accumulo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From ke...@deenlo.com
Subject Re: Review Request 21557: ACCUMULO-2766 fix wal group commit
Date Fri, 16 May 2014 17:58:30 GMT


> On May 16, 2014, 5:30 p.m., John Vines wrote:
> > server/tserver/src/main/java/org/apache/accumulo/tserver/log/DfsLogger.java, line
153
> > <https://reviews.apache.org/r/21557/diff/1/?file=583591#file583591line153>
> >
> >     I think that before this loop the closeLock and/or closed should be checked
and if it is closed, then LogClosedExceptions should be placed on all of the LogWorks, not
the Exception received.

I am not sure this is needed, but I have not completely convinced myself yet.  Theoretically
DfsLogger.close() should wait for the background thread to finish all work before it acutally
closes the log file.  However, I am not sure why this synchronization in the background thread
was added, which makes me unsure.  


- kturner


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/21557/#review43223
-----------------------------------------------------------


On May 16, 2014, 5:17 p.m., kturner wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/21557/
> -----------------------------------------------------------
> 
> (Updated May 16, 2014, 5:17 p.m.)
> 
> 
> Review request for accumulo.
> 
> 
> Bugs: ACCUMULO-2766
>     https://issues.apache.org/jira/browse/ACCUMULO-2766
> 
> 
> Repository: accumulo
> 
> 
> Description
> -------
> 
> A possible fix for ACCUMULO-2766.
> 
> 
> Diffs
> -----
> 
>   server/tserver/src/main/java/org/apache/accumulo/tserver/log/DfsLogger.java eb04f09

> 
> Diff: https://reviews.apache.org/r/21557/diff/
> 
> 
> Testing
> -------
> 
> Only performance testing so far.  This patch dramatically improves performance.
> 
> 
> Thanks,
> 
> kturner
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message