How to install Gravitino

Install Gravitino from scratch

note

Gravitino supports running on Java 8, 11, and 17. Make sure you have Java installed and JAVA_HOME configured correctly. To confirm the Java version, run the ${JAVA_HOME}/bin/java -version command.

Get the Gravitino binary distribution package

Before installing Gravitino, make sure you have the Gravitino binary distribution package. You can download the latest Gravitino binary distribution package from GitHub, or you can build it yourself by following the instructions in How to Build Gravitino.

• If you build Gravitino yourself using the ./gradlew compileDistribution command, you can find the Gravitino binary distribution package in the distribution/package directory.

• If you build Gravitino yourself using the ./gradlew assembleDistribution command, you can get the compressed Gravitino binary distribution package with the name gravitino-<version>-bin.tar.gz in the distribution directory with sha256 checksum file gravitino-<version>-bin.tar.gz.sha256.

The Gravitino binary distribution package contains the following files:

|── ...
└── distribution/package
    |── bin/gravitino.sh            # Gravitino server Launching scripts.
    |── catalogs
    |   └── hive/                   # Hive catalog dependencies and configurations.
    |   └── lakehouse-iceberg/      # Apache Iceberg catalog dependencies and configurations.
    |   └── jdbc-mysql/             # JDBC MySQL catalog dependencies and configurations.
    |   └── jdbc-postgresql/        # JDBC PostgreSQL catalog dependencies and configurations.
    |── conf/                       # All configurations for Gravitino.
    |   ├── gravitino.conf          # Gravitino server configuration.
    |   ├── gravitino-env.sh        # Environment variables, etc., JAVA_HOME, GRAVITINO_HOME, and more.
    |   └── log4j2.properties       # log4j configuration for the Gravitino server.
    |── libs/                       # Gravitino server dependencies libraries.
    |── logs/                       # Gravitino server logs. Automatically created after the Gravitino server starts.
    |── data/                       # Default directory for the Gravitino server to store data.
    └── scripts/                    # Extra scripts for Gravitino.

Initialize the RDBMS (Optional)

If you want to use the relational backend storage, you need to initialize the RDBMS firstly. For the details on how to initialize the RDBMS, please check How to use relational backend storage.

Configure the Gravitino server

The Gravitino server configuration file is conf/gravitino.conf. You can configure the Gravitino server by modifying this file. Basic configurations are already added to this file. All the configurations are listed in Gravitino Server Configurations.

Configure the Gravitino server log

The Gravitino server log configuration file is conf/log4j2.properties. Gravitino uses Log4j2 as the Logging system. You can Log4j2 to do the log configuration.

Configure the Gravitino server environment

The Gravitino server environment configuration file is conf/gravitino-env.sh. Gravitino exposes several environment variables. You can modify them in this file.

Configure Gravitino catalogs

Gravitino supports multiple catalogs. You can configure the catalog-level configurations by modifying the related configuration file in the catalogs/<catalog-provider>/conf directory. The configurations you set here apply to all the catalogs of the same type you create.

For example, if you want to configure the Hive catalog, you can modify the file catalogs/hive/conf/hive.conf. The detailed configurations are listed in the specific catalog documentation.

note

Gravitino takes the catalog configurations in the following order:

1. Catalog properties specified in catalog creation API or REST API.
2. Catalog configurations specified in the catalog configuration file.

The catalog properties can override the catalog configurations specified in the configuration file.

Gravitino supports passing in catalog-specific configurations if you add gravitino.bypass.. For example, if you want to pass in the HMS-specific configuration hive.metastore.client.capability.check to the underlying Hive client in the Hive catalog, add the gravitino.bypass. prefix to it.

Also, Gravitino supports loading catalog specific configurations from external files. For example, you can put your own hive-site.xml file in the catalogs/hive/conf directory, and Gravitino loads it automatically.

Start Gravitino server

After configuring the Gravitino server, start the Gravitino server by running:

./bin/gravitino.sh start

You can access the Gravitino Web UI by typing http://localhost:8090 in your browser. or you can run

curl -v -X GET -H "Accept: application/vnd.gravitino.v1+json" -H "Content-Type: application/json" http://localhost:8090/api/version

to make sure Gravitino is running.

info

If you need to debug the Gravitino server, enable the GRAVITINO_DEBUG_OPTS environment variable in the conf/gravitino-env.sh file. Then create a Remote JVM Debug configuration in IntelliJ IDEA and debug gravitino.server.main.

Install Gravitino using Docker

Get the Gravitino Docker image

Gravitino publishes the Docker image to Docker Hub. Run the Gravitino Docker image by running:

docker run -d -i -p 8090:8090 datastrato/gravitino:<version>

Access the Gravitino Web UI by typing http://localhost:8090 in your browser, or you can run

curl -v -X GET -H "Accept: application/vnd.gravitino.v1+json" -H "Content-Type: application/json" http://localhost:8090/api/version

to make sure Gravitino is running.

Install Gravitino using Docker compose

The published Gravitino Docker image only contains the Gravitino server with basic configurations. If you want to experience the whole Gravitino system with other components, use the Docker compose file.

For the details, review the Gravitino playground repository and playground example.