDirName	Name of the directory where a block device is mounted to kubelet.	No	volumeDevices	After a block device is successfully mounted, the directory structure of the mount path is as follows: /var/lib/kubelet/plugins/kubernetes.io/csi/{kubeletVolumeDevicesDirName}/publish/{specName}/{podUID}
node.huaweiCsiDriver.kubeApiQps	Number of requests that can be sent by the client to kube-apiserver per second.	No	5
node.huaweiCsiDriver.kubeApiBurst	Maximum number of requests that can be sent by the client in a short period.	No	10

csiDriver Parameters

The csiDriver parameters include the basic configurations for running Huawei CSI, such as Huawei driver name and multipathing type.

Table 4 csiDriver parameters

Parameter	Description	Mandatory	Default Value	Remarks
csiDriver.driverName	Registered driver name.	Yes	csi.huawei.com	Use the default value. For the CCE Agile platform, modify this field. For example, csi.oceanstor.com.
csiDriver.endpoint	Communication endpoint.	Yes	/csi/csi.sock	Use the default value.
csiDriver.connectorThreads	Maximum number of disks that can be concurrently scanned/detached. The value is an integer ranging from 1 to 10.	Yes	4	A larger value indicates that more concurrent disk scanning and detaching operations are performed on a single node at the same time. When DM-Multipath is used, a large number of concurrent requests may cause unknown problems and affect the overall time.
csiDriver.volumeUseMultipath	Whether to use multipathing software. The value is a Boolean value.	Yes	true	It is strongly recommended that multipathing software be enabled to enhance the redundancy and performance of storage links.
csiDriver.scsiMultipathType	Multipathing software used when the storage protocol is fc or iscsi. The following parameter values can be configured: DM-multipath HW-UltraPath HW-UltraPath-NVMe	Mandatory when volumeUseMultipath is set to true.	DM-multipath	The DM-multipath value is recommended.
csiDriver.nvmeMultipathType	Multipathing software used when the storage protocol is NVMe. The following parameter values can be configured: HW-UltraPath-NVMe NVMe-Native	Mandatory when volumeUseMultipath is set to true.	HW-UltraPath-NVMe	roce-nvme and fc-nvme support only HW-UltraPath-NVMe. The tcp-nvme protocol supports only NVMe-Native.
csiDriver.scanVolumeTimeout	Timeout interval for waiting for multipathing aggregation when DM-Multipath is used on the host. The value ranges from 1 to 600 seconds.	Yes	3	-
csiDriver.execCommandTimeout	Timeout interval for running commands on the host.	Yes	30	In scenarios such as mounting and capacity expansion, the CSI plug-in needs to run some host commands, for example, running the mount command to mount a file system. This parameter is used to control the timeout interval for running a single command.
csiDriver.enableRoCEConnect	Whether to enable CSI automatic disk scanning when the roce-nvme or tcp-nvme protocol is used.	No	true	If an external tool is used to establish an NVMe_over_RoCE or NVMe_over_TCP connection, set this parameter to false. For example, if the SNSD automatic connection function is enabled on storage devices, you can set this parameter to false.
csiDriver.allPathOnline	Whether to check that the number of paths aggregated by DM-Multipath or NVMe-Native is equal to the actual number of online paths. The following parameter values can be configured: true: The drive letter mounting condition is met only when the number of paths aggregated by multipathing software is equal to the actual number of online paths. false: By default, the number of paths aggregated by multipathing software is not checked. As long as virtual drive letters are generated upon aggregation, the drive letter mounting condition is met.	This parameter is mandatory when csiDriver.scsiMultipathType is set to DM-multipath or csiDriver.nvmeMultipathType is set to NVMe-Native.	false	-
csiDriver.backendUpdateInterval	Interval for updating backend capabilities. The value ranges from 60 to 600 seconds.	Yes	60	-
csiDriver.controllerLogging.module	Record type of the controller log. The following parameter values can be configured: file console	Yes	file	When the value is file, logs are retained in the specified directory of the node. When the Pod where CSI is located is destroyed, logs are still retained. When the value is console, logs are retained in the temporary space of the Pod where CSI is located. When the Pod where CSI is located is destroyed, the logs are also destroyed.
csiDriver.controllerLogging.level	Output level of the controller log. The following parameter values can be configured: debug info warning error fatal	Yes	info	-
csiDriver.controllerLogging.fileDir	Directory of the controller log in file output mode.	Yes	/var/log/huawei	Ensure that the directory has sufficient space for storing logs. It is recommended that the space be greater than or equal to 200 MB.
csiDriver.controllerLogging.fileSize	Size of a single controller log file in file output mode.	Yes	20M	-
csiDriver.controllerLogging.maxBackups	Maximum number of controller log file backups in file output mode.	Yes	9	-
csiDriver.nodeLogging.module	Record type of the node log. The following parameter values can be configured: file console	Yes	file	When the value is file, logs are retained in the specified directory of the node. When the Pod where CSI is located is destroyed, logs are still retained. When the value is console, logs are retained in the temporary space of the Pod where CSI is located. When the Pod where CSI is located is destroyed, the logs are also destroyed.
csiDriver.nodeLogging.level	Output level of the node log. The following parameter values can be configured: debug info warning error fatal	Yes	info	-
csiDriver.nodeLogging.fileDir	Directory of the node log in file output mode.	Yes	/var/log/huawei	Ensure that the directory has sufficient space for storing logs. It is recommended that the space be greater than or equal to 200 MB.
csiDriver.nodeLogging.fileSize	Size of a single node log file in file output mode.	Yes	20M	-
csiDriver.nodeLogging.maxBackups	Maximum number of node log file backups in file output mode.	Yes	9	-
csiDriver.reportNodeIP	Whether to report the node IP address.	No	false	After this item is enabled, the CSI node will automatically report the node IP address.
csiDriver.enablePerNodeSecret	Whether to create a Secret for each node.	No	fasle	By default, the information required by CSI on all nodes is saved to a Secret. After this function is enabled, an independent Secret is created for each node to save information. NOTICE: After this function is enabled, the following conditions must be met: The K8s node name must meet the K8s naming rules in Names. The storage backend whose name is the same as that of the K8s node cannot exist in the cluster.

