hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Wangda Tan (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-4734) Merge branch:YARN-3368 to trunk
Date Thu, 14 Apr 2016 00:51:25 GMT

    [ https://issues.apache.org/jira/browse/YARN-4734?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15240333#comment-15240333
] 

Wangda Tan commented on YARN-4734:
----------------------------------

Thanks for comments, [~aw].

For your comments, actually attached patches are still WIP and I used it to figure out issues
like ASF licensing warnings. Will send out a mail to yarn-dev mail list when patch is ready
to be reviewed from my POV. I will add "wip" to patch name later to avoid confusing.

For your concerns:
bq. Definitely need some clarification from ASF legal whether we can merge licenses like that.
My hunch is no, but IANAL.
I can see some projects like Spark are using merged licenses. See LEGAL-226/SPARK-10833. And
I will send query about standard formats of Licenses as well.

bq. The dist and tmp directories should be inside target and not in the root of the module.
This makes a ton of other problems go away.
Will do

bq. Why is there a separate profile for this? What UI do I get if I don't build with this
profile? This also means the precommit hooks won't work until the hadoop personality is modified
(which means the above precommit testing is mostly useless)
Since it requires additional tools to build it (npm & bower), we cannot ask developers
to install them until it *officially* supported by YARN.
To make sure it can be run by Jenkins, can we modify Yetus (or Hadoop dev support script)
to make Jenkins can build / test it by adding the additional profile?

bq. Double check the license headers. At least one of 'em was using the old text.
It seems all existing YARN docs (*.md) are using old header. (If that's the older header you
mentioned above). I will fix yarnui2.md and rest of YARN docs can be fixed separately.

bq. Why isn't YarnUI2.md's content in BUILDING.txt? Why does an end user care about this information?
My understanding is BUILDING.txt should only contain how to build components. YarnUI2.md is
majorly about how to deploy and start new UI server. Contributor/volunteer can try it follow
the steps.

bq. The Apache RAT issues
Will fix

bq. Why does "hadoop-yarn-ui/src/main/resources/META-INF/NOTICE.txt" mention Tez?
Will fix.

bq. hadoop-yarn-ui/src/main/webapp/package.json should have it's version pulled from maven.
>From my investigation, we cannot pass down the version to ember to build the package from
CLI, it has to be picked from package.json.
We basically have two choices:
1) Modify package.json automatically in maven building script and pass down the version from
maven.
2) Give a separate version to yarn-ui module. 0.0.0 in the patch doesn't make sense at all,
how about call it 0.1?
I'm not sure how we dealed with version of libhadoop.so. Would like to hear your thoughts.



> Merge branch:YARN-3368 to trunk
> -------------------------------
>
>                 Key: YARN-4734
>                 URL: https://issues.apache.org/jira/browse/YARN-4734
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>            Reporter: Wangda Tan
>            Assignee: Wangda Tan
>         Attachments: YARN-4734.1.patch, YARN-4734.2.patch, YARN-4734.3.patch, YARN-4734.4.patch,
YARN-4734.5.patch
>
>
> YARN-2928 branch is planned to merge back to trunk shortly, it depends on changes of
YARN-3368. This JIRA is to track the merging task.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message