DanielLeens commented on issue #12766: URL: https://github.com/apache/seatunnel/issues/12766#issuecomment-6191894357
Classification: A / Connector-V2 Hudi authentication. The new evidence changes this from an environment-only investigation into a confirmed connector gap. On the current Hudi path, `HudiSinkOptions` does not expose the claimed Kerberos settings, `HudiSinkConfig.of(...)` does not retain them, and `createHoodieJavaWriteClient(...)` reaches the Hudi client without calling the existing keytab helper. The same client-creation path is used by both the cached client manager and the write-client provider. Your official-distribution confirmation and the absence of a worker-level login make that gap material. There is no exact open Hudi/Kerberos PR. A focused contribution is welcome, but it must not simply call a JVM-global login helper next to `executeWrite`. Please first propose the backward-compatible option contract and cover all client-creation paths. The implementation must: 1. establish the intended identity before every Hudi client is initialized, including single-table and multi-table paths; 2. preserve non-Kerberos behavior and avoid logging principals, keytab paths, or credentials; 3. define the UGI/credential lifecycle and concurrent-job behavior explicitly, rather than silently relying on process-global identity; 4. add unit coverage for option parsing and authentication handoff without committing a keytab, plus an integration/reproduction boundary in the documentation. Please keep the sanitized worker evidence on this issue and link one focused PR here once that contract is settled. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
