You are viewing a plain text version of this content. The canonical link for it is here.
Posted to notifications@accumulo.apache.org by "Keith Turner (JIRA)" <ji...@apache.org> on 2014/04/15 19:23:15 UTC

[jira] [Resolved] (ACCUMULO-2674) Bulk test may not work in LongClean

     [ https://issues.apache.org/jira/browse/ACCUMULO-2674?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Keith Turner resolved ACCUMULO-2674.
------------------------------------

    Resolution: Not a Problem

All of the bulk nodes throw work onto a thread pool.  So the test quickly gets to the verify node.   The verify node blocks until everything in the thread pool is finished, then verifies the data.  So the test will probably always get to the verify step very quickly.   However the bulk test could very easily take more then 1 hour because the verify node could take a really long time, and nodes are not interrupted.   

> Bulk test may not work in LongClean
> -----------------------------------
>
>                 Key: ACCUMULO-2674
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-2674
>             Project: Accumulo
>          Issue Type: Bug
>    Affects Versions: 1.6.0
>            Reporter: Keith Turner
>
> The Bulk random walk test performs a verification step in its terminal node.  If the terminal node is never executed, then the test does not have much purpose.  The LongClean module runs bulk for 1 hour and then exits.  In this case the test may exit before verification occurs.
> Another issue is that bulk starts a thread pool which will it seems will not be cleaned up.
> Shard suffers from a similar issue, its terminal node performs a verification.  Howerver it also does intermediate verifications, so its not as bad. Bulk does no intermediate verification steps.



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