[ 
https://issues.apache.org/jira/browse/YARN-233?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13503377#comment-13503377
 ] 

Bikas Saha commented on YARN-233:
---------------------------------

Its common to suffix patches as .1.patch, .2.patch etc so that its easy to 
download and diff etc.

Changes look good. 

For the kill -0 change I think I would do it a little differently. I would 
create containerExists() or containerIsAlive() and use that for the alive check 
and create a killContainer(TYPE kill) for terminating containers. That does not 
leak implementation details of Unix like Signal handling into the Hadoop code. 
Might remove the need to change the native code also. What do you think?

About ProcessIdFileReader. I am not sure about the reasoning for that code 
since it predates me. I do see that checking formats helps us differentiate 
among OS implementations though I am not sure if that was the original 
reasoning :P e.g. if we are on Linux we should check that we are getting 
expected numeric pids. If we are on windows we should check that we are getting 
containerIds. This helps fail fast. I am not sure if the upper level code would 
fail if we asked it to use wrong process identifiers.


                
> YARN support for execution of container on Windows
> --------------------------------------------------
>
>                 Key: YARN-233
>                 URL: https://issues.apache.org/jira/browse/YARN-233
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: nodemanager
>    Affects Versions: trunk-win
>            Reporter: Chris Nauroth
>            Assignee: Chris Nauroth
>         Attachments: YARN-233-branch-trunk-win.patch, 
> YARN-233-branch-trunk-win.patch
>
>
> YARN container launch code requires updates for compatibility with Windows.  
> Instead of launching and killing tasks via generated bash scripts and kill, 
> use generated cmd scripts and winutils for task management.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to