One of the subtasks doesn't have any input(i.e. bytes received are 0), and checkpoint of this subtask is stuck.
Is it normal, or is it a bug?
The flink version is 1.6.2.
Add more information.
Below is the topology.
And the third task(i.e. 'Window(TumblingProcessingTimeWindows(30000), ProcessingTimeTrigger, ScalaReduceFunction, PassThroughWindowFunction) -> Map -> Sink: Unnamed') has five subtasks whose 'Bytes received' is 0. And checkpoint stuck on these five subtasks. One of the subtasks is shown as below.
And every time I restart the application, 'Bytes received 0' happens on different number of subtasks.
Why?
Flink 1.6.2 had some bugs related to checkpointing. Can you upgrade to 1.6.4, perhaps, and see if that resolves the issue?