[
https://issues.apache.org/jira/browse/FLINK-5668?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15927823#comment-15927823
]
Stephan Ewen commented on FLINK-5668:
-------------------------------------
[~bill.liu8904] Okay, I see. This is a bigger change, though. I would suggest
to have a FLIP for that, because there are many aspects to discuss:
- The new flink-on-yarn-job is currently written to store the {{JobGraph}}
directly in HDFS as well so that it is part of the default resources of each
{{JobManager}}
- In HA mode, a writable persistent store is needed for BLOBs as well
- For flink-on-yarn-clusters (multiple jobs), we would need a persistent
writable store to store the {{JobGraphs}} of currently running jobs.
The general theme is: Some persistent store is needed currently, at least for
high-availability modes. Decoupling Yarn from a persistent store pushes the
responsibility to another layer.
> passing taskmanager configuration through taskManagerEnv instead of file
> ------------------------------------------------------------------------
>
> Key: FLINK-5668
> URL: https://issues.apache.org/jira/browse/FLINK-5668
> Project: Flink
> Issue Type: Improvement
> Components: YARN
> Reporter: Bill Liu
> Original Estimate: 48h
> Remaining Estimate: 48h
>
> When create a Flink cluster on Yarn, JobManager depends on HDFS to share
> taskmanager-conf.yaml with TaskManager.
> It's better to share the taskmanager-conf.yaml on JobManager Web server
> instead of HDFS, which could reduce the HDFS dependency at job startup.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
