Hive on MR3

Using executable scripts:

Using Helm:

  • On Minikube shows how to use Helm to operate Hive on MR3 on Minikube.
  • On a Multi-node Cluster shows how to use Helm to operate Hive on MR3 on Kubernetes with multiple nodes.

Spark on MR3

Using executable scripts:

Hive on MR3, Spark on MR3, Ranger, Grafana, MR3-UI, Superset

The recommended way to run Hive/Spark on MR3 is to use TypeScript code.

Using TypeScript code:

A web-based interface MR3 Cloud is also available.

mr3-typescript-components.png

  • The user can connect to a public HiveServer2 (via JDBC/ODBC) which is exposed to the outside of the Kubernetes cluster. Multiple HiveServer2 instances can be created.
  • The user can connect to Apache HTTP Server which serves as a gateway to Superset, Ranger, Grafana, MR3-UI, and Spark UI.
  • Superset connects to an internal HiveServer2 which is not exposed to the outside of the Kubernetes cluster. Superset with an internal HiveServer2 is optional.
  • All HiveServer2 instances share a common MR3 DAGAppMaster and its ContainerWorkers.
  • Multiple Spark drivers sharing a common MR3 DAGAppMaster and its ContainerWorkers can run inside the Kubernetes cluster.
  • HiveServer2 and Spark drivers are automatically configured to share Metastore.
For asking questions, please visit MR3 Google Group.