If Huawei CSI has been deployed in your container environment, ensure that the value of csiDriver.driverName is the same as that configured during previous deployment. Otherwise, existing volumes or snapshots provisioned by Huawei CSI in the system cannot be managed by the newly deployed Huawei CSI.

Other Parameters

Other parameters include some features of the CSI plug-in or the policies for obtaining images.

Table 5 Other parameters

Parameter	Description	Mandatory	Default Value	Remarks
kubernetes.namespace	Kubernetes namespace where Huawei CSI is running, which can be customized. The name must consist of lowercase letters, digits, and hyphens (-), for example, my-name and 123-abc.	No	huawei-csi	-
kubeletConfigDir	Working directory of kubelet.	Yes	/var/lib/kubelet	Use the default value. For the Tanzu platform, change the value of this field to /var/vcap/data/kubelet. For the CCE Agile platform, change the value of this field to /mnt/paas/kubernetes/kubelet.
sidecarImagePullPolicy	Pull policy of the sidecar image.	Yes	IfNotPresent	-
huaweiImagePullPolicy	Pull policy of the huawei-csi image.	Yes	IfNotPresent	-
imagePullSecrets	Used by the Kubernetes cluster to pass the identity authentication of an image registry to pull private images.	No	-	For details, see Pull an Image from a Private Registry.
CSIDriverObject.isCreate	Whether to create the CSIDriver object.	Yes	false	The CSIDriver feature is a GA version in Kubernetes v1.18. Therefore, to use this feature, the Kubernetes version must be later than v1.18. If the Kubernetes version is earlier than v1.18, set this parameter to false.
CSIDriverObject.attachRequired	Whether the CSI plug-in skips the attach operation. The following parameter values can be configured: true: The attach operation is required. false: The attach operation is skipped.	Yes	true	The attachRequired parameter can be configured in Kubernetes v1.18. If CSIDriverObject.isCreate is set to true and attachRequired is set to false, the huawei-csi plug-in will not deploy the csi-attacher sidecar. If NAS storage is used, this parameter can be set to false. If SAN storage is used, set this parameter to true.
CSIDriverObject.fsGroupPolicy	Whether the ownership and permissions of a basic volume can be changed before the volume is mounted. The following parameter values can be configured: "ReadWriteOnceWithFSType": The volume ownership and permission can be changed only when fsType is specified and acfsTypecessModes of the volume contains ReadWriteOnce. "File": Kubernetes can use fsGroup to change the permissions and ownership of a volume to match fsGroup requested by a user in the Pod security policy, regardless of fsGroup or accessModes. "None": A volume is mounted without any change. "null": The fsGroupPolicy parameter is not set.	No	null	The fsGroupPolicy parameter can be configured in Kubernetes v1.20, and takes effect only when CSIDriverObject.isCreate is set to true. This feature is a Beta version in Kubernetes v1.20 but a GA version in Kubernetes v1.23. Therefore, the Kubernetes version must be later than v1.20.
leaderElection.leaseDuration	Leader duration.	No	8s	This parameter takes effect only in the multi-controller scenario.
leaderElection.renewDeadline	Time for the leader to be re-elected.	No	6s	This parameter takes effect only in the multi-controller scenario.
leaderElection.retryPeriod	Leader election retry time.	No	2s	This parameter takes effect only in the multi-controller scenario.
service.ipFamilyPolicy	IP protocol stack selection policy of a service.	No	SingleStack	The value can be: SingleStack: The service uses only one IP address family (IPv4 or IPv6). PreferDualStack: The service preferentially uses dual stack. If the cluster does not support dual stack, single stack is used. RequireDualStack: The service must use dual stack. If the cluster does not support dual stack, the service will fail to be created.
service.ipFamilies	List of IP protocol stacks supported by a service.	No	IPv4	The value can be: IPv4 IPv6
resources	You can allocate container resources related to huawei-csi-controller and huawei-csi-node: resources.<component>.<container-name> <component> supports the following values: node: configured when the sidecar in the huawei-csi-node component is configured. controller: configured when the sidecar in the huawei-csi-controller component is configured. controller.<container-name> supports the following values: livenessProbe csiProvisioner csiAttacher csiResizer csiSnapshotter snapshotController storageBackendController storageBackendSidecar huaweiCsiExtender huaweiCsiDriver node.<container-name> supports the following values: huaweiCsiDriver livenessProbe csiNodeDriverRegistrar	No	For the default values of different container resources, see Huawei CSI Resource Management.	The following uses livenessProbe of huawei-csi-controller as an example: resources: controller: limits: cpu: 100m memory: 128Mi requests: cpu: 10m memory: 128Mi

