hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Vinod Kumar Vavilapalli (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-2848) [HOD] If a cluster directory is deleted, hod -o list must show it, and deallocate should work.
Date Tue, 18 Mar 2008 12:19:24 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-2848?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12579818#action_12579818
] 

Vinod Kumar Vavilapalli commented on HADOOP-2848:
-------------------------------------------------

Some problems:
   * allocate a cluster, delete the cluster directory and then create the directory. hod list
again shows nothing. What should be the behaviour here?
   * allocate a cluster, delete the cluster directory, create the directory and then use the
same directory for another allocation. The second allocation goes through without fuss and
the first allocation still runs without the user knowing it. Here, I guess, we definitely
need some fix.

Other than that, a very minor thing : deallocating such clusters says "Invalid cluster directory
'<cluster-id>' specified. Deleted cluster with cluster id <cluster-id>". 'Deleting
cluster' here looked a bit confusing to me on the first look, could it be something like 'freed
up' or 'removed from queue'?

> [HOD] If a cluster directory is deleted, hod -o list must show it, and deallocate should
work.
> ----------------------------------------------------------------------------------------------
>
>                 Key: HADOOP-2848
>                 URL: https://issues.apache.org/jira/browse/HADOOP-2848
>             Project: Hadoop Core
>          Issue Type: Bug
>          Components: contrib/hod
>    Affects Versions: 0.16.0
>            Reporter: Hemanth Yamijala
>            Assignee: Hemanth Yamijala
>             Fix For: 0.17.0
>
>         Attachments: 2848.patch
>
>
> Currently if the cluster directory is deleted, all state about the cluster is lost. While
this in itself is not a problem, at least recovery in the sense of being able to list the
torque job id and deallocation to clear up the nodes should happen correctly.

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


Mime
View raw message