FLINK-13708 is also very critical IMO. This would cause invalid flink job (doubled output)
https://issues.apache.org/jira/browse/FLINK-13708 Jark Wu <imj...@gmail.com> 于2019年9月23日周一 下午2:03写道： > Hi everyone, > > It has already been a month since we released Flink 1.9.0. > We already have many important bug fixes from which our users can benefit > in the release-1.9 branch (83 resolved issues). > Therefore, I propose to create the next bug fix release for Flink 1.9. > > Most notable fixes are: > > - [FLINK-13526] When switching to a non existing catalog or database in the > SQL Client the client crashes. > - [FLINK-13568] It is not possible to create a table with a "STRING" data > type via the SQL DDL. > - [FLINK-13941] Prevent data-loss by not cleaning up small part files from > S3. > - [FLINK-13490][jdbc] If one column value is null when reading JDBC, the > following values will all be null. > - [FLINK-14107][kinesis] When using event time alignment with the Kinsesis > Consumer the consumer might deadlock in one corner case. > > Furthermore, I would like the following critical issues to be merged before > 1.9.1 release: > > - [FLINK-14118] Reduce the unnecessary flushing when there is no data > available for flush which can save 20% ~ 40% CPU. (reviewing) > - [FLINK-13386] Fix A couple of issues with the new dashboard have already > been filed. (PR is created, need review) > - [FLINK-14010][yarn] The Flink YARN cluster can get into an inconsistent > state in some cases, where > leaderhship for JobManager, ResourceManager and Dispatcher components is > split between two master processes. (PR is created, need review) > > I would volunteer as release manager and kick off the release process once > blocker issues has been merged. What do you think? > > If there is any other blocker issues need to be fixed in 1.9.1, please let > me know. > > Cheers, > Jark > -- Best Regards Jeff Zhang
