[jira] [Commented] (ACCUMULO-2719) Shell fails to find instance_id when using viewfs

["Josh Elser (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ACCUMULO-2719?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13977288#comment-13977288 ] 

Josh Elser commented on ACCUMULO-2719:
--------------------------------------

[~dlmarion], can you elaborate on your configuration here (notably the instance.* values in accumulo-site.xml and your viewfs mount table)? You're trying to run Accumulo over ViewFS, but as I understand it, viewfs presents a single Hadoop FileSystem over multiple DFS instances using mount-points, so I'm confused as to the environment.

You mention {{instance.volumes}}, but I would have thought that you'd use viewfs to try to alleviate the need to specify multiple volumes when you only want to pin a certain table to a given filesystem (and not spread files across many filesystems).

Also, how/where did {{accumulo init}} write the instance_id?

> Shell fails to find instance_id when using viewfs
> -------------------------------------------------
>
>                 Key: ACCUMULO-2719
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-2719
>             Project: Accumulo
>          Issue Type: Bug
>          Components: shell
>    Affects Versions: 1.6.0
>            Reporter: Dave Marion
>            Priority: Minor
>
> Shell fails to find instance id at /accumulo/instance_id when just the name service roots are configured in viewfs. Fails at ZooUtil.getInstanceIDFromHDFS(line 63). I'm thinking the code should prepend one of the instance.volumes.
> Workaround is to explicitly map /accumulo/instance_id in viewfs to one of the locations.



--
This message was sent by Atlassian JIRA
(v6.2#6252)