You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Till Rohrmann (JIRA)" <ji...@apache.org> on 2018/03/20 18:05:01 UTC
[jira] [Resolved] (FLINK-8881) Accumulators not updated for running
jobs
[ https://issues.apache.org/jira/browse/FLINK-8881?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Till Rohrmann resolved FLINK-8881.
----------------------------------
Resolution: Fixed
Fixed via
master: 95d4c0170339585cb4876c4bafc1c2a42c44be3c
1.5.0: 6657aa2090b6b0f56f329152b3aaa1c147e73380
> Accumulators not updated for running jobs
> -----------------------------------------
>
> Key: FLINK-8881
> URL: https://issues.apache.org/jira/browse/FLINK-8881
> Project: Flink
> Issue Type: Bug
> Components: DataSet API, DataStream API, Distributed Coordination, JobManager, TaskManager
> Affects Versions: 1.5.0
> Reporter: Chesnay Schepler
> Assignee: Chesnay Schepler
> Priority: Blocker
> Labels: flip-6
> Fix For: 1.5.0
>
>
> The FLIP-6 {{TaskExecutor}} is never sending the current state of accumulators to the JobMaster. They are only updated if the job is finished.
> The legacy TaskManager did this regularly as part of the heartbeat to the JobManager.
> This is a regression and blocks the porting of some tests (like the {{SavepointMigrationTestBase}}) that makes use of accumulators to determine when the job shutdown condition is fulfilled.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)