[ 
https://issues.apache.org/jira/browse/YARN-2369?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14527328#comment-14527328
 ] 

Jason Lowe commented on YARN-2369:
----------------------------------

bq. any thoughts on things that are missing from this latest patch or problems 
with the design?

+1 to the approach of using a whitelist for appendable variables.  I think this 
is a reasonable approach to preserve compatibility, especially for admin vs. 
user variables like mapreduce.admin.user.env vs. mapred.child.env.  It can be 
important for admin path settings like LD_LIBRARY_PATH to not be completely 
smashed by user settings, as users love to forget to specify the path to native 
hadoop libs when specifying LD_LIBRARY_PATH for their own libs.

> Environment variable handling assumes values should be appended
> ---------------------------------------------------------------
>
>                 Key: YARN-2369
>                 URL: https://issues.apache.org/jira/browse/YARN-2369
>             Project: Hadoop YARN
>          Issue Type: Bug
>    Affects Versions: 2.2.0
>            Reporter: Jason Lowe
>            Assignee: Dustin Cote
>         Attachments: YARN-2369-1.patch, YARN-2369-2.patch
>
>
> When processing environment variables for a container context the code 
> assumes that the value should be appended to any pre-existing value in the 
> environment.  This may be desired behavior for handling path-like environment 
> variables such as PATH, LD_LIBRARY_PATH, CLASSPATH, etc. but it is a 
> non-intuitive and harmful way to handle any variable that does not have 
> path-like semantics.



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

Reply via email to