How do I deploy a DataCache CRD? - Elastic Container Instance

Pulling large datasets into ECI containers often causes slow pod startup. The DataCache CRD lets Kubernetes users pre-pull data into a cache so containers start with data already available. This topic describes how to deploy the DataCache CRD controller in a Kubernetes cluster using Helm (recommended) or a YAML file.

Billing

Cluster type	Billing
Container Service for Kubernetes (ACK) cluster	The DataCache CRD deploys on an existing Elastic Compute Service (ECS) node. No additional charges apply.
ACK Serverless cluster	Deploying the DataCache CRD creates a pod with 0.25 vCPU and 0.5 GiB of memory. You are charged for this pod: `Price = (0.25 × unit price of vCPUs + 0.5 × unit price of memory) × running duration`. For pricing details, see Billing of elastic container instances.

Prerequisites

Before you begin, ensure that you have:

An AccessKey ID and AccessKey secret with permissions to call DataCache API operations. For details, see Create an AccessKey pair
Important
The Alibaba Cloud account or RAM user associated with the AccessKey pair must have permissions to call DataCache-related API operations.
The region ID of the cluster

Deploy using Helm (recommended)

Helm is the recommended deployment method. It handles dependency management and simplifies upgrades.

Helm 3.7.0 or later

Connect to the Kubernetes cluster.
Check your Helm version.
```
helm version
```
If your Helm version is 3.7.x, enable OCI registry support.
```
export HELM_EXPERIMENTAL_OCI=1
```
Important
Helm 3.7.x requires this flag to pull charts from OCI-compatible registries. Helm 3.8.0 and later enable OCI support by default — skip this step if you are on 3.8.0 or later.
Set the environment variables.
```
export REPO="oci://eci-release-registry.cn-hangzhou.cr.aliyuncs.com/datacache/datacache-operator"
export AK='<AccessKey ID>'
export SK='<AccessKey secret>'
export REGION='<Region ID>'
```
Important
The REPO value points to the Alibaba Cloud Container Registry hosting the DataCache operator chart. Do not change the region ID embedded in the URL.

Deploy the DataCache CRD.

helm install datacache-operator $REPO --set accessKey=$AK --set secretKey=$SK --set regionId=$REGION

Verify the deployment.

kubectl get pods -n datacache-controller

The controller pod should be in Running state:

NAME                                          READY   STATUS    RESTARTS   AGE
aliyun-datacache-controller-xxxxxxxxx-xxxxx   1/1     Running   0          30s

Helm earlier than 3.7.0

Connect to the Kubernetes cluster.
Check your Helm version.
```
helm version
```
Install the Container Registry plugin for Helm. If you have access to GitHub, run:
```
helm plugin install https://github.com/AliyunContainerService/helm-acr
```
If your network connection to GitHub is unstable, use the Alibaba Cloud mirror instead:
```
git clone https://github.com/AliyunContainerService/helm-acr.git
sed -i 's/github.com/helm-acr-releases.oss-cn-hangzhou.aliyuncs.com/g' helm-acr/scripts/install_plugin.sh
helm plugin install helm-acr
```
Important
The sed command replaces the GitHub download URL in the install script with an Alibaba Cloud mirror URL (helm-acr-releases.oss-cn-hangzhou.aliyuncs.com), which is more reliable on networks with restricted access to GitHub.

Add the DataCache Helm repository.

helm repo add datacache acr://eci-release-chart.cn-hangzhou.cr.aliyuncs.com/datacache/datacache-operator
helm repo update

Set the environment variables.

export REPO="datacache/datacache-operator"
export AK='<AccessKey ID>'
export SK='<AccessKey secret>'
export REGION='<Region ID>'

Deploy the DataCache CRD.

helm install datacache-operator $REPO --set accessKey=$AK --set secretKey=$SK --set regionId=$REGION

Verify the deployment.

kubectl get pods -n datacache-controller

The controller pod should be in Running state:

NAME                                          READY   STATUS    RESTARTS   AGE
aliyun-datacache-controller-xxxxxxxxx-xxxxx   1/1     Running   0          30s

Deploy using a YAML file

Connect to the Kubernetes cluster.

Create a file named aliyun-datacache-controller.yaml with the following content. Replace the three placeholder values marked with *** in the template: the AccessKey ID (line 13), AccessKey secret (line 14), and region ID (line 211).

