• [+] expand all

Vespa quick start using Kubernetes

This guide describes how to install and run Vespa on a single machine using Kubernetes, K8s. See Getting Started for troubleshooting, next steps and other guides. Also see Vespa example on GKE.

  1. Validate environment:

    Refer to Docker memory for details and troubleshooting: 

    $ docker info | grep "Total Memory"
  2. Start Kubernetes cluster with minikube: 
    $ minikube start --driver docker --memory 4096
  3. Clone the Vespa sample apps: 
    $ git clone --depth 1 https://github.com/vespa-engine/sample-apps.git
$ export VESPA_SAMPLE_APPS=`pwd`/sample-apps
  4. Create Kubernetes configuration files: 
    $ cat << EOF > service.yml
apiVersion: v1
kind: Service
metadata:
  name: vespa
  labels:
    app: vespa
spec:
  selector:
    app: vespa
  type: NodePort
  ports:
  - name: container
    port: 8080
    targetPort: 8080
    protocol: TCP
  - name: config
    port: 19071
    targetPort: 19071
    protocol: TCP
EOF

    $ cat << EOF > statefulset.yml
apiVersion: apps/v1
kind: StatefulSet
metadata:
  name: vespa
  labels:
    app: vespa
spec:
  replicas: 1
  serviceName: vespa
  selector:
    matchLabels:
      app: vespa
  template:
    metadata:
      labels:
        app: vespa
    spec:
      containers:
      - name: vespa
        image: vespaengine/vespa
        imagePullPolicy: Always
        env:
        - name: VESPA_CONFIGSERVERS
          value: vespa-0.vespa.default.svc.cluster.local
        securityContext:
          privileged: true
          runAsUser: 0
        ports:
        - containerPort: 8080
          protocol: TCP
        readinessProbe:
          httpGet:
            path: /state/v1/health
            port: 19071
            scheme: HTTP
EOF
  5. Start the service: 
    $ kubectl apply -f service.yml -f statefulset.yml
  6. Wait for the service to enter a running state: 
    $ kubectl get pods --watch
    Wait for STATUS Running: 
        NAME      READY   STATUS              RESTARTS   AGE
    vespa-0   0/1     ContainerCreating   0          8s
    vespa-0   0/1     Running             0          2m4s
  7. Start port forwarding from localhost to pod: 
    $ kubectl port-forward vespa-0 19071 8080 &
  8. Wait for configserver start - wait for 200 OK: 
    $ curl -s --head http://localhost:19071/state/v1/health
  9. Zip the application package: 
    $ (cd ${VESPA_SAMPLE_APPS}/album-recommendation && \
  zip -r ../../application.zip .)
  10. Deploy and activate the application package: 
    $ curl --header Content-Type:application/zip --data-binary @application.zip \
  localhost:19071/application/v2/tenant/default/prepareandactivate
  11. Ensure the application is active - wait for 200 OK:

    This normally takes a minute or so:

    $ curl -s --head http://localhost:8080/state/v1/health
  12. Feed documents: 
    $ curl -s -H "Content-Type:application/json" \
  --data-binary @${VESPA_SAMPLE_APPS}/album-recommendation/ext/A-Head-Full-of-Dreams.json \
  http://localhost:8080/document/v1/mynamespace/music/docid/a-head-full-of-dreams
$ curl -s -H "Content-Type:application/json" \
  --data-binary @${VESPA_SAMPLE_APPS}/album-recommendation/ext/Love-Is-Here-To-Stay.json \
  http://localhost:8080/document/v1/mynamespace/music/docid/love-is-here-to-stay
$ curl -s -H "Content-Type:application/json" \
  --data-binary @${VESPA_SAMPLE_APPS}/album-recommendation/ext/Hardwired...To-Self-Destruct.json \
  http://localhost:8080/document/v1/mynamespace/music/docid/hardwired-to-self-destruct

  13. Make a query:

    $ curl -X POST -H "Content-Type: application/json" --data '
  {
    "yql": "select * from sources * where true",
    "input.query(user_profile)": "{{cat:pop}:0.8,{cat:rock}:0.2,{cat:jazz}:0.1}"
  }' \
  http://localhost:8080/search/

    Read more in the Query API.

  14. Run a document get request: 
    $ curl -s http://localhost:8080/document/v1/mynamespace/music/docid/love-is-here-to-stay

  15. Clean up:

    Stop the running container:

    $ kubectl delete service,statefulsets vespa

    Stop port forwarding:

    $ killall kubectl

    Stop minikube:

    $ minikube stop