MR3 Cloud is a web-based interface for facilitating the execution of Hive on MR3 on public clouds and Kubernetes. The user can set parameters in an interactive way to generate resources for creating a Kubernetes cluster or YAML files for creating various Kubernetes objects. After downloading such resources or YAML files, the user can execute a command specific to the public cloud (e.g., eksctl for Amazon EKS) or execute kubectl to start Hive on MR3.

MR3 Cloud generates YAML files for starting the following components all at once:

• Hive on MR3
• Apache Ranger
• Superset
• Grafana
• MR3-UI

The user can access Hive on MR3 by connecting directly to HiveServer2 (using either JDBC or ODBC) and all the other components via an Apache server (using either HTTP or HTTPS). HiveServer2 runs behind a LoadBalancer, while the Apache server runs behind another LoadBalancer.

Internally we run a second instance of HiveServer2 which is not exposed to the outside of the Kubernetes cluster and serves requests only from Superset. We also run a Timeline server to collect history logs from MR3 DAGAppMaster, and a Prometheus server to collect metrics from MR3 DAGAppMaster.

For asking questions, please visit MR3 Google Group.

Sign In

Open MR3 Cloud at https://cloud.datamonad.com. In order to use the service of MR3 Cloud, the user should login with OAuth. If you need an OAuth provider not listed in the menu, please contact us.

Load

The Load page shows a preset configuration.

• Choose Amazon EKS to generate YAML files for Amazon EKS.
• Choose Kubernetes - General to generate YAML files for a Kubernetes cluster.

The user can also load a configuration that was previously saved.

Setting parameters

Each page has one or more sections, some of which may have been disabled depending on the current configuration. If all the input fields in a section are okay, a Ready message is displayed.

If some input fields are incomplete or given wrong parameters, error messages are displayed.

Every input field has an info button which opens a popup window describing the input field.

Executing Hive on MR3

• For executing Hive on MR3 on Amazon EKS, see On Amazon EKS.
• For executing Hive on MR3 on Kubernetes, see On Kubernetes.
• For using Kerberos for authentication, see Using Kerberos.
• For using SSL encryption, see Using SSL.

Ranger

The password for the user admin is specified in the field Administrator Password in the section Ranger. The user should manually create a Ranger service with the name specified in the field Ranger Service for Hive in the same section.

Then the Ranger service should be properly configured. In particular, the policy.download.auth.users field in Config Properties panel should be set correctly to hive (which is the primary in the service principal if Kerberos is used for authentication).

Grafana

The password for the user admin is initialized to admin, and can be changed after the first login. The user can watch MR3 on the dashboard MR3 created by MR3 Cloud.

Superset

The password for the user admin is initialized to admin, and can be changed after the first login. The user should manually register a database source using a Hive URI pointing to the internal instance of HiveServer2.

• If Kerberos is not used, the Hive URI should be hive://hiveserver2-internal.hivemr3.svc.cluster.local:9852/default.
• If Kerberos is used, the HIVE URI should be hive://hiveserver2-internal.hivemr3.svc.cluster.local:9852/default?auth=KERBEROS&kerberos_service_name=hive.