[
https://issues.apache.org/jira/browse/FLINK-25914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17490280#comment-17490280
]
Roman Khachatryan commented on FLINK-25914:
-------------------------------------------
I've double-checked that the regression is caused by FLINK-25479 by reverting
53c3525d0f4d4b3117edf1f8fcce0ff2ecfdda19:
!image-2022-02-10-15-01-06-613.png|width=750!
The change made Task to always serialize its state snapshot before sending it
to JM.
Before that, on mini-cluster, it was not serialized if the Akka connection is
local. That caused issues with sharing JVM objects between JM and TM.
I'm going to close the ticket because this should not affect production
environments.
> Performance regression in checkpointSingleInput.UNALIGNED on 31.01.2022
> -----------------------------------------------------------------------
>
> Key: FLINK-25914
> URL: https://issues.apache.org/jira/browse/FLINK-25914
> Project: Flink
> Issue Type: Bug
> Components: Benchmarks, Runtime / Checkpointing
> Affects Versions: 1.15.0
> Reporter: Piotr Nowojski
> Assignee: Roman Khachatryan
> Priority: Critical
> Attachments: image-2022-02-10-15-01-06-613.png
>
>
> http://codespeed.dak8s.net:8000/timeline/?ben=checkpointSingleInput.UNALIGNED&env=2
> suspected range e2f609c929..cd8ea8d5b2
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
