hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Junping Du (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-7162) Remove XML excludes file format
Date Thu, 14 Sep 2017 22:02:00 GMT

    [ https://issues.apache.org/jira/browse/YARN-7162?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16167024#comment-16167024
] 

Junping Du commented on YARN-7162:
----------------------------------

bq.  if you wanted to have nodeA do 50, nodeB do 70, etc, you can't do that without the XML
or JSON format. I'm not sure how useful that is.
It really depend on how important we think this timeout is. In the beginning, we think timeout
value is not that matter, so we have a simple, client-side tracking on timeout values. But
later on, [~danzhi] from AWS claim that timeout value matters a lot, and we need a more reliable
way to track it - through RM. Also, multiple nodes has different timeout values is the case
they want to cover because the precisely for each node's timeout may also be important. I
can quickly think of some use cases, such as: admin could plan to decommission some nodes
earlier than other nodes. Or some nodes's containers/applications (in node label case) are
more important/critical than other nodes, so worth to wait a bit longer, etc.


> Remove XML excludes file format
> -------------------------------
>
>                 Key: YARN-7162
>                 URL: https://issues.apache.org/jira/browse/YARN-7162
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: graceful
>    Affects Versions: 2.9.0, 3.0.0-beta1
>            Reporter: Robert Kanter
>            Assignee: Robert Kanter
>            Priority: Blocker
>         Attachments: YARN-7162.001.patch, YARN-7162.branch-2.001.patch
>
>
> YARN-5536 aims to replace the XML format for the excludes file with a JSON format.  However,
it looks like we won't have time for that for Hadoop 3 Beta 1.  The concern is that if we
release it as-is, we'll now have to support the XML format as-is for all of Hadoop 3.x, which
we're either planning on removing, or rewriting using a pluggable framework.  
> [This comment in YARN-5536|https://issues.apache.org/jira/browse/YARN-5536?focusedCommentId=16126194&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-16126194]
proposed two quick solutions to prevent this compat issue.  In this JIRA, we're going to remove
the XML format.  If we later want to add it back in, YARN-5536 can add it back, rewriting
it to be in the pluggable framework.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org


Mime
View raw message