volcano.sh/volcano@v1.9.0/example/MindSpore-example/mindspore_gpu/mindspore-gpu.yaml (about)

     1  apiVersion: batch.volcano.sh/v1alpha1
     2  kind: Job
     3  metadata:
     4    name: mindspore-gpu
     5  spec:
     6    minAvailable: 3
     7    schedulerName: volcano
     8    plugins:
     9      ssh: []
    10      svc: []
    11    tasks:
    12      - replicas: 1
    13        name: mpimaster
    14        template:
    15          spec:
    16            containers:
    17              - command:
    18                  - /bin/bash
    19                  - -c
    20                  - |
    21                    mkdir -p /var/run/sshd; /usr/sbin/sshd;
    22                    MPI_HOST=`cat /etc/volcano/mpiworker.host | tr "\n" ","`;
    23                    sleep 10;
    24                    mpiexec --allow-run-as-root --host ${MPI_HOST} -np 2 --prefix /usr/local/openmpi-3.1.5 python /tmp/gpu-test.py;
    25                    sleep 3600;
    26                image: lyd911/mindspore-gpu-example:0.2.0
    27                name: mpimaster
    28                ports:
    29                  - containerPort: 22
    30                    name: mpijob-port
    31                workingDir: /home
    32            restartPolicy: OnFailure
    33      - replicas: 2
    34        name: mpiworker
    35        template:
    36          spec:
    37            containers:
    38              - command:
    39                  - /bin/bash
    40                  - -c
    41                  - |
    42                    mkdir -p /var/run/sshd; /usr/sbin/sshd -D; 
    43                image: lyd911/mindspore-gpu-example:0.2.0
    44                name: mpiworker
    45                resources:
    46                  limits:
    47                    nvidia.com/gpu: "1"
    48                ports:
    49                  - containerPort: 22
    50                    name: mpijob-port
    51                workingDir: /home
    52            restartPolicy: OnFailure
    53   
    54  ---