[
https://issues.apache.org/jira/browse/SPARK-30108?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16991381#comment-16991381
]
Herman van Hövell commented on SPARK-30108:
-------------------------------------------
[~Ankitraj] Great to hear! Here are some pointers, the goal of this ticket to
make the current AggregatingAccumulator to only add the result of a partition
once. For this we need to do a couple of things:
- We need to add some tracking mechanism to the AggregatingAccumulator that
keeps track of the partitions that have been processed. A (compressible) bit
map should be enough.
- We need a way to tell the accumulator's merge method which partition it is
processing. One way of doing this would be making a small change in the way the
DAGScheduler interacts with the accumulator, and also pass Task/Partition info
on calling update. A less intrusive way of doing this would be to pass back
partition info with the to-be-merged accumulator.
> Add robust accumulator for observable metrics
> ---------------------------------------------
>
> Key: SPARK-30108
> URL: https://issues.apache.org/jira/browse/SPARK-30108
> Project: Spark
> Issue Type: New Feature
> Components: SQL
> Affects Versions: 3.0.0
> Reporter: Herman van Hövell
> Priority: Major
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
