Container Service for Kubernetes:Use static volumes with CPFS General-purpose Edition

How it works

The main workflow for mounting a CPFS General Purpose static storage volume in an ACK cluster is as follows.

1. Create a PV: "Register" an existing CPFS General Purpose file system in the cluster by declaring its mount address, capacity, access mode, and other parameters.

2. Create a PVC: Applications "request" registered storage resources through a PVC. The PVC automatically binds to a PV that meets the specified conditions.

3. Mount in the application: Mount the bound PVC to the application Pod as a persistent directory inside the container.

Step 1: Create a PV

Create a PV in the cluster to "register" an existing CPFS file system.

1. Save the following content as a cpfs-pv.yaml file.

   apiVersion: v1
   kind: PersistentVolume
   metadata:
     name: cpfs-pv
     labels:  # Define labels for subsequent PVC binding
       alicloud-pvname: cpfs-pv
   spec:
     accessModes:  # Access mode
     - ReadWriteMany
     capacity:
       storage: 20Gi  # Define the total capacity of the PV
     csi:
       driver: nasplugin.csi.alibabacloud.com  # Fixed to this value
       volumeAttributes:
         mountProtocol: cpfs-nfs   # Use the NFS protocol for mounting
         path: "/share"    # The mount directory in the mount address
         volumeAs: subpath    # Subdirectory
         server: "cpfs-******-******.cn-shanghai.cpfs.aliyuncs.com" # The mount point domain name in the mount address
       volumeHandle: cpfs-pv  # Unique identifier of the volume, must be consistent with metadata.name       
     mountOptions:  # NFS mount options
     - rsize=1048576,wsize=1048576,hard,timeo=600,retrans=2,noresvport
     - vers=3

   Key parameter description:

   Parameter	Description
   accessModes	Access mode for the PV.
   capacity.storage	Storage capacity of the volume.
   csi.driver	Driver type. When using CPFS General Purpose, this is fixed to nasplugin.csi.alibabacloud.com.
   csi.volumeAttributes.server	The mount point domain name in the mount address of the CPFS General Purpose protocol service export directory.
   csi.volumeAttributes.path	The mount directory in the mount address of the CPFS General Purpose protocol service export directory, for example, /share. You can also set it to a subdirectory, for example, /share/dir.
   csi.volumeAttributes.mountProtocol	cpfs-nfs indicates using the NFS protocol to mount CPFS.
   csi.volumeAttributes.volumeAs	subpath indicates creating a subdirectory type PV.
   csi.volumeHandle	Must be consistent with the PV name.

2. Run the following command to create the PV.

   kubectl apply -f cpfs-pv.yaml

3. View the PV status.

   kubectl get pv cpfs-pv

   Expected output:

   NAME      CAPACITY   ACCESS MODES   RECLAIM POLICY   STATUS      CLAIM   STORAGECLASS   REASON   AGE
   cpfs-pv   20Gi       RWX            Retain           Available                                   2s

Step 2: Create a PVC

Create a PVC to "request" the use of the storage resources registered in the previous step.

1. Save the following content as cpfs-pvc.yaml.

   apiVersion: v1
   kind: PersistentVolumeClaim
   metadata:
     name: cpfs-pvc
   spec:
     accessModes:
     - ReadWriteMany  # Access mode
     resources:
       requests: 
         storage: 20Gi  # Requested storage capacity
     selector:  # 
       matchLabels:
         alicloud-pvname: cpfs-pv  # Bind to the previously created PV

   PV parameters	Description
   accessModes	Access mode that the PVC requests from the PV.
   selector	Uses the label on the PV for matching and binding.
   resources.requests.storage	The storage capacity allocated to the Pod. It must not exceed the PV capacity.

2. Create the PV and PVC.

   kubectl apply -f cpfs-pvc.yaml

3. View the PVC status to confirm that it has been successfully bound to the PV.

   kubectl get pvc cpfs-pvc

   The following is an example of the output:

   NAME       STATUS   VOLUME    CAPACITY   ACCESS MODES   STORAGECLASS   VOLUMEATTRIBUTESCLASS   AGE
   cpfs-pvc   Bound    cpfs-pv   20Gi       RWO                           <unset>                 18m

Step 3: Create the application and mount the volume

Create an application workload and declare the use of the previously created PVC in its configuration.

1. Save the following content as cpfs-test.yaml.

   apiVersion: apps/v1
   kind: StatefulSet
   metadata:
     name: cpfs-sts
   spec:
     selector:
       matchLabels:
         app: nginx
     serviceName: "nginx"
     replicas: 2
     template:
       metadata:
         labels:
           app: nginx
       spec:
         containers:
         - name: nginx
           image: anolis-registry.cn-zhangjiakou.cr.aliyuncs.com/openanolis/nginx:1.14.1-8.6
           volumeMounts:
           - name: cpfs-pvc
             mountPath: /data
         volumes:
         - name: cpfs-pvc
           persistentVolumeClaim:
             claimName: cpfs-pvc

2. Create the StatefulSet.

   kubectl apply -f cpfs-test.yaml

3. View the Pod status.

   kubectl get pods -l app=nginx

   Expected output:

   NAME         READY   STATUS    RESTARTS   AGE
   cpfs-sts-0   1/1     Running   0          58s
   cpfs-sts-1   1/1     Running   0          50s

4. Check the mount point inside the container to confirm that CPFS has been mounted successfully.

   kubectl exec cpfs-sts-0 -- mount | grep /data

   The following expected output indicates that the CPFS General Purpose static volume has been mounted successfully.

   cpfs-******-******.cn-shanghai.cpfs.aliyuncs.com:/share on /data type nfs (rw,relatime,vers=3,rsize=1048576,wsize=1048576,namlen=255,hard,nolock,noresvport,proto=tcp,port=30000,timeo=600,retrans=2,sec=sys,mountaddr=127.0.1.255,mountvers=3,mountport=30000,mountproto=tcp,local_lock=all,addr=127.0.1.255)

Verify shared storage and persistent storage

After the application is successfully deployed, you can verify whether the storage volume works as expected.

Recommendations for production environments

• Performance tuning: CPFS is suitable for high-throughput and high-IOPS scenarios. You can adjust NFS mount options in the mountOptions of the PV (such as rsize and wsize) to optimize performance and meet specific workload requirements.

• Data consistency: As shared storage, CPFS does not guarantee data consistency when multiple Pods concurrently write to the same file. To avoid file corruption, you need to implement distributed locks at the application layer or use an append-only write mode.

• Storage volume management: The static storage volume approach requires manual PV management. For scenarios that require automatically creating and isolating storage volumes for multiple applications, we recommend using CNFS to manage CPFS dynamic storage volumes. For more information, see Manage dynamically provisioned volumes of General-purpose CPFS.