[ 
https://issues.apache.org/jira/browse/HADOOP-4546?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12678079#action_12678079
 ] 

Chris Douglas commented on HADOOP-4546:
---------------------------------------

bq. I don't see any reason why this can't be included in 0.18.4 as the original 
code that is patched has not been changed over the last few releases.

That's not one of the criteria Hadoop uses to determine whether a patch should 
be backported. That the code is stable is an argument _against_ backporting 
this for two reasons. First, there are no known issues with it and any changes 
may destabilize it, requiring more maintenance releases. Second, applying the 
patch isn't unduly burdensome, neither does it require any special build 
instruction, so interested parties can apply this themselves. Hadoop's release 
criteria are conservative, in part because most organizations want to backport 
some subset of later releases into stable branches.

With few exceptions, the only patches that get pushed into older branches are 
those that fix behavior that was unintentionally and incompatibly broken in 
that branch or earlier. If it can be demonstrated that a previous version of 
Hadoop was compatible with AIX, then the patch can be pushed into previous 
releases, but lacking that it requires a special dispensation.

> Minor fix in dfs to make hadoop work in AIX
> -------------------------------------------
>
>                 Key: HADOOP-4546
>                 URL: https://issues.apache.org/jira/browse/HADOOP-4546
>             Project: Hadoop Core
>          Issue Type: Improvement
>          Components: dfs
>    Affects Versions: 0.18.0, 0.19.0
>         Environment: AIX 
>            Reporter: Arun Venugopal
>            Assignee: Bill Habermaas
>             Fix For: 0.21.0
>
>         Attachments: 4546-1.patch, 4546-2.patch, dfs_aix_fix.patch, 
> hadoop-4546.patch, TestDFVariations.java
>
>
> HDFS uses df command to get the disk space. The output format of df command 
> is different in AIX compared to Linux and Solaris. Checking the OS type and 
> then reading the df output as per the format will fix this issue and thus 
> allows hadoop to be fully functional in AIX environment.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to