You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@beam.apache.org by "Brian Hulette (Jira)" <ji...@apache.org> on 2022/05/24 01:30:00 UTC

[jira] [Commented] (BEAM-14481) OOM regression caused by Batched DoFn worker changes

    [ https://issues.apache.org/jira/browse/BEAM-14481?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17541229#comment-17541229 ] 

Brian Hulette commented on BEAM-14481:
--------------------------------------

It turns out the issue in internal pipelines was not an OOM, there was already memory pressure in this pipeline. The issue appears to be a bug in a minor changed to the worker code that affected metric reporting.

> OOM regression caused by Batched DoFn worker changes
> ----------------------------------------------------
>
>                 Key: BEAM-14481
>                 URL: https://issues.apache.org/jira/browse/BEAM-14481
>             Project: Beam
>          Issue Type: Bug
>          Components: sdk-py-core
>            Reporter: Brian Hulette
>            Priority: P1
>          Time Spent: 50m
>  Remaining Estimate: 0h
>
> We've observed some OOMs in internal pipelines. We've narrowed the root cause down to [https://github.com/apache/beam/pull/17384] - the worker changes for Batched DoFns.
> I still don't understand the root cause, but given this change can lead to a regression in pipelines that don't use the feature I think we should roll it back.



--
This message was sent by Atlassian Jira
(v8.20.7#820007)