Yes at the moment, this could happen. However, the superfluous `TaskManager` should be released after it idled around for too long. Moreover, I'm currently working on making the `SlotManager` aware of how many outstanding slots he has requested. That way he should not allocate additional containers in case of a failover of the `ExecutionGraph`.
[ Full content available at: https://github.com/apache/flink/pull/5931 ] This message was relayed via gitbox.apache.org for [email protected]
