Container Service for Kubernetes:Mount a statically provisioned OSS volume

Usage notes

• OSS buckets do not support dynamically provisioned persistent volumes (PVs). We recommend that you do not use OSS buckets across accounts.

• kubelet is restarted when an ACK cluster is updated. As a result, ossfs is restarted and the mounted OSS directory becomes unavailable. In this case, you must recreate the pods to which the OSS volume is mounted. You can add health check settings in the YAML file to restart the pods and remount the OSS volume when the OSS directory becomes unavailable.

  Note

  If the csi-plugin and csi-provisioner components that you use are of V1.18.8.45 or later, the preceding issue does not occur.

• If the securityContext.fsgroup parameter is specified in the application template, kubelet performs the chmod or chown operation after the volume is mounted, which increases the time consumption.

  Note

  For more information about how to speed up the mounting process when the securityContext.fsgroup parameter is specified, see the Why does it require a long period of time to mount an OSS volume? section of the "FAQ about OSS volumes" topic.

• When you use ossfs to perform List operations, HTTP requests are sent to OSS to retrieve the metadata of the requested files. If a directory contains large numbers of files, such as more than 100,000 files, ossfs occupies large amounts of system memory. As a result, Out of Memory (OOM) errors may occur in pods. The actual number of files that may cause OOM errors is related to the node memory. To resolve this issue, you can divide the directory or mount a subdirectory in the OSS bucket.

• ossfs is applicable to concurrent read scenarios. We recommend that you set the access modes of persistent volume claims (PVCs) and PVs to ReadOnlyMany in concurrent read scenarios. To handle write operations, we recommend that you use OSS SDKs or ossutil to split reads and writes. For more information, see Best practice for OSS read/write splitting.

  Note

  When you use ossfs to write data to an OSS volume whose access mode is ReadWriteMany, take note of the following items:

  • ossfs cannot guarantee the consistency of data written by concurrent write operations.

  • After the OSS volume is mounted to a pod, if you log on to the pod or the host of the pod and delete or modify a file in the mount path, the source file in the OSS bucket is also deleted or modified. To prevent accidental deletion of important data, you can enable version control for the OSS bucket. For more information, see Overview.

• If you want to upload a file larger than 10 MB in size to OSS, you can split the file into multiple parts and separately upload the parts. To prevent the parts from incurring additional storage fees if a multipart upload task is interrupted, you can use one of the following methods to delete the parts that are no longer needed:

  • Manually delete parts. For more information, see Delete parts.

  • Configure lifecycle rules to automatically delete parts. For more information, see Lifecycle rules based on the last modified time.

Use RRSA authentication to mount a statically provisioned OSS volume

You can use the RRSA feature to enforce access control on different PVs that are deployed in an ACK cluster. This implements fine-grained API permission control on PVs and reduces security risks. For more information, see Use RRSA to authorize different pods to access different cloud services.

(Optional) Step 1: Create a RAM role

If you use the RRSA feature for the first time in your cluster, perform the following steps. If you have used RRSA authentication to mount OSS volumes in the cluster, skip this step.

1. Log on to the ACK console and enable RRSA. For more information, see Use RRSA to authorize different pods to access different cloud services.

2. Create a RAM role for mounting OSS volumes by using RRSA authentication. The RAM role is assumed by your cluster when it uses RRSA to mount OSS volumes.

   When you create the RAM role, select IdP for the Select Trusted Entity parameter. In this example, the name of the RAM role is demo-role-for-rrsa.

   1. Log on to the RAM console with your Alibaba Cloud account.

   2. In the left-side navigation pane, choose Identities > Roles. On the Roles page, click Create Role.

   3. In the Create Role panel, select IdP for Select Trusted Entity and click Next.

   4. In the Configure Role step, configure the parameters and click OK.

      The following table describes the parameters that are configured in this example.

      Parameter	Description
      RAM Role Name	The name of the RAM role. In this example, demo-role-for-rrsa is used.
      Note	Optional. The description of the RAM role.
      IdP Type	The type of the identity provider (IdP). Select OIDC.
      Select IdP	The IdP that you want to use. Select a value in the ack-rrsa-<cluster_id> format. <cluster_id> indicates the ID of your cluster.
      Conditions	oidc:iss: Use the default value. oidc:aud: Select sts.aliyuncs.com. oidc:sub: Set the condition operator to StringEquals. In this example, system:serviceaccount:ack-csi-fuse:csi-fuse-ossfs is used.

Step 2: Grant permissions to the demo-role-for-rrsa role

1. Create the following custom policies to grant OSS permissions to the RAM user. For more information, see Create custom policies.

   Replace mybucket in the following policies with the name of your OSS bucket.

