[
https://issues.apache.org/jira/browse/FALCON-1162?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14514088#comment-14514088
]
karan kumar commented on FALCON-1162:
-------------------------------------
[~venkat.ramachan], Just ran it on my machine also; feed scheduling was
breaking when cluster staging dir perms were 755
I made the changes on these lines
https://issues.apache.org/jira/browse/FALCON-910?focusedCommentId=14255793&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-14255793
I somehow overlooked this scenario, mybad. The patch LGTM. Just ensure you also
make changes to the documentation at:
https://github.com/apache/falcon/blob/master/docs/src/site/twiki/EntitySpecification.twiki
Thanks for catching this.
> Cluster submit succeeds when staging HDFS dir does not have 777 (ALL)
> permission
> ---------------------------------------------------------------------------------
>
> Key: FALCON-1162
> URL: https://issues.apache.org/jira/browse/FALCON-1162
> Project: Falcon
> Issue Type: Bug
> Components: common
> Affects Versions: 0.6
> Reporter: Venkat Ramachandran
> Assignee: Venkat Ramachandran
> Priority: Blocker
> Fix For: 0.6.1
>
> Attachments: falcon-1162.1.patch, falcon-1162.2.patch
>
>
> Staging HDFS dir specified in the cluster definition should have WORLD
> WRITABLE permission. It seems Cluster entity submit used to validate this
> condition in order to avoid further runtime failures.
> But, this check had been reverted as part of the FALCON-910 commits (only
> checks for 755) as below:
> ClusterEntityParser.java
> {code}
> checkPathOwnerAndPermission(cluster.getName(),
> stagingLocation.getPath(), fs, HadoopClientFactory.READ_EXECUTE_PERMISSION,
> false);
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
