[
https://issues.apache.org/jira/browse/FLINK-14312?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Zhu Zhu updated FLINK-14312:
----------------------------
Description:
Logical pipelined region partitioning is needed by FLINK-14060 to determine
{{JobVertex}} slot sharing group.
It would be better if we can leverage
{{PipelinedRegionComputeUtil#computePipelinedRegions}} to do this. So that we
do not need to implement a very similar logic twice.
This would require a unification of the interfaces of FailoverTopology and
JobGraph.
With changes from FLINK-14453, we can build {{LogicalPipelinedRegions}} from
{{JobGraph}} by:
1. Introduce {{LogicalTopology}}
2. Implement {{DefaultLogicalTopology}} as an adapter of {{JobGraph}} to
{{LogicalTopology}}
3. Add {{DefaultLogicalTopology#getLogicalPipelinedRegions}} to return the
logical pipelined regions of a JobGraph
was:
Logical pipelined region partitioning is needed by FLINK-14060 to determine
{{JobVertex}} slot sharing group.
It would be better if we can leverage
{{PipelinedRegionComputeUtil#computePipelinedRegions}} to do this. So that we
do not need to implement a very similar logic twice.
This would require a unification of the interfaces of FailoverTopology and
JobGraph.
With changes from FLINK-14453, we can build {{LogicalPipelinedRegions}} from
{{JobGraph}} by:
1. Introduce {{LogicalTopology}}
2. Implement {{DefaultLogicalTopology}} as an adapter of {{JobGraph}} to
{{LogicalTopology}}
3. Add {{DefaultLogicalTopology#getLogicalPipelinedTopology}} to return the
logical pipelined regions of a JobGraph
> Support building logical pipelined regions from JobGraph
> --------------------------------------------------------
>
> Key: FLINK-14312
> URL: https://issues.apache.org/jira/browse/FLINK-14312
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / Coordination
> Affects Versions: 1.10.0
> Reporter: Zhu Zhu
> Assignee: Zhu Zhu
> Priority: Major
> Fix For: 1.10.0
>
>
> Logical pipelined region partitioning is needed by FLINK-14060 to determine
> {{JobVertex}} slot sharing group.
> It would be better if we can leverage
> {{PipelinedRegionComputeUtil#computePipelinedRegions}} to do this. So that we
> do not need to implement a very similar logic twice.
> This would require a unification of the interfaces of FailoverTopology and
> JobGraph.
> With changes from FLINK-14453, we can build {{LogicalPipelinedRegions}} from
> {{JobGraph}} by:
> 1. Introduce {{LogicalTopology}}
> 2. Implement {{DefaultLogicalTopology}} as an adapter of {{JobGraph}} to
> {{LogicalTopology}}
> 3. Add {{DefaultLogicalTopology#getLogicalPipelinedRegions}} to return the
> logical pipelined regions of a JobGraph
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