   • Policy that grants read-only permissions on OSS

     Policy document

     {
         "Statement": [
             {
                 "Action": [
                     "oss:Get*",
                     "oss:List*"
                 ],
                 "Effect": "Allow",
                 "Resource": [
                     "acs:oss:*:*:mybucket",
                     "acs:oss:*:*:mybucket/*"
                 ],
             }
         ],
         "Version": "1"
     }

   • Policy that grants read and write permissions on OSS

     Policy document

     {
         "Statement": [
             {
                 "Action": "oss:*",
                 "Effect": "Allow",
                 "Resource": [
                     "acs:oss:*:*:mybucket",
                     "acs:oss:*:*:mybucket/*"
                 ],
             }
         ],
         "Version": "1"
     }

2. Optional. If the files in the OSS bucket are encrypted by using a customer master key (CMK) in Key Management Service (KMS), you need to grant KMS permissions to the RAM user. For more information, see the Encrypt an OSS volume section of the "Encrypt an OSS volume" topic.

3. Grant the required permissions to the demo-role-for-rrsa role. For more information, see Grant permissions to a RAM role.

   Note

   If you want to use an existing RAM role, you can modify the trust policy of the RAM role to which the required OSS permissions are granted. For more information, see the Use an existing RAM role and grant the required permissions to the RAM role section of the "Use RRSA to authorize different pods to access different cloud services" topic.

Step 3: Create a PV and a PVC

1. Create a PV that uses RRSA authentication.

   1. Use the following sample template to create a PV configuration file named pv-rrsa.yaml. In this file, RRSA authentication is enabled.

      apiVersion: v1
      kind: PersistentVolume
      metadata:
        name: pv-oss
        labels:    
          alicloud-pvname: pv-oss
      spec:
        capacity:
          storage: 5Gi
        accessModes:
          - ReadOnlyMany
        persistentVolumeReclaimPolicy: Retain
        csi:
          driver: ossplugin.csi.alibabacloud.com
          volumeHandle: pv-oss # Specify the name of the PV. 
          volumeAttributes:
            bucket: "oss"
            url: "oss-cn-hangzhou.aliyuncs.com"
            otherOpts: "-o umask=022 -o max_stat_cache_size=0 -o allow_other"
            authType: "rrsa"
            roleName: "demo-role-for-rrsa"

      Parameter	Description
      name	The name of the PV.
      labels	The labels that are added to the PV.
      storage	The available storage of the OSS bucket.
      accessModes	The access mode. Valid values: ReadOnlyMany and ReadWriteMany. If you select ReadOnlyMany, ossfs mounts the OSS bucket in read-only mode.
      persistentVolumeReclaimPolicy	The reclaim policy of the PV.
      driver	The type of driver. In this example, the parameter is set to ossplugin.csi.alibabacloud.com. This indicates that the OSS CSI component is used.
      volumeHandle	The name of the PV.
      bucket	The OSS bucket that you want to mount.
      url	The endpoint of the OSS bucket that you want to mount. If the node and the OSS bucket reside in the same region, or are connected through a virtual private cloud (VPC), use an internal endpoint of the OSS bucket. If the node and the OSS bucket reside in different regions, use a public endpoint of the OSS bucket. Endpoint formats: Internal endpoint: http://oss-{{regionName}}-internal.aliyuncs.com or https://oss-{{regionName}}-internal.aliyuncs.com Public endpoint: http://oss-{{regionName}}.aliyuncs.com or https://oss-{{regionName}}.aliyuncs.com. Important The actual endpoint is displayed on the Overview page in the OSS console. The internal access port for vpc100-oss-{{regionName}}.aliyuncs.com is deprecated. Make sure to update your configuration accordingly.
      otherOpts	You can configure custom parameters in the -o *** -o *** format for the OSS volume. Example: -o umask=022 -o max_stat_cache_size=0 -o allow_other. umask: modifies the permission mask of files in ossfs. For example, if you specify umask=022, the permission mask of files in ossfs changes to 755. By default, the permission mask of files uploaded by using OSS SDKs or console is 640 in ossfs. Therefore, we recommend that you specify the umask parameter if you want to split reads and writes. max_stat_cache_size: the maximum number of files whose metadata can be cached. Metadata caching can accelerate ls operations. However, if you modify files by using methods such as OSS SDKs, console, and ossutil, the cached metadata of the files is not synchronously updated. As a result, the cached metadata becomes outdated, and the results of ls operations may be inaccurate. allow_other: allows other users to access the mounted directory. However, these users cannot access the files in the directory. For more information about other parameters, see Options supported by ossfs.
      path	The path relative to the root directory of the OSS bucket to be mounted. Default value: /. This parameter is supported by csi-plugin of V1.14.8.32-c77e277b-aliyun and later. For ossfs versions earlier than 1.91, you must create this path in OSS in advance. For more information, see Features of ossfs 1.91 and later and ossfs performance benchmarking.
      authType	The authentication method. Set this parameter to rrsa, which indicates that RRSA authentication is used.
      roleName	The name of the RAM role that you want to use. Set this parameter to the name of the RAM role that you created or modified in Step 1. If you need to configure different permissions for different PVs, you can create multiple RAM roles and specify the RAM roles in the roleName parameter.

      Note

      For more information about how to use specified Alibaba Cloud Resource Names (ARNs) or ServiceAccounts in RRSA authentication, see the How do I use the specified ARNs or ServiceAccount in RRSA authentication? section of the "FAQ about OSS volumes" topic.

