Github user StephanEwen commented on the issue:
https://github.com/apache/flink/pull/2580
Looks quite good. I would suggest one change, though:
Can we avoid copying the offsets in the checkpoint into a new map (with
increment by one) and passing that to the ZooKeeper Offset Committer or the
Kafka Offset Committer? I am just not a big fan of copying things back and
forth (especially in "prepareSnaoshot()", which we want to keep as lightweight
as possible). Instead, can we have the contract that the offset committers
always commit "+1" from the value they get (pretty much as it was in the 0.9
committer after FLINK-4618)?
Concerning the tests, is the stability issue fixed there?
What I frequently do is push the same commit to 10 different newly created
branches to keep Travis busy over night with 10 test runs and see if I see a
stability issue.
---
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.
---