pnowojski opened a new pull request #18420:
URL: https://github.com/apache/flink/pull/18420


   Previously, if there was a node that was self-unioned with itself,
   it was creating a situation with two identical StreamEdges. Both
   with the same partitioning, from the same source node to the same
   target node.
   
   This was causing issues when constructing output collectors and
   picking the correct RecordWriters, as StreamTask was not able to
   uniquely identify given StreamEdge and was assigning the same
   RecordWriter to both of the edges. As a result all stream elements
   were sent twice through the same RecordWriter. It was actually pretty
   harmless apart of calculating the combined watermark downstream,
   since all watermarks were always comming just from one single
   edge/inputgate, and the unused edges were always stuck with
   min watermark.
   
   As a solution we are making sure that StreamEdges are unique
   by introducing a uniqueId field, incremented for every pair
   of StreamEdges connecting the same node.
   
   ## Verifying this change
   
   This change adds a new `checkState` for early detection of similar problems 
and a new ITCase.
   
   ## Does this pull request potentially affect one of the following parts:
   
     - Dependencies (does it add or upgrade a dependency): (yes / **no**)
     - The public API, i.e., is any changed class annotated with 
`@Public(Evolving)`: (yes / **no**)
     - The serializers: (yes / **no** / don't know)
     - The runtime per-record code paths (performance sensitive): (yes / **no** 
/ don't know)
     - Anything that affects deployment or recovery: JobManager (and its 
components), Checkpointing, Kubernetes/Yarn, ZooKeeper: (yes / **no** / don't 
know)
     - The S3 file system connector: (yes / **no** / don't know)
   
   ## Documentation
   
     - Does this pull request introduce a new feature? (yes / **no**)
     - If yes, how is the feature documented? (**not applicable** / docs / 
JavaDocs / not documented)
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


Reply via email to