Deployment

Prerequisites

Access to a running Airy Core installation Get started
A dialogflow agent up and running in google cloud see documentation

In order to deploy dialogflow-connector you can apply the following Kubernetes Deployment descriptor to your Airy Core Kubernetes cluster. You can also adapt the following configuration to work with your favorite deployment tool like terraform.

Get google cloud credentials see documentation
Create configmap using the snippet called dialogflow-config.yaml
Deploy the connector by using the snippet called dialogflow-deployment.yaml

There are some variables that need to be taken into account:

Dialogflow project ID given by the Google Cloud Console: see documentation
The confidence level tells the connector the amount of confidence it should have in order to suggest or give a reply. This value is a percentage so it should be between 0 and 1.
The processor waiting time and check periods (in milliseconds) adjust the enrichment of the message before it is being processed by Dialogflow.
- The processor waiting time represents the amount of time before the message can be enriched by other connectors in the airy cluster before sending it to Dialogflow (for example: 2500).
- The processor check time represents the amount of time before the message can be enriched by other connectors before forwarding it to Dialogflow (for example: 5000).
Default language: if the language in which the message is written cannot be determined, it will default to this value (for example: en for English).

dialogflow-config.yaml

apiVersion: v1
data:
  dialogflow-credentials.json: <the content of google json credentials>
kind: ConfigMap
metadata:
  name: dialogflow-config

kubectl -n staging create -f dialogflow-configmap.yaml

dialogflow-deployment.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: enterprise-dialogflow-connector
  namespace: staging
  labels:
    app: dialogflow-connector
    type: api
    core.airy.co/managed: 'false'
    core.airy.co/mandatory: 'false'
    core.airy.co/component: enterprise-dialogflow-connector
spec:
  replicas: 1
  selector:
    matchLabels:
      app: dialogflow-connector
  strategy:
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 1
    type: RollingUpdate
  template:
    metadata:
      labels:
        app: dialogflow-connector
    spec:
      containers:
        - name: app
          image: ghcr.io/airyhq/enterprise/dialogflow-connector:release

          imagePullPolicy: Always
          envFrom:
            - configMapRef:
                name: security
            - configMapRef:
                name: kafka-config
          env:
            - name: DIALOGFLOW_PROJECT_ID
              value: <your dialogflow project id>
            - name: GOOGLE_APPLICATION_CREDENTIALS
              value: '/app/dialogflow-credentials.json'
            - name: CONNECTOR_SUGGESTION_CONFIDENCE_LEVEL
              value: <suggestion confidence level a value between 0 and 1>
            - name: CONNECTOR_REPLY_CONFIDENCE_LEVEL
              value: <reply confidence level a value between 0 and 1>
            - name: CONNECTOR_STORE_MESSAGES_PROCESSOR_MAX_WAIT_MILLIS
              value: 5000
            - name: CONNECTOR_STORE_MESSAGES_PROCESSOR_CHECK_PERIOD_MILLIS
              value: 2500
            - name: CONNECTOR_DEFAULT_LANGUAGE
              value: 'en'
            - name: SERVICE_NAME
              value: dialogflow-connector
            - name: POD_NAMESPACE
              valueFrom:
                fieldRef:
                  apiVersion: v1
                  fieldPath: metadata.namespace
            - name: REQUESTED_CPU
              valueFrom:
                resourceFieldRef:
                  containerName: app
                  resource: requests.cpu
            - name: LIMIT_CPU
              valueFrom:
                resourceFieldRef:
                  containerName: app
                  resource: limits.cpu
            - name: LIMIT_MEMORY
              valueFrom:
                resourceFieldRef:
                  containerName: app
                  resource: limits.memory
          volumeMounts:
            - name: dialogflow-config
              mountPath: /app/dialogflow-credentials.json
              subPath: dialogflow-credentials.json
          livenessProbe:
            httpGet:
              path: /actuator/health
              port: 8080
              httpHeaders:
                - name: Health-Check
                  value: health-check
            initialDelaySeconds: 60
            periodSeconds: 10
            failureThreshold: 3
      volumes:
        - name: dialogflow-config
          configMap:
            name: dialogflow-config
---
apiVersion: v1
kind: Service
metadata:
  name: enterprise-dialogflow-connector
  namespace: default
  labels:
    core.airy.co/prometheus: spring
spec:
  ports:
    - name: web
      port: 80
      targetPort: 8080
      protocol: TCP
  type: NodePort
  selector:
    app: enterprise-dialogflow-connector

kubectl apply -f dialogflow-deployment.yaml