Ensure that the namespace entered in kubernetes.namespace exists on Kubernetes. If the namespace does not exist, run the following command to create it. In this example, the namespace for running Huawei CSI is huawei-csi.
kubectl create namespace huawei-csi

2 - Manual Installation

This section describes how to manually install Huawei CSI.

Currently, only the Kubernetes platform supports manual installation of Huawei CSI.

Procedure

Use a remote access tool, such as PuTTY, to log in to any master node in the cluster through the management IP address.
Copy the manual directory in the Kubernetes CSI component package to any directory on the master node.
Run the following command to create a namespace. huawei-csi is used as an example namespace.
```
kubectl create ns huawei-csi
```
Go to the manual/esdk working directory. For details about the path, see Table 1 .
```
cd manual/esdk
```
Run the following command to update the storage backend CRD.
```
kubectl apply -f ./crds/backend/
```
(Optional) Check snapshot-dependent components by following the instructions provided in Checking Volume Snapshot-Dependent Components . After confirming that the components are correct, run the following command to update the snapshot CRD. If the Kubernetes version is earlier than v1.20, skip this step.
```
kubectl apply -f ./crds/snapshot-crds/ --validate=false
```
(Optional) Run the following command to install CSIDriver. If the CSIDriver feature is not used, you can skip this step. For details, see the CSIDriver feature.
```
kubectl apply -f ./deploy/csidriver.yaml
```
Run the following command to install the huawei-csi-controller service.
If the Kubernetes version is earlier than v1.20, modify the ./deploy/huawei-csi-controller.yaml file as follows:
- If the Kubernetes version is earlier than v1.20, the snapshot feature is not supported. In this case, delete the snapshot-related container configurations items csi-snapshotter and snapshot-controller. If the Kubernetes version is earlier than v1.17, modify the ./deploy/huawei-csi-controller.yaml file as follows:
- If the Kubernetes version is earlier than v1.17, the snapshot feature is not supported. In this case, delete the snapshot-related container configurations items csi-snapshotter and snapshot-controller.
- If the Kubernetes version is earlier than v1.17, the csi-provisioner sidecar image provided by the Kubernetes community does not support the –leader-election parameter. Therefore, the leader-election parameter of the csi-provisioner container is deleted and only single-copy deployment is supported.
- Modify the dependent image version based on the version requirements in Checking the Images on Which CSI Depends .
```
kubectl apply -f ./deploy/huawei-csi-controller.yaml
```
Run the following command to install the huawei-csi-node service.
```
kubectl apply -f ./deploy/huawei-csi-node.yaml 
```

Run the following command to check whether the services are started.

kubectl get pod -n huawei-csi

The following is an example of the command output. If the Pod status is Running, the installation is successful.

NAME                                     READY   STATUS    RESTARTS   AGE
huawei-csi-controller-68745d489c-v5xkj   9/9     Running   0          13m
huawei-csi-node-4hbqp                    3/3     Running   0          13m
huawei-csi-node-f7dkf                    3/3     Running   0          13m
huawei-csi-node-xrntc                    3/3     Running   0          13m

In the multi-copy controller deployment scenario, you can modify the spec.replica field of the Deployment resource in the ./deploy/huawei-csi-controller.yaml file to specify the number of copies. After the modification, run the kubectl apply -f ./deploy/huawei-csi-controller.yaml command to make the configuration take effect.