Gravitino web UI

This document primarily outlines how users can manage metadata within Gravitino using the web UI, the graphical interface is accessible through a web browser as an alternative to writing code or using the REST interface.

Currently, you can integrate OAuth settings to view, add, modify, and delete metalakes, create catalogs, and view catalogs, schemas, and tables, among other functions.

Build and deploy the Gravitino Web UI and open it in a browser at http://<gravitino-host>:<gravitino-port>, by default is http://localhost:8090.

Initial page

The web UI homepage displayed in Gravitino depends on the configuration parameter for OAuth mode, see the details in Security.

Set parameter for gravitino.authenticator, simple or oauth. Simple mode is the default authentication option.

tip

After changing the configuration, make sure to restart the Gravitino server.

<path-to-gravitino>/bin/gravitino.sh restart

Simple mode

gravitino.authenticator = simple

Set the configuration parameter gravitino.authenticator to simple, and the web UI displays the homepage (Metalakes).

At the top-right, the UI displays the current Gravitino version.

The main content displays the existing metalake list.

Oauth mode

gravitino.authenticator = oauth

Set the configuration parameter gravitino.authenticator to oauth, and the web UI displays the login page.

caution

If both OAuth and HTTPS are set, due to the different security permission rules of various browsers, to avoid cross-domain errors, it is recommended to use the Chrome browser for access and operation.

Such as Safari need to enable the developer menu, and select Disable Cross-Origin Restrictions from the develop menu.

1. Enter the values corresponding to your specific configuration. For detailed instructions, please refer to Security.

2. Clicking on the LOGIN button takes you to the homepage.

At the top-right, there is an icon button that takes you to the login page when clicked.

Manage metadata

All the manage actions are performed by using the REST API

Metalake

Create metalake

On the homepage, clicking on the CREATE METALAKE button displays a dialog to create a metalake.

Creating a metalake needs these fields:

1. Name(required): the name of the metalake.
2. Comment(optional): the comment of the metalake.
3. Properties(optional): Click on the ADD PROPERTY button to add custom properties.

There are 3 actions you can perform on a metalake.

Show metalake details

Clicking on the action icon in the table cell.

You can see the detailed information of this metalake in the drawer component on the right.

Edit metalake

Clicking on the action icon in the table cell.

Displays the dialog for modifying fields of the selected metalake.

Delete metalake

Clicking on the action icon in the table cell.

Displays a confirmation dialog, clicking on the SUBMIT button deletes this metalake.

Catalog

Clicking on a metalake name in the table views catalogs in a metalake.

If this is the first time, it shows no data until after creating a catalog.

Clicking on the left arrow icon button takes you to the metalake page.

Clicking on the Tab - DETAILS views the details of the catalog on the metalake catalogs page.

On the left side of the page is a tree list, and the icons of the catalog correspond to their type and provider.

• Catalog
• Schema
• Table

Hover your mouse over the corresponding icon to the data changes to a reload icon . Click on this icon to reload the currently selected data.

Create catalog

Clicking on the CREATE CATALOG button displays the dialog to create a catalog.

Creating a catalog requires these fields:

1. Catalog name(required): the name of the catalog
2. Type(required): relational/fileset/messaging, the default value is relational
3. Provider(required):
   1. Type relational - hive/iceberg/mysql/postgresql/doris
   2. Type fileset - hadoop
   3. Type messaging - kafka
4. Comment(optional): the comment of this catalog
5. Properties(each provider must fill in the required property fields specifically)

Providers

Required properties in various providers

1. Type relational

Hive

Follow the Apache Hive catalog document.

Key	Description
metastore.uris	The Hive metastore URIs e.g. thrift://127.0.0.1:9083

Iceberg

Follow the Lakehouse Iceberg catalog document.

the parameter catalog-backend provides two values: hive, and jdbc.

