[ https://issues.apache.org/jira/browse/FLINK-12823?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16862690#comment-16862690 ]
Jingsong Lee commented on FLINK-12823: -------------------------------------- (y)(y)(y) The basic requirement is that batch jobs need to be set up with BLOCKING ResultPartitionType to avoid diamond-shaped DAG deadlocks. NIT: I think StreamTransformation are more accurate than DataStream. > PartitionTransformation supports DataExchangeMode property > ---------------------------------------------------------- > > Key: FLINK-12823 > URL: https://issues.apache.org/jira/browse/FLINK-12823 > Project: Flink > Issue Type: Improvement > Components: API / DataStream > Reporter: Biao Liu > Assignee: Biao Liu > Priority: Major > Fix For: 1.9.0 > > > Currently {{ResultPartitionType}} of {{DataStream}} job is hard-coded with > {{PIPELINED_BOUNDED}}. Since {{DataStream}} would support batch mode, there > should be way to set the {{ResultPartitionType}}. > So I propose introducing a new property {{DataExchangeMode}} of > {{PartitionTransformation}}. As a first step, this would be an internal API. > Maybe it would be exposed as a public API in the future. -- This message was sent by Atlassian JIRA (v7.6.3#76005)