autoscale_job.yaml

---
apiVersion: elastic.iml.github.io/v1alpha1
kind: ElasticJob
metadata:
  name: deepctr-auto-scale
  namespace: dlrover
spec:
  distributionStrategy: ParameterServerStrategy
  resourceLimits:
    cpu: "15"
    memory: "30000Mi"
  replicaSpecs:
    ps:
      autoScale: True
      template:
        spec:
          restartPolicy: Never
          containers:
            - name: main
              # yamllint disable-line rule:line-length
              image: registry.cn-hangzhou.aliyuncs.com/intell-ai/dlrover:deeprec_criteo_v2
              imagePullPolicy: Always
              command:
                - /bin/bash
                - -c
                - "cd ./examples/tensorflow/criteo_deeprec \
                  && python -m dlrover.trainer.entry.local_entry \
                  --platform=Kubernetes --conf=train_conf.TrainConf \
                  --enable_auto_scaling=True"
              volumeMounts:
                - name: pvc-nas
                  mountPath: /nas
          volumes:
            - name: pvc-nas
              persistentVolumeClaim:
                claimName: pvc-nas
    worker:
      autoScale: True
      template:
        spec:
          restartPolicy: Never
          containers:
            - name: main
              # yamllint disable-line rule:line-length
              image: registry.cn-hangzhou.aliyuncs.com/intell-ai/dlrover:deeprec_criteo_v2
              imagePullPolicy: Always
              command:
                - /bin/bash
                - -c
                - "cd ./examples/tensorflow/criteo_deeprec \
                  && python -m dlrover.trainer.entry.local_entry \
                  --platform=Kubernetes --conf=train_conf.TrainConf \
                  --enable_auto_scaling=True"
              volumeMounts:
                - name: pvc-nas
                  mountPath: /nas
          volumes:
            - name: pvc-nas
              persistentVolumeClaim:
                claimName: pvc-nas
    evaluator:
      autoScale: True
      replicas: 1
      template:
        spec:
          restartPolicy: Never
          containers:
            - name: main
              # yamllint disable-line rule:line-length
              image: registry.cn-hangzhou.aliyuncs.com/intell-ai/dlrover:deeprec_criteo_v2
              imagePullPolicy: Always
              command:
                - /bin/bash
                - -c
                - "cd ./examples/tensorflow/criteo_deeprec \
                  && python -m dlrover.trainer.entry.local_entry \
                  --platform=Kubernetes --conf=train_conf.TrainConf \
                  --enable_auto_scaling=True"
              volumeMounts:
                - name: pvc-nas
                  mountPath: /nas
          volumes:
            - name: pvc-nas
              persistentVolumeClaim:
                claimName: pvc-nas
    dlrover-master:
      template:
        spec:
          restartPolicy: Never
          containers:
            - name: main
              imagePullPolicy: Always