使用 AWS ParallelCluster API - AWS ParallelCluster

本文為英文版的機器翻譯版本,如內容有任何歧義或不一致之處,概以英文版為準。

使用 AWS ParallelCluster API

在本教學課程中,您可以使用 Amazon API Gateway API 和 AWS ParallelCluster CloudFormation 範本建置和測試 。然後,您可以使用 上可用的範例用戶端 GitHub 來使用 API。如需使用 的詳細資訊API,請參閱 AWS ParallelCluster API

本教學課程摘錄自 HPC For Public Sector Customer Workshop

使用 AWS ParallelCluster 命令列界面 (CLI) 或 時API,您只需支付建立或更新 AWS ParallelCluster 映像和叢集時建立 AWS 的資源。如需詳細資訊,請參閱AWS 所使用的 服務 AWS ParallelCluster

AWS ParallelCluster UI 建立在無伺服器架構上,您可以在 AWS 免費方案類別中用於大多數案例。如需詳細資訊,請參閱AWS ParallelCluster 使用者介面成本

必要條件

保留在您的家用使用者目錄中並啟用您的虛擬環境:
  1. 安裝實用的JSON命令列處理器。

    $ sudo yum groupinstall -y "Development Tools" sudo yum install -y jq python3-devel
  2. 執行下列命令以取得您的 AWS ParallelCluster 版本,並將其指派給環境變數。

    $ PCLUSTER_VERSION=$(pcluster version | jq -r '.version') echo "export PCLUSTER_VERSION=${PCLUSTER_VERSION}" |tee -a ~/.bashrc
  3. 建立環境變數,並將您的區域 ID 指派給該變數。

    $ export AWS_DEFAULT_REGION="us-east-1" echo "export AWS_DEFAULT_REGION=${AWS_DEFAULT_REGION}" |tee -a ~/.bashrc
  4. 執行下列命令來部署 API。

    API_STACK_NAME="pc-api-stack" echo "export API_STACK_NAME=${API_STACK_NAME}" |tee -a ~/.bashrc
    aws cloudformation create-stack \ --region ${AWS_DEFAULT_REGION} \ --stack-name ${API_STACK_NAME} \ --template-url https://${AWS_DEFAULT_REGION}-aws-parallelcluster.s3.${AWS_DEFAULT_REGION}.amazonaws.com/parallelcluster/${PCLUSTER_VERSION}/api/parallelcluster-api.yaml \ --capabilities CAPABILITY_NAMED_IAM CAPABILITY_AUTO_EXPAND \ --parameters ParameterKey=EnableIamAdminAccess,ParameterValue=true { "StackId": "arn:aws:cloudformation:us-east-1:123456789012:stack/my-api-stack/abcd1234-ef56-gh78-ei90-1234abcd5678" }

    程序完成後,繼續下一個步驟。

  1. 登入 AWS Management Console。

  2. 導覽至 Amazon API Gateway 主控台

  3. 選擇您的API部署。

    Amazon API Gateway 主控台,其中包含您可以從中選擇的閘道清單。
  4. 選擇階段並選取階段。

    您可以從中選擇的階段的主控台檢視。您也可以檢視 URL API Gateway 為 提供的 API。
  5. 請注意URL,APIGateway 提供存取或叫用您的 API。它以藍色反白顯示。

  6. 選擇資源 ,然後選擇 GET 下的 /clusters

  7. 選擇 TEST 圖示,然後向下捲動並選擇TEST圖示。

    資源API和測試機制的主控台檢視。

    您的 回應/clusters GET隨即出現。

    資源API、測試機制和測試請求回應的主控台檢視。

複製 AWS ParallelCluster 原始程式碼、cdapi目錄,並安裝 Python 用戶端程式庫。

  1. $ git clone -b v${PCLUSTER_VERSION} https://github.com/aws/aws-parallelcluster aws-parallelcluster-v${PCLUSTER_VERSION} cd aws-parallelcluster-v${PCLUSTER_VERSION}/api
    $ pip3 install client/src
  2. 導覽回您的主要使用者目錄。

  3. 匯出用戶端在執行時使用URL的API閘道基礎。

    $ export PCLUSTER_API_URL=$( aws cloudformation describe-stacks --stack-name ${API_STACK_NAME} --query 'Stacks[0].Outputs[?OutputKey==`ParallelClusterApiInvokeUrl`].OutputValue' --output text ) echo "export PCLUSTER_API_URL=${PCLUSTER_API_URL}" |tee -a ~/.bashrc
  4. 匯出用戶端用來建立叢集的叢集名稱。

    $ export CLUSTER_NAME="test-api-cluster" echo "export CLUSTER_NAME=${CLUSTER_NAME}" |tee -a ~/.bashrc
  5. 執行下列命令,以存放範例用戶端用來存取 的憑證API。

