You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by GitBox <gi...@apache.org> on 2019/05/07 16:11:55 UTC
[GitHub] [flink] azagrebin opened a new pull request #8362: [FLINK-11391]
Introduce shuffle master interface
azagrebin opened a new pull request #8362: [FLINK-11391] Introduce shuffle master interface
URL: https://github.com/apache/flink/pull/8362
## What is the purpose of the change
This PR introduces ShuffleMaster interface. JobMaster and ExecutionGraph use ShuffleMaster to register resources for produced partition of intermediate results. This is part of the overall effort to make shuffle implementation pluggable.
## Brief change log
- Introduce ShuffleMaster interface
- Introduce ProducerShuffleDescriptor to describe producer for ShuffleMaster
- Introduce PartitionShuffleDescriptor to describe partition for ShuffleMaster
- Introduce ShuffleDeploymentDescriptor: partition handle produced by ShuffleMaster
to hand over to task and its local shuffle service
- Introduce UnknownShuffleDeploymentDescriptor for eager consumer scheduling
when producer is unknown yet
- Register produced partitions when execution gets slot
- Refactor TaskDeploymentDescriptor creation into a TaskDeploymentDescriptorFactory
- Refactor ResultPartitionDeploymentDescriptor to consist of
PartitionShuffleDescriptor and ShuffleDeploymentDescriptor
- Refactor InputGateDeploymentDescriptor and PartitionInfo to use ShuffleDeploymentDescriptor
instead of InputChannelDeploymentDescriptor
- Introduce DefaultShuffleMaster and DefaultShuffleDeploymentDescriptor implementations
for existing shuffle service based on Netty communication and local files
- Refactor SingleInputGate.create and updateInputChannel to determine partition location based on
consumer resource id and producer id from DefaultShuffleDeploymentDescriptor
- Adjust tests
## Verifying this change
unit tests
## Does this pull request potentially affect one of the following parts:
- Dependencies (does it add or upgrade a dependency): (no)
- The public API, i.e., is any changed class annotated with `@Public(Evolving)`: (no)
- The serializers: (no)
- The runtime per-record code paths (performance sensitive): (no)
- Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Yarn/Mesos, ZooKeeper: (yes)
- The S3 file system connector: (no)
## Documentation
- Does this pull request introduce a new feature? (no)
- If yes, how is the feature documented? (not applicable)
----------------------------------------------------------------
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:
users@infra.apache.org
With regards,
Apache Git Services