node-template.yaml

apiVersion: machinelearning.seldon.io/v1
kind: SeldonDeployment
metadata:
  name: router
spec:
  protocol: v2
  name: router
  annotations:
    proxy.istio.io/config: |
      terminationDrainDuration: {{ distrpution_time }}s
  predictors:
  - name: router
    annotations:
      seldon.io/no-engine: "true"
    componentSpecs:
    - spec:
        terminationGracePeriodSeconds: {{ distrpution_time }}
        # nodeSelector:
        #   router-node: "true"
        containers:
        - image: sdghafouri/router:router
          name: router
          imagePullPolicy: Always
          resources:
            requests:
              cpu: '{{ cpu_request }}'
              memory: '{{ memory_request }}'
            limits:
              cpu: '{{ cpu_limit }}'
              memory: '{{ memory_limit }}'
          env:
            - name: MODEL_LISTS
              value: '{{ model_lists }}'
            # router should be multi-process to sustain througput
            - name: MLSERVER_PARALLEL_WORKERS
              value: "8"
            - name: DROP_LIMIT
              value: '{{drop_limit}}'
            - name: LOGS_ENABLED
              value: '{{ logs_enabled }}'
          readinessProbe:
            failureThreshold: 3
            initialDelaySeconds: 0
            periodSeconds: 1
            successThreshold: 1
            tcpSocket:
              port: 9000
            timeoutSeconds: 1
          lifecycle:
            preStop:
              exec:
                command:
                - /bin/sh
                - -c
                - /bin/sleep {{ distrpution_time }}
      replicas: {{ replicas }}
    graph:
      name: router
      type: MODEL
      children: []
    labels:
      sidecar.istio.io/inject: "true"