    $ export PCLUSTER_API_USER_ROLE=$( aws cloudformation describe-stacks --stack-name ${API_STACK_NAME} --query 'Stacks[0].Outputs[?OutputKey==`ParallelClusterApiUserRole`].OutputValue' --output text ) echo "export PCLUSTER_API_USER_ROLE=${PCLUSTER_API_USER_ROLE}" |tee -a ~/.bashrc
  1. 將下列用戶端程式碼範例複製到您主要使用者目錄中test_pcluster_client.py的 。用戶端程式碼會發出執行下列動作的請求:

    • 建立 叢集

    • 描述叢集。

    • 列出叢集。

    • 描述運算機群。

    • 描述叢集執行個體。

    # Copyright 2021 Amazon.com, Inc. or its affiliates. All Rights Reserved. # SPDX-License-Identifier: MIT-0 # # Permission is hereby granted, free of charge, to any person obtaining a copy of this # software and associated documentation files (the "Software"), to deal in the Software # without restriction, including without limitation the rights to use, copy, modify, # merge, publish, distribute, sublicense, and/or sell copies of the Software, and to # permit persons to whom the Software is furnished to do so. # # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, # INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A # PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT # HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION # OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE # SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. # # Author: Evan F. Bollig (Github: bollig) import time, datetime import os import pcluster_client from pprint import pprint from pcluster_client.api import ( cluster_compute_fleet_api, cluster_instances_api, cluster_operations_api ) from pcluster_client.model.create_cluster_request_content import CreateClusterRequestContent from pcluster_client.model.cluster_status import ClusterStatus region=os.environ.get("AWS_DEFAULT_REGION") # Defining the host is optional and defaults to http://localhost # See configuration.py for a list of all supported configuration parameters. configuration = pcluster_client.Configuration( host = os.environ.get("PCLUSTER_API_URL") ) cluster_name=os.environ.get("CLUSTER_NAME") # Enter a context with an instance of the API client with pcluster_client.ApiClient(configuration) as api_client: cluster_ops = cluster_operations_api.ClusterOperationsApi(api_client) fleet_ops = cluster_compute_fleet_api.ClusterComputeFleetApi(api_client) instance_ops = cluster_instances_api.ClusterInstancesApi(api_client) # Create cluster build_done = False try: with open('cluster-config.yaml', encoding="utf-8") as f: body = CreateClusterRequestContent(cluster_name=cluster_name, cluster_configuration=f.read()) api_response = cluster_ops.create_cluster(body, region=region) except pcluster_client.ApiException as e: print("Exception when calling create_cluster: %s\n" % e) build_done = True time.sleep(60) # Confirm cluster status with describe_cluster while not build_done: try: api_response = cluster_ops.describe_cluster(cluster_name, region=region) pprint(api_response) if api_response.cluster_status == ClusterStatus('CREATE_IN_PROGRESS'): print('. . . working . . .', end='', flush=True) time.sleep(60) elif api_response.cluster_status == ClusterStatus('CREATE_COMPLETE'): print('READY!') build_done = True else: print('ERROR!!!!') build_done = True except pcluster_client.ApiException as e: print("Exception when calling describe_cluster: %s\n" % e) # List clusters try: api_response = cluster_ops.list_clusters(region=region) pprint(api_response) except pcluster_client.ApiException as e: print("Exception when calling list_clusters: %s\n" % e) # DescribeComputeFleet try: api_response = fleet_ops.describe_compute_fleet(cluster_name, region=region) pprint(api_response) except pcluster_client.ApiException as e: print("Exception when calling compute fleet: %s\n" % e) # DescribeClusterInstances try: api_response = instance_ops.describe_cluster_instances(cluster_name, region=region) pprint(api_response) except pcluster_client.ApiException as e: print("Exception when calling describe_cluster_instances: %s\n" % e)
  2. 建立叢集組態。

    $ pcluster configure --config cluster-config.yaml
  3. API 用戶端程式庫會自動偵測來自環境變數 (例如 AWS_ACCESS_KEY_IDAWS_SECRET_ACCESS_KEYAWS_SESSION_TOKEN) 或 的組態詳細資訊$HOME/.aws。下列命令會將您目前的IAM角色切換至指定的 ParallelClusterApiUserRole.