   2. Run the following command to create a PV that uses RRSA authentication:

      kubectl create -f pv-rrsa.yaml

2. Run the following command to create a statically provisioned PVC:

   kubectl create -f pvc-oss.yaml

   The following pvc-oss.yaml sample file is used to create the statically provisioned PVC:

   apiVersion: v1
   kind: PersistentVolumeClaim
   metadata:
     name: pvc-oss
   spec:
     accessModes:
       - ReadOnlyMany
     resources:
       requests:
         storage: 5Gi
     selector:
       matchLabels:
         alicloud-pvname: pv-oss

   Parameter	Description
   name	The name of the PVC.
   accessModes	The access mode. Valid values: ReadOnlyMany and ReadWriteMany. If you select ReadOnlyMany, ossfs mounts the OSS bucket in read-only mode.
   storage	The capacity claimed by the PVC. The claimed capacity cannot exceed the capacity of the PV that is bound to the PVC.
   alicloud-pvname	The labels that are used to select and bind a PV to the PVC. The labels must be the same as those of the PV to be bound to the PVC.

   In the left-side navigation pane, choose Volumes > Persistent Volume Claims. On the Persistent Volume Claims page, you can find the created PVC.

Step 4: Create an application

Create an application named oss-static and mount the PV to the application.

Run the following command to create a file named oss-static.yaml:

kubectl create -f oss-static.yaml

The following oss-static.yaml sample file is used to create the application:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: oss-static
  labels:
    app: nginx
spec:
  replicas: 2
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: nginx
        ports:
        - containerPort: 80
        volumeMounts:
          - name: pvc-oss
            mountPath: "/data"
          - name: pvc-oss
            mountPath: "/data1"
        livenessProbe:
          exec:
            command:
            - sh
            - -c
            - cd /data
          initialDelaySeconds: 30
          periodSeconds: 30
      volumes:
        - name: pvc-oss
          persistentVolumeClaim:
            claimName: pvc-oss

• livenessProbe: the health check configurations. For more information, see OSS volume overview.

• mountPath: the path to which the OSS bucket is mounted in the container.

• claimName: the name of the PVC that the application uses.

Use AccessKey authentication to mount a statically provisioned OSS volume as a RAM user

kubectl create -f oss-static.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: oss-static
  labels:
    app: nginx
spec:
  replicas: 2
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: nginx
        ports:
        - containerPort: 80
        volumeMounts:
          - name: pvc-oss
            mountPath: "/data"
          - name: pvc-oss
            mountPath: "/data1"
        livenessProbe:
          exec:
            command:
            - sh
            - -c
            - cd /data
          initialDelaySeconds: 30
          periodSeconds: 30
      volumes:
        - name: pvc-oss
          persistentVolumeClaim:
            claimName: pvc-oss

Check whether the OSS volume can persist and share data

1. View the pod that runs the oss-static application and the files in the OSS bucket.

   1. Run the following command to query the pod that runs the oss-static application:

      kubectl get pod

      Expected output:

      NAME                             READY   STATUS    RESTARTS   AGE
      oss-static-66fbb85b67-d****      1/1     Running   0          1h
      oss-static-66fbb85b67-l****      1/1     Running   0          1h

   2. Create a tmpfile file.

      • If the OSS volume is mounted in ReadWriteMany mode, run the following commands to create the tmpfile file in the /data path:

        kubectl exec oss-static-66fbb85b67-d**** -- touch /data/tmpfile
        kubectl exec oss-static-66fbb85b67-l**** -- touch /data/tmpfile

      • If the OSS volume is mounted in ReadOnlyMany mode, use the OSS console or cp command (upload objects) to upload the tmpfile file to the corresponding path in the OSS bucket.

2. Run the following commands to query files in the /data path of the oss-static-66fbb85b67-d**** pod and query files in the /data1 path of the oss-static-66fbb85b67-l**** pod:

   kubectl exec oss-static-66fbb85b67-d**** -- ls /data | grep tmpfile
   kubectl exec oss-static-66fbb85b67-l**** -- ls /data1 | grep tmpfile

   Expected output:

   tmpfile

   The output indicates that the file exists in both pods. This means that the pods share the data stored in the OSS volume.

   Note

   If no output is returned, check whether the version of the CSI component is 1.20.7 or later. For more information, see csi-plugin.

3. Run the following command to delete the oss-static-66fbb85b67-d**** pod:

   kubectl delete pod oss-static-66fbb85b67-d****

   Expected output:

   pod "oss-static-66fbb85b67-d****" deleted

4. Verify that the file still exists after the pod is deleted.

   1. Run the following command to query the pod that is recreated:

      kubectl get pod

      Expected output:

      NAME                             READY   STATUS    RESTARTS   AGE
      oss-static-66fbb85b67-l****      1/1     Running   0          1h
      oss-static-66fbb85b67-z****      1/1     Running   0          40s

   2. Run the following command to query files in the /data path of the pod:

      kubectl exec oss-static-66fbb85b67-z**** -- ls /data | grep tmpfile

      Expected output:

      tmpfile

      The output indicates that the temfile file still exists. This means that the OSS volume can persist data.