Key	Description
catalog-backend	hive, or jdbc

• hive

Key	Description
uri	Iceberg catalog URI config
warehouse	Iceberg catalog warehouse config

• jdbc

Key	Description
uri	Iceberg catalog URI config
warehouse	Iceberg catalog warehouse config
jdbc-driver	"com.mysql.jdbc.Driver" or "com.mysql.cj.jdbc.Driver" for MySQL, "org.postgresql.Driver" for PostgreSQL
jdbc-user	jdbc username
jdbc-password	jdbc password

MySQL

Follow the JDBC MySQL catalog document.

Key	Description
jdbc-driver	JDBC URL for connecting to the database. e.g. com.mysql.jdbc.Driver or com.mysql.cj.jdbc.Driver
jdbc-url	e.g. jdbc:mysql://localhost:3306
jdbc-user	The JDBC user name
jdbc-password	The JDBC password

PostgreSQL

Follow the JDBC PostgreSQL catalog document.

Key	Description
jdbc-driver	e.g. org.postgresql.Driver
jdbc-url	e.g. jdbc:postgresql://localhost:5432/your_database
jdbc-user	The JDBC user name
jdbc-password	The JDBC password
jdbc-database	e.g. pg_database

Doris

Follow the JDBC Doris catalog document.

Key	Description
jdbc-driver	JDBC URL for connecting to the database. e.g. com.mysql.jdbc.Driver
jdbc-url	e.g. jdbc:mysql://localhost:9030
jdbc-user	The JDBC user name
jdbc-password	The JDBC password

tip

Due to the current limitation of the web interface, which only allows for viewing, the functionality to create or modify schema, tables, or filesets is not available. Please refer to the documentation to use the REST API for these operations.

2. Type fileset

Hadoop

Follow the Hadoop catalog document.

3. Type messaging

Kafka

Follow the Kafka catalog document.

Key	Description
bootstrap.servers	The Kafka broker(s) to connect to, allowing for multiple brokers by comma-separating them

After verifying the values of these fields, clicking on the CREATE button creates a catalog.

Show catalog details

Clicking on the action icon in the table cell.

You can see the detailed information of this catalog in the drawer component on the right.

Edit catalog

Clicking on the action icon in the table cell.

Displays the dialog for modifying fields of the selected catalog.

Only the name, comment, and custom fields in properties can be modified, other fields such as type, provider, and default fields in properties cannot be modified.

The fields that are not allowed to be modified cannot be selected and modified in the web UI.

Delete catalog

Clicking on the action icon in the table cell.

Displays a confirmation dialog, clicking on the SUBMIT button deletes this catalog.

Schema

Table

Fileset

Topic

Feature capabilities

Page	Capabilities
Metalake	View ✔ / Create ✔ / Edit ✔ / Delete ✔
Catalog	View ✔ / Create ✔ / Edit ✔ / Delete ✔
Schema	View ✔ / Create ✘ / Edit ✘ / Delete ✘
Table	View ✔ / Create ✘ / Edit ✘ / Delete ✘
Fileset	View ✔ / Create ✘ / Edit ✘ / Delete ✘
Topic	View ✔ / Create ✘ / Edit ✘ / Delete ✘

E2E test

End-to-end testing for web frontends is conducted using the Selenium testing framework, which is Java-based.

Test cases can be found in the project directory: integration-test/src/test/java/com/datastrato/gravitino/integration/test/web/ui, where the pages directory is designated for storing definitions of frontend elements, among others. The root directory contains the actual steps for the test cases.

tip

While writing test cases, running them in a local environment may not pose any issues.

However, due to the limited performance capabilities of GitHub Actions, scenarios involving delayed DOM loading—such as the time taken for a popup animation to open—can result in test failures.

To circumvent this issue, it is necessary to manually insert a delay operation, for instance, by adding such as Thread.sleep(sleepTimeMillis).

This ensures that the test waits for the completion of the delay animation before proceeding with the next operation, thereby avoiding the problem.

It is advisable to utilize the waits methods inherent to Selenium as a substitute for Thread.sleep().