MartijnVisser commented on PR #254:
URL:
https://github.com/apache/flink-connector-jdbc/pull/254#issuecomment-6298252444
@eskabetxe Do you want to take a look at this draft PR? It deviates from the
accepted FLIP hence it's still a draft, because I want to send it to the Dev ML:
1. Instead of having a V2 class, we change this in place.
JdbcDynamicTableSink will return an internal, writer-only
Sink<RowData> that wraps JdbcWriter, with no committer and no writer
state. That is the
non-XA variant the FLIP describes, under the existing factory. Kinesis
(FLINK-24905) is
the precedent for not keeping the old sink behind an option;
Elasticsearch (FLINK-26638,
reverted in 1.15 because a legacy behavior was missing) is the reason the
legacy
behavior is pinned with tests first (FLINK-40742). The DataStream
JdbcSink is untouched.
2. No deprecation window. There is nothing for a user to migrate: same
identifier, same
options, same changelog semantics, and the writer registers no state, so
jobs with operator
uids restore their savepoints in both directions. The sink task is
renamed "<sink>: Writer",
the lineage dataset gets the physical table name instead of an empty one,
and the standard
writer metrics appear at zero. A statement set without uids in which a
JDBC sink receives a
rowtime column cannot restore an older savepoint, because the planner
puts a timestamp
inserter in front of every Sink V2 sink.
--
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]
