Deploy Kyuubi On Kubernetes#


If you want to deploy Kyuubi on Kubernetes, you’d better get a sense of the following things.

Kyuubi Official Docker Image#

You can find the official docker image at Apache Kyuubi (Incubating) Docker Hub.

Build Kyuubi Docker Image#

You can build custom Docker images from the ${KYUUBI_HOME}/bin/ contained in the binary package.


  - Build and push image with tag "v1.4.0" to
    $0 -r -t v1.4.0 build
    $0 -r -t v1.4.0 push

  - Build and push with tag "v1.4.0" and Spark-3.2.1 as base image to
    $0 -r -t v1.4.0 -b BASE_IMAGE=repo/spark:3.2.1 build
    $0 -r -t v1.4.0 push

  - Build and push for multiple archs to
    $0 -r -t v1.4.0 -X build

  - Build with Spark placed "/path/spark"
    $0 -s /path/spark build
  - Build with Spark Image myrepo/spark:3.1.0
    $0 -S /opt/spark -b BASE_IMAGE=myrepo/spark:3.1.0 build

${KYUUBI_HOME}/bin/ use Kyuubi Version as default docker tag and always build ${repo}/kyuubi:${tag} image.

The script can also help build external Spark into a Kyuubi image that acts as a client for submitting tasks by -s ${SPARK_HOME}.

Of course, if you have an image that contains the Spark binary package, you don’t have to copy Spark locally. Make your Spark Image as BASE_IMAGE by using the -S ${SPARK_HOME_IN_DOCKER} and -b BASE_IMAGE=${SPARK_IMAGE} arguments.

You can use ${KYUUBI_HOME}/bin/ -h for more parameters.


Multiple YAML files are provided under ${KYUUBI_HOME}/docker/ to help you deploy Kyuubi.

You can deploy single-node Kyuubi through ${KYUUBI_HOME}/docker/kyuubi-pod.yaml or ${KYUUBI_HOME}/docker/kyuubi-deployment.yaml.

Also, you can use ${KYUUBI_HOME}/docker/kyuubi-service.yaml to deploy Kyuubi Service.


You can configure Kyuubi the old-fashioned way by placing kyuubi-default.conf inside the image. Kyuubi do not recommend using this way on Kubernetes.

Kyuubi provide ${KYUUBI_HOME}/docker/kyuubi-configmap.yaml to build Configmap for Kyuubi.

You can find out how to use it in the comments inside the above file.

If you want to know kyuubi engine on kubernetes configurations, you can refer to Deploy Kyuubi engines on Kubernetes


If you do not use Service or HostNetwork to get the IP address of the node where Kyuubi deployed. You should connect like:

kubectl exec -it kyuubi-example -- /bin/bash
${KYUUBI_HOME}/bin/beeline -u 'jdbc:hive2://localhost:10009'

Or you can submit tasks directly through local beeline:

${KYUUBI_HOME}/bin/beeline -u 'jdbc:hive2://${hostname}:${port}'

As using service nodePort, port means nodePort and hostname means any hostname of kubernetes node.

As using HostNetwork, port means kyuubi containerPort and hostname means hostname of node where Kyuubi deployed.


Kyuubi will provide other connection methods in the future, like Ingress, Load Balance.