    $ eval $(aws sts assume-role --role-arn ${PCLUSTER_API_USER_ROLE} --role-session-name ApiTestSession | jq -r '.Credentials | "export AWS_ACCESS_KEY_ID=\(.AccessKeyId)\nexport AWS_SECRET_ACCESS_KEY=\(.SecretAccessKey)\nexport AWS_SESSION_TOKEN=\(.SessionToken)\n"')

    要注意的錯誤:

    如果您看到類似下列的錯誤,表示您已擔任 ParallelClusterApiUserRole 您的 AWS_SESSION_TOKEN 已過期。

    An error occurred (AccessDenied) when calling the AssumeRole operation: 
    User: arn:aws:sts::XXXXXXXXXXXX:assumed-role/ParallelClusterApiUserRole-XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX/ApiTestSession 
    is not authorized to perform: sts:AssumeRole on resource: arn:aws:iam::XXXXXXXXXXXX:role/ParallelClusterApiUserRole-XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX

    捨棄角色,然後重新執行aws sts assume-role命令以使用 ParallelClusterApiUserRole.

    $ unset AWS_SESSION_TOKEN unset AWS_SECRET_ACCESS_KEY unset AWS_ACCESS_KEY_ID

    若要提供目前使用者API存取許可,您必須展開資源政策

  4. 執行下列命令來測試範例用戶端。

    $ python3 test_pcluster_client.py {'cluster_configuration': 'Region: us-east-1\n' 'Image:\n' ' Os: alinux2\n' 'HeadNode:\n' ' InstanceType: t2.micro\n' ' Networking . . . :\n' ' SubnetId: subnet-1234567890abcdef0\n' ' Ssh:\n' ' KeyName: adpc\n' 'Scheduling:\n' ' Scheduler: slurm\n' ' SlurmQueues:\n' ' - Name: queue1\n' ' ComputeResources:\n' ' - Name: t2micro\n' ' InstanceType: t2.micro\n' ' MinCount: 0\n' ' MaxCount: 10\n' ' Networking . . . :\n' ' SubnetIds:\n' ' - subnet-1234567890abcdef0\n', 'cluster_name': 'test-api-cluster'} {'cloud_formation_stack_status': 'CREATE_IN_PROGRESS', 'cloudformation_stack_arn': 'arn:aws:cloudformation:us-east-1:123456789012:stack/test-api-cluster/abcd1234-ef56-gh78-ij90-1234abcd5678', 'cluster_configuration': {'url': 'https://parallelcluster-021345abcdef6789-v1-do-not-delete...}, 'cluster_name': 'test-api-cluster', 'cluster_status': 'CREATE_IN_PROGRESS', 'compute_fleet_status': 'UNKNOWN', 'creation_time': datetime.datetime(2022, 4, 28, 16, 18, 47, 972000, tzinfo=tzlocal()), 'last_updated_time': datetime.datetime(2022, 4, 28, 16, 18, 47, 972000, tzinfo=tzlocal()), 'region': 'us-east-1', 'tags': [{'key': 'parallelcluster:version', 'value': '3.1.3'}], 'version': '3.1.3'} . . . . . . working . . . {'cloud_formation_stack_status': 'CREATE_COMPLETE', 'cloudformation_stack_arn': 'arn:aws:cloudformation:us-east-1:123456789012:stack/test-api-cluster/abcd1234-ef56-gh78-ij90-1234abcd5678', 'cluster_configuration': {'url': 'https://parallelcluster-021345abcdef6789-v1-do-not-delete...}, 'cluster_name': 'test-api-cluster', 'cluster_status': 'CREATE_COMPLETE', 'compute_fleet_status': 'RUNNING', 'creation_time': datetime.datetime(2022, 4, 28, 16, 18, 47, 972000, tzinfo=tzlocal()), 'head_node': {'instance_id': 'i-abcdef01234567890', 'instance_type': 't2.micro', 'launch_time': datetime.datetime(2022, 4, 28, 16, 21, 46, tzinfo=tzlocal()), 'private_ip_address': '172.31.27.153', 'public_ip_address': '52.90.156.51', 'state': 'running'}, 'last_updated_time': datetime.datetime(2022, 4, 28, 16, 18, 47, 972000, tzinfo=tzlocal()), 'region': 'us-east-1', 'tags': [{'key': 'parallelcluster:version', 'value': '3.1.3'}], 'version': '3.1.3'} READY!
  1. 將下列範例用戶端程式碼複製到 delete_cluster_client.py。用戶端程式碼會發出刪除叢集的請求。

