model_tfserving_rest.yaml

apiVersion: machinelearning.seldon.io/v1
kind: SeldonDeployment
metadata:
  name: rest-tfserving
spec:
  name: resttfserving
  predictors:
    - protocol: tensorflow
      transport: rest
      componentSpecs:
        - spec:
            containers:
              - args:
                  - --port=8500
                  - --rest_api_port=8501
                  - --model_name=recommender
                  - --model_base_path=s3://fdp-killrweather-data/kubeflow/recommeder/model-updated/recommender
                image: tensorflow/serving:1.14.0
                name: recommender
                ports:
                  - containerPort: 8501
                    name: http
                env:
                  - name: "AWS_REGION"
                    value: "eu-west-1"
                  - name: "S3_REGION"
                    value: "eu-west-1"
                  - name: "S3_ENDPOINT"
                    value: "s3.eu-west-1.amazonaws.com"
                  - name: "S3_USE_HTTPS"
                    value: "1"
                  - name: "S3_VERIFY_SSL"
                    value: "1"
                  - name: "AWS_ACCESS_KEY_ID"
                    valueFrom: { secretKeyRef: { name: "s3-credentials", key: "accessKey" } }
                  - name: "AWS_SECRET_ACCESS_KEY"
                    valueFrom: { secretKeyRef: { name: "s3-credentials", key: "secretKey" } }
                volumes:
                  - name: secret-volume
                    secret:
                      secretName: s3-credentials
      graph:
        name: recommender
        type: MODEL
        endpoint:
          service_port: 8501
      name: model
      replicas: 1