[
https://issues.apache.org/jira/browse/HADOOP-3045?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12581364#action_12581364
]
Hemanth Yamijala commented on HADOOP-3045:
------------------------------------------
Mahadev, understand your intent. The problem is that the tarball path is on the
ringmaster node, not the local file system. Hence we cannot check before
allocation, only when the ringmaster runs. Typically the tarball file is a path
that's common to both the node from where hod is running, and where the
ringmaster runs - but that need not be the case always.
If we could take the tarball file as a local file path, and transfer it
ourselves to the ringmaster node first, then we could check.
Alternatively, if we could get the errors from the ringmaster node to the
client node, that would improve the error reporting.
Given the more general purpose utility of the second case, I think we should
move in that direction.
However, both approaches are significant changes over the existing
implementation.
> hod -t option fails silently if the tarball specified does not exist.
> ---------------------------------------------------------------------
>
> Key: HADOOP-3045
> URL: https://issues.apache.org/jira/browse/HADOOP-3045
> Project: Hadoop Core
> Issue Type: Bug
> Components: contrib/hod
> Affects Versions: 0.16.1
> Reporter: Mahadev konar
> Assignee: Hemanth Yamijala
> Priority: Minor
>
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
