SEZ9 commented on issue #12267: URL: https://github.com/apache/seatunnel/issues/12267#issuecomment-5691789862
Thanks @1362227089 for filling in the details — HighGo 4.5.8, SM3 authentication for the user, `postgresql-42.7.13.jar` on the classpath, and an independent connection with the same `jdbc:postgresql` URL and credentials succeeding. That gives us a much clearer picture. Where this stands: the Postgres-CDC connector uses `org.postgresql.Driver` and the PostgreSQL/Debezium contract, and the run stops at the driver's authentication handshake with unsupported authentication type `13` before any snapshot, replication slot, publication, or WAL decoding work begins. Since SM3 is HighGo-specific and the vendor guide you linked documents `com.highgo.jdbc.Driver` with a `jdbc:highgo:` URL, this is not something the existing connector can pick up by placing `HgdbJdbc-6.2.5.jar` next to the PostgreSQL jar — mixing drivers is not a supported compatibility path. So I don't see a confirmed defect in Postgres-CDC here; it's a compatibility gap that would need its own proposal. One thing that would help clarify the boundary: for the independent connection that succeeded, can you confirm which driver it used (`postgresql-42.7.13.jar` or the HighGo jar) and whether the user was still on SM3 for that test? If it went through the HighGo driver or a different auth method, that would explain the difference from the connector run. Remaining asks to move a compatibility proposal forward: 1. The relevant `pg_hba.conf` rule for this user/endpoint, and whether the server allows a PostgreSQL-standard method (e.g. scram-sha-256 or md5) for a CDC user — if so, please retry the connector with that user and share the full log. That is the quickest way to see whether snapshot and logical replication work on HighGo at all. 2. Public documentation for the SM3 authentication protocol and the license of the HighGo JDBC driver, so we can evaluate whether a separate HighGo driver path is feasible as a dependency. 3. If you get past authentication, a minimal sanitized snapshot-only run and a logical-replication run against the same database/schema/table, with complete logs. Please don't change replication ports or CDC config based on the authentication failure — nothing in the current log points there. Once we have the above, we can decide whether to convert this into a compatibility feature request. <!-- streview-comment:1088 --> -- 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]
