Support AWS Glue as a Hive metastore
Extended Hive catalog initialization to use the standard metastore client-factory contract, including AWS Glue Data Catalog configuration from inline or Hive/Hadoop configuration sources.
apache/seatunnel · #11895
Cloud catalog feature
SeaTunnel's Hive sink can now use AWS Glue Data Catalog or another compatible client factory without a Thrift metastore URI.
Problem
Hive sink initialization required `metastore_uri`, excluding Glue and compatible metastores whose clients are created through `hive.metastore.client.factory.class` rather than Thrift endpoints.
Approach
Loads factory and catalog properties from supported configuration sources, gives the factory route precedence when configured, creates the metastore client through Hive's standard contract, and fails early when neither route or required runtime classes are available. Existing Thrift behavior remains intact.
Impact and scope
- Enables managed AWS Glue catalogs without requiring a separate Thrift metastore service.
- Supports other compatible Hive metastore factories through a standard extension contract.
- Preserves existing deployments and delegates authentication to the client implementation, including AWS's default credential chain.
Validation
- Covered inline and `hive-site.xml` factories, precedence, missing classes, missing configuration, and option exposure.
- All 98 Hive connector tests passed on Java 11, followed by the complete Apache Build.
- Both authored commits and the GitHub-verified merge credit Goutam Adwant; upstream approval closed GH-9856.