accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Josh Elser (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ACCUMULO-1919) Size of visibility cache in AccumuloFileOutputFormat should be configurable
Date Tue, 03 Jun 2014 15:39:02 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-1919?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14016784#comment-14016784
] 

Josh Elser commented on ACCUMULO-1919:
--------------------------------------

Thanks for the patch, [~dallaybatta]. Overall, it looks pretty good, but I have some suggestions.

You added the {{VISIBILITY_DISTRIBUTION_COUNT}} to the {{Property}} class to match what the
other options in FileOutputConfigurator was doing, which makes sense at a glance. However,
those existing properties are also valid when running an Accumulo instance, where the option
for the size of the visibility cache is only relevant when running MapReduce. I think it would
make sense to define this property somewhere in the MapReduce code, maybe {{ConfiguratorBase.GeneralOpts}},
instead of {{Property}}. I also think naming the constant {{VISIBILITY_CACHE_SIZE}} would
also be more intuitive than {{VISIBILITY_DISTRIBUTION_COUNT}}.

The description in the javadoc for {{FileOutputConfigurator.setValidVisibilityCacheSize(Class,
Configuration, int)}} has some duplicated text: "Sets the Sets the ..."

The good: patch looks to be in the right format, code formatting looks good, javadoc included,
and unit test additions.

> Size of visibility cache in AccumuloFileOutputFormat should be configurable
> ---------------------------------------------------------------------------
>
>                 Key: ACCUMULO-1919
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-1919
>             Project: Accumulo
>          Issue Type: Improvement
>          Components: client
>    Affects Versions: 1.5.0, 1.6.0
>            Reporter: Mike Drob
>            Priority: Minor
>              Labels: newbie
>         Attachments: ACCUMULO-1919.patch
>
>
> AccumuloFileOutputFormat uses an LRUMap hard-coded to max size of 1000. It is possible
that the client writing the data knows more about the visibility distribution and can specify
a better value if this is made configurable.



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

Mime
View raw message