apiVersion: v1
kind: Namespace
metadata:
  name: datacache-controller
---
apiVersion: v1
kind: Secret
metadata:
  name: accesskey.eci.aliyun.com
  namespace: datacache-controller
type: Opaque
stringData:
  accessKey: ***            # Enter the AccessKey ID.
  secretKey: ***            # Enter the AccessKey secret.
---
apiVersion: apiextensions.k8s.io/v1
kind: CustomResourceDefinition
metadata:
  name: datacaches.eci.aliyun.com
spec:
  group: eci.aliyun.com
  scope: Cluster
  names:
    kind: DataCache
    plural: datacaches
    singular: datacache
    shortNames:
      - edc
    categories:
      - all
  versions:
    - name: v1alpha1
      served: true
      storage: true
      subresources:
        status: {}
      schema:
        openAPIV3Schema:
          type: object
          required:
            - spec
          properties:
            spec:
              type: object
              required:
                - path
                - dataSource
              properties:
                bucket:
                  type: string
                  default: default
                  description: the name of a bucket. The default name is default. The bucket named eci-system is reserved to store common caches of Elastic Container Instance and cannot be used as a value.
                path:
                  type: string
                  description: the directory of the bucket.
                dataSource:
                  type: object
                  description: the cache data source.
                  properties:
                    type:
                      type: string
                      description: the property of the cache data source. Valid values: NAS, OSS, URL, and SNAPSHOT.
                    options:
                      type: object
                      description: the options for a specific type of cache data source. For more information, see the DataCache documentation.
                      x-kubernetes-preserve-unknown-fields: true
                size:
                  type: integer
                  default: 20
                  description: the size of the data cache. You must plan the size. Default value: 20 GiB, which is the minimum size of an elastic block storage (EBS).
                retentionDays:
                  type: integer
                  description: the retention period of the data cache. By default, data caches never expire.
                netConfig:
                  type: object
                  properties:
                    securityGroupId:
                      type: string
                      description: the ID of the security group.
                    vSwitchId:
                      type: string
                      description: the ID of the vSwitch.
                    eipInstanceId:
                      type: string
                      description: the ID of the elastic IP address (EIP). If no source network address translation (SNAT) entry is created for the vSwitch, you can bind an EIP to pull Internet data.
                    eipCreateParam:
                      type: object
                      description: the parameter that is used to create an EIP. You can specify EipInstanceId or EipCreateParam.
                      properties:
                        bandwidth:
                          type: integer
                          description: the public bandwidth. Unit: MB. Default value: 5.
                        commonBandwidthPackage:
                          type: string
                          description: the EIP bandwidth plan.
                        internetChargeType:
                          type: string
                          description: the billing method of the EIP.
                        publicIpAddressPoolId:
                          type: string
                          description: the IP address pool.
                        ISP:
                          type: string
                          description: the EIP line type.
            status:
              type: object
              properties:
                status:
                  type: string
                progress:
                  type: string
                snapshotId:
                  type: string
                dataCacheId:
                  type: string
      additionalPrinterColumns:
        - name: Age
          type: date
          jsonPath: .metadata.creationTimestamp
        - name: dataCacheId
          type: string
          jsonPath: .status.dataCacheId
        - name: Status
          type: string
          jsonPath: .status.status
        - name: Progress
          type: string
          jsonPath: .status.progress
        - name: Bucket
          type: string
          jsonPath: .spec.bucket
        - name: Path
          type: string
          jsonPath: .spec.path
---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: aliyun-datacache-controller
  namespace: datacache-controller
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: aliyun-datacache-controller
  namespace: datacache-controller
rules:
  - apiGroups:
      - "eci.aliyun.com"
    resources:
      - datacaches
      - datacaches/status
    verbs:
      - "*"
  - apiGroups:
      - ""
    resources:
      - events
    verbs:
      - create
      - patch
      - update
  - apiGroups:
      - ""
    resources:
      - configmaps
    verbs:
      - get
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: aliyun-datacache-controller
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: aliyun-datacache-controller
subjects:
  - kind: ServiceAccount
    name: aliyun-datacache-controller
    namespace: datacache-controller
---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: aliyun-datacache-controller
  namespace: datacache-controller
spec:
  selector:
    matchLabels:
      app: aliyun-datacache-controller
  replicas: 1
  template:
    metadata:
      labels:
        app: aliyun-datacache-controller
    spec:
      serviceAccountName: aliyun-datacache-controller
      containers:
      - name: datacache-operator
        image: eci-release-registry.cn-hangzhou.cr.aliyuncs.com/datacache/datacache-operator:latest
        resources:
          limits:
            memory: "512Mi"
            cpu: "250m"
          requests:
            memory: "64Mi"
            cpu: "50m"
        env:
        - name: REGION_ID
          value: *** # Specify the actual region ID.
        - name: ACCESS_KEY
          valueFrom:
            secretKeyRef:
              name: accesskey.eci.aliyun.com
              key: accessKey
        - name: SECRET_KEY
          valueFrom:
            secretKeyRef:
              name: accesskey.eci.aliyun.com
              key: secretKey

Deploy the DataCache CRD.

kubectl apply -f aliyun-datacache-controller.yaml

Verify the deployment.

kubectl get pods -n datacache-controller

The controller pod should be in Running state:

NAME                                          READY   STATUS    RESTARTS   AGE
aliyun-datacache-controller-xxxxxxxxx-xxxxx   1/1     Running   0          30s