nzw921rx commented on issue #13223:
URL: https://github.com/apache/seatunnel/issues/13223#issuecomment-6791273991

   > [@nzw921rx](https://github.com/nzw921rx) Thanks. I will start Step 1 with 
`MultiTableSinkReplicaMode replicaMode()` as you describe, set the Iceberg E2E 
and the six doc examples to 1, and add the restore check in 
`MultiTableSink.restoreWriter()` with a test and docs.
   > 
   > One point before I write it. With `UNSUPPORTED` as the default and 
`primaryKey()` no longer enough, Jdbc is also rejected until it declares 
`KEYED`. On dev `f5374addb3` that affects:
   > 
   > * five E2E configs with `multi_table_sink_replica = 2`, all Jdbc sinks 
without exactly-once: `mysqlcdc_to_mysql_with_schema_change.conf` and 
`mysqlcdc_to_mysql_with_multi_db_same_name_schema_change.conf` in 
`connector-cdc-mysql-e2e`, and `mysqlcdc_to_dm_with_schema_change.conf`, 
`mysqlcdc_to_postgres_with_schema_change.conf` and 
`mysqlcdc_to_sqlserver_with_schema_change.conf` in `connector-jdbc-e2e-ddl`
   > * the EN and ZH Jdbc examples with replicas in OceanBase, Oracle, 
PostgreSql, `schema-evolution.md` and `multi-table.md`
   > * users who run Jdbc with replicas today, whose jobs would fail after 
upgrading until the JDBC row lands
   > 
   > Should this PR set these to 1 as well and leave Jdbc to the JDBC row, or 
should `JdbcSinkWriter` declare `KEYED` here when its key is present, so the 
current Jdbc behavior stays while the JDBC row verifies it? I can take the JDBC 
normal row right after this one.
   > 
   > Also, dev does not create writers at runtime yet; only 
[#11206](https://github.com/apache/seatunnel/pull/11206) does. I will put the 
validation in one method that such a path can call before it registers a writer.
   
   In Phase 1, JDBC should override the new replicaMode() method to return 
KEYED, preserving its existing routing behavior.


-- 
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]

Reply via email to