使用 Elastic Fabric Adapter 在 Amazon EKS 上运行机器学习训练

将以下内容复制到名为 efa-cluster.yaml 的文件中。将 example values 替换为您自己的值。您可以将 p5.48xlarge 替换为不同的实例，但是如果您进行替换，请确保 availabilityZones 的值为第 1 步中针对实例类型返回的可用区。

apiVersion: eksctl.io/v1alpha5
kind: ClusterConfig

metadata:
  name: my-efa-cluster
  region: region-code
  version: "1.XX"

iam:
  withOIDC: true

availabilityZones: ["us-west-2a", "us-west-2c"]

managedNodeGroups:
  - name: my-efa-ng
    instanceType: p5.48xlarge
    minSize: 1
    desiredCapacity: 2
    maxSize: 3
    availabilityZones: ["us-west-2a"]
    volumeSize: 300
    privateNetworking: true
    efaEnabled: true

在现有集群中创建托管节点组。

eksctl create nodegroup -f efa-cluster.yaml

如果您没有现有集群，可以运行以下命令来创建集群和节点组。

eksctl create cluster -f efa-cluster.yaml

创建 MPIJob 清单：

将以下内容复制到名为 nccl-tests.yaml 的文件中：

apiVersion: kubeflow.org/v2beta1
kind: MPIJob
metadata:
  name: nccl-tests
spec:
  runPolicy:
    cleanPodPolicy: Running
    backoffLimit: 20
  slotsPerWorker: 8
  mpiReplicaSpecs:
    Launcher:
      replicas: 1
      template:
         spec:
          restartPolicy: OnFailure
          containers:
          - image: public.ecr.aws/hpc-cloud/nccl-tests:latest
            imagePullPolicy: IfNotPresent
            name: test-nccl-launcher
            env:
             - name: PATH
               value: $PATH:/opt/amazon/efa/bin:/usr/bin
             - name: LD_LIBRARY_PATH
               value: /opt/amazon/openmpi/lib:/opt/nccl/build/lib:/opt/amazon/efa/lib:/opt/aws-ofi-nccl/install/lib:/usr/local/nvidia/lib:$LD_LIBRARY_PATH
             - name: NCCL_DEBUG
               value: INFO
             - name: NCCL_BUFFSIZE
               value: '8388608'
             - name: NCCL_P2P_NET_CHUNKSIZE
               value: '524288'
             - name: NCCL_TUNER_PLUGIN
               value: /opt/aws-ofi-nccl/install/lib/libnccl-ofi-tuner.so
            command:
            - /opt/amazon/openmpi/bin/mpirun
            - --allow-run-as-root
            - --tag-output
            - -np
            - "16"
            - -N
            - "8"
            - --bind-to
            - none
            - -x
            - PATH
            - -x
            - LD_LIBRARY_PATH
            - -x
            - NCCL_DEBUG=INFO
            - -x
            - NCCL_BUFFSIZE
            - -x
            - NCCL_P2P_NET_CHUNKSIZE
            - -x
            - NCCL_TUNER_PLUGIN
            - --mca
            - pml
            - ^cm,ucx
            - --mca
            - btl
            - tcp,self
            - --mca
            - btl_tcp_if_exclude
            - lo,docker0,veth_def_agent
            - /opt/nccl-tests/build/all_reduce_perf
            - -b
            - "8"
            - -e
            - "16G"
            - -f
            - "2"
            - -g
            - "1"
            - -c
            - "1"
            - -n
            - "100"
    Worker:
      replicas: 2
      template:
        spec:
          nodeSelector:
            node.kubernetes.io/instance-type: "p5.48xlarge"
          containers:
          - image: public.ecr.aws/hpc-cloud/nccl-tests:latest
            imagePullPolicy: IfNotPresent
            name: nccl-tests-worker
            volumeMounts:
            - name: shmem
              mountPath: /dev/shm
            resources:
              limits:
                nvidia.com/gpu: 8
                hugepages-2Mi: 5120Mi
                vpc.amazonaws.com/efa: 32
                memory: 32000Mi
              requests:
                nvidia.com/gpu: 8
                hugepages-2Mi: 5120Mi
                vpc.amazonaws.com/efa: 32
                memory: 32000Mi
          volumes:
          - name: shmem
            hostPath:
              path: /dev/shm

应用 NCCL-tests MPIJob：

通过应用清单提交 MPIJob。这将创建两个 p5.48xlarge Amazon EC2 实例。

kubectl apply -f nccl-tests.yaml

示例输出如下。

mpijob.kubeflow.org/nccl-tests created

验证作业是否启动了容器组（pod）：

查看正在运行的容器组（pod）。

kubectl get pods

示例输出如下。

NAME                             READY   STATUS     RESTARTS   AGE
nccl-tests-launcher-nbql9    0/1     Init:0/1   0          2m49s
nccl-tests-worker-0          1/1     Running    0          2m49s
nccl-tests-worker-1          1/1     Running    0          2m49s

MPI Operator 创建一个启动程序容器组（pod）和 2 个工作线程容器组（pod）（每个节点上一个）。

使用日志来验证作业是否成功运行：

查看 nccl-tests-launcher 容器组（pod）的日志。将 nbql9 替换为输出中的值。

kubectl logs -f nccl-tests-launcher-nbql9