Github user StephanEwen commented on the issue: https://github.com/apache/flink/pull/4220 We have to also consider anther aspect: Changing this from a synchronous call to an asynchronous callback makes new types of races possible, against which we need to guard. We need to make sure that the `restart()` call cannot succeed if there was another cycle of failure that goes into `RESTARTING`. I would suggest to address https://issues.apache.org/jira/browse/FLINK-6667 first, so that the callback can check that the `globalModVersion` is unchanged upon restart. Then there is no anger in moving this to an asynchronous callback.
--- 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 infrastruct...@apache.org or file a JIRA ticket with INFRA. ---