[jira] [Commented] (ACCUMULO-4619) Add splits hung forever

["Keith Turner (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ACCUMULO-4619?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15957247#comment-15957247 ] 

Keith Turner commented on ACCUMULO-4619:
----------------------------------------

While looking into fixing this I learned something interesting.  The existing code supplies a thread factory to the executor service that supplies a Thread with a LoggingRunnable.  So it seems like the LoggingRunnable should have logged the Error.  However, in this case there was thread pool code that caught the exception and ate it before it got to LoggingRunnable.

So the LoggingRunnable created by org.apache.accumulo.core.util.NamingThreadFactory may not be doing much for us when used in a thread pool.

> Add splits hung forever
> -----------------------
>
>                 Key: ACCUMULO-4619
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-4619
>             Project: Accumulo
>          Issue Type: Bug
>    Affects Versions: 1.7.3, 1.8.1
>            Reporter: Keith Turner
>            Assignee: Keith Turner
>             Fix For: 1.7.4, 1.8.2, 2.0.0
>
>
> I ran into a problem where addsplits was hanging forever and no errors were printed by Accumulo even when I set the log level to trace.  I finally figured out the problem was that I had the wrong version of thrift on the classpath (0.9.1 vs 0.9.3).  Accumulo 1.8.1 was trying to use a method that did not exist in thrift 0.9.1 :(  
> This missing method caused a NoSuchMethodError to be thrown.  The Accumulo split code only caught Exception and therefore did not handle this.  Also nothing was printed for this.   
> The split code should be fixed to not silently drop errors.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)