You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@beam.apache.org by "Kyle Weaver (Jira)" <ji...@apache.org> on 2020/05/04 20:21:00 UTC
[jira] [Resolved] (BEAM-9877) Eager size estimation of large
group-by-key iterables cause expensive / duplicate reads
[ https://issues.apache.org/jira/browse/BEAM-9877?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Kyle Weaver resolved BEAM-9877.
-------------------------------
Fix Version/s: 2.21.0
Resolution: Fixed
> Eager size estimation of large group-by-key iterables cause expensive / duplicate reads
> ---------------------------------------------------------------------------------------
>
> Key: BEAM-9877
> URL: https://issues.apache.org/jira/browse/BEAM-9877
> Project: Beam
> Issue Type: Bug
> Components: sdk-java-core
> Reporter: Tudor Marian
> Assignee: Tudor Marian
> Priority: Major
> Fix For: 2.21.0
>
> Time Spent: 1h
> Remaining Estimate: 0h
>
> BatchGroupAlsoByWindowViaIteratorsFn WindowReiterator can cause very expensive duplicate reading of data from a (Co-)GroupByKey for hot keys with many values due to PCollection size estimation. Instead, it should perform lazy estimation like GroupingShuffleReader and GroupingShuffleEntryInterator (or perform no estimation at all).
--
This message was sent by Atlassian Jira
(v8.3.4#803005)