    # Copyright 2021 Amazon.com, Inc. or its affiliates. All Rights Reserved. # SPDX-License-Identifier: MIT-0 # # Permission is hereby granted, free of charge, to any person obtaining a copy of this # software and associated documentation files (the "Software"), to deal in the Software # without restriction, including without limitation the rights to use, copy, modify, # merge, publish, distribute, sublicense, and/or sell copies of the Software, and to # permit persons to whom the Software is furnished to do so. # # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, # INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A # PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT # HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION # OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE # SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. # # Author: Evan F. Bollig (Github: bollig) import time, datetime import os import pcluster_client from pprint import pprint from pcluster_client.api import ( cluster_compute_fleet_api, cluster_instances_api, cluster_operations_api ) from pcluster_client.model.create_cluster_request_content import CreateClusterRequestContent from pcluster_client.model.cluster_status import ClusterStatus region=os.environ.get("AWS_DEFAULT_REGION") # Defining the host is optional and defaults to http://localhost # See configuration.py for a list of all supported configuration parameters. configuration = pcluster_client.Configuration( host = os.environ.get("PCLUSTER_API_URL") ) cluster_name=os.environ.get("CLUSTER_NAME") # Enter a context with an instance of the API client with pcluster_client.ApiClient(configuration) as api_client: cluster_ops = cluster_operations_api.ClusterOperationsApi(api_client) # Delete the cluster gone = False try: api_response = cluster_ops.delete_cluster(cluster_name, region=region) except pcluster_client.ApiException as e: print("Exception when calling delete_cluster: %s\n" % e) time.sleep(60) # Confirm cluster status with describe_cluster while not gone: try: api_response = cluster_ops.describe_cluster(cluster_name, region=region) pprint(api_response) if api_response.cluster_status == ClusterStatus('DELETE_IN_PROGRESS'): print('. . . working . . .', end='', flush=True) time.sleep(60) except pcluster_client.ApiException as e: gone = True print("DELETE COMPLETE or Exception when calling describe_cluster: %s\n" % e)
  2. 執行下列命令來刪除叢集。

    $ python3 delete_cluster_client.py {'cloud_formation_stack_status': 'DELETE_IN_PROGRESS', 'cloudformation_stack_arn': 'arn:aws:cloudformation:us-east-1:123456789012:stack/test-api-cluster/abcd1234-ef56-gh78-ij90-1234abcd5678', 'cluster_configuration': {'url': 'https://parallelcluster-021345abcdef6789-v1-do-not-delete...}, 'cluster_name': 'test-api-cluster', 'cluster_status': 'DELETE_IN_PROGRESS', 'compute_fleet_status': 'UNKNOWN', 'creation_time': datetime.datetime(2022, 4, 28, 16, 50, 47, 943000, tzinfo=tzlocal()), 'head_node': {'instance_id': 'i-abcdef01234567890', 'instance_type': 't2.micro', 'launch_time': datetime.datetime(2022, 4, 28, 16, 53, 48, tzinfo=tzlocal()), 'private_ip_address': '172.31.17.132', 'public_ip_address': '34.201.100.37', 'state': 'running'}, 'last_updated_time': datetime.datetime(2022, 4, 28, 16, 50, 47, 943000, tzinfo=tzlocal()), 'region': 'us-east-1', 'tags': [{'key': 'parallelcluster:version', 'value': '3.1.3'}], 'version': '3.1.3'} . . . . . . working . . . {'cloud_formation_stack_status': 'DELETE_IN_PROGRESS', 'cloudformation_stack_arn': 'arn:aws:cloudformation:us-east-1:123456789012:stack/test-api-cluster/abcd1234-ef56-gh78-ij90-1234abcd5678', 'cluster_configuration': {'url': 'https://parallelcluster-021345abcdef6789-v1-do-not-delete...}, 'cluster_name': 'test-api-cluster', 'cluster_status': 'DELETE_IN_PROGRESS', 'compute_fleet_status': 'UNKNOWN', 'creation_time': datetime.datetime(2022, 4, 28, 16, 50, 47, 943000, tzinfo=tzlocal()), 'last_updated_time': datetime.datetime(2022, 4, 28, 16, 50, 47, 943000, tzinfo=tzlocal()), 'region': 'us-east-1', 'tags': [{'key': 'parallelcluster:version', 'value': '3.1.3'}], 'version': '3.1.3'} . . . working . . . DELETE COMPLETE or Exception when calling describe_cluster: (404) Reason: Not Found . . . HTTP response body: {"message":"Cluster 'test-api-cluster' does not exist or belongs to an incompatible ParallelCluster major version."}
  3. 完成測試後,請取消設定環境變數。

    $ unset AWS_SESSION_TOKEN unset AWS_SECRET_ACCESS_KEY unset AWS_ACCESS_KEY_ID

您可以使用 AWS Management Console 或 AWS CLI 來刪除您的 API。

  1. 從 AWS CloudFormation 主控台中選擇API堆疊,然後選擇刪除

  2. 如果使用 ,請執行下列命令 AWS CLI。

    使用 AWS CloudFormation。

    $ aws cloudformation delete-stack --stack-name ${API_STACK_NAME}