Github user aljoscha commented on the issue:
https://github.com/apache/flink/pull/4239
I did a first high-level review of the code. I think it's good so far!
Before we can merge this, however, we need a few more things around it:
- A section in the Kafka doc about the new exactly-once mode, how it can
be configured etc.
- A big disclaimer (possibly in an "alert" box) about the interplay with
the transaction timeout and what the caveats there are
- A section in the Javadocs about the aforementioned caveats
- A check that ensures that the transaction timeout is set to a reasonably
high setting (say 1 hour) when exactly-once semantics are enabled. (With an
override setting that allows the user to set a lower transaction time out if
they want to.)
Also, this has interplay with #4616 but we can resolve that by merging them
in any order and fixing up the later changes when merging.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
