zhijiangW opened a new pull request #10029:  [FLINK-14553][runtime] Respect 
non-blocking output in StreamTask#processInput
URL: https://github.com/apache/flink/pull/10029
 
 
   ## What is the purpose of the change
   
   The non-blocking output was introduced in FLINK-14396 and FLINK-14498 to 
solve the problem of handling the checkpoint barrier in the case of 
backpressure.
       
   In order to make the whole process through, `StreamInputProcessor` should be 
allowed to process input elements if the output is also available.
      
   The default core size of `LocalBufferPool` for `ResultPartition` should also 
be increased by 1 for two considerations:
   
   - `StreamTask` can only process input if there is at-least one available 
buffer on output side, so it might cause stuck problem if the minimum pool size 
is exactly equal to the number of subpartitions, because every subpartition 
might maintain a partial unfilled buffer.
   
   - Increases one more buffer for every output `LocalBufferPool` to void 
performance regression if processing input is based on at-least one buffer 
available on output side.
   
   ## Brief change log
   
     - *Rename `AvailabilityProvider#isAvailable` as 
`AvailabilityProvider#getAvailableFuture`*
     - *Refactor `ResultPartitionWriter` interface to abstract class*
     - *Refactor `AvailabilityProvider` to provide default `#isAvailable` and 
`#isVolatileAvailable` methods*
     - *Adjust `StreamTask` stack to consider output status while `processInput`
     - *Increase one more minimum size for output `LocalBufferPool`*
   
   ## Verifying this change
   
   Some new unit tests would be added later after confirming the core process 
make sense.
   
   ## 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, Yarn/Mesos, 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.
 
For queries about this service, please contact Infrastructure at:
[email protected]


With regards,
Apache Git Services

Reply via email to