> On Oct. 29, 2013, 4:27 p.m., Bill Havanki wrote:
> > This is really crying out to be implemented using thread interruption 
> > instead of a flag check. However, it looks like the ZooStore implementation 
> > wouldn't deal with InterruptedExceptions well.

In my experience doing both can be a good strategy.   I tried using thread 
interruption in the tablet server before, but Hadoop client code was 
eating/ignoring the exceptions.   I think a lot code eats these exceptions.  It 
seems like InterruptedException would have worked so much better if they 
extended RuntimeException.


- kturner


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15002/#review27689
-----------------------------------------------------------


On Oct. 29, 2013, 4:28 p.m., John Vines wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/15002/
> -----------------------------------------------------------
> 
> (Updated Oct. 29, 2013, 4:28 p.m.)
> 
> 
> Review request for accumulo.
> 
> 
> Bugs: ACCUMULO-1307
>     https://issues.apache.org/jira/browse/ACCUMULO-1307
> 
> 
> Repository: accumulo
> 
> 
> Description
> -------
> 
> First pass at shutting down FATE. Maybe makes sense to daemonize the thread 
> and add a daemonized monitor thread that 'manages' them somehow in case 
> operation is stuck?
> 
> 
> Diffs
> -----
> 
>   fate/src/main/java/org/apache/accumulo/fate/Fate.java bd36edb 
>   server/src/main/java/org/apache/accumulo/server/master/Master.java c029ae5 
> 
> Diff: https://reviews.apache.org/r/15002/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> John Vines
> 
>

Reply via email to