[jira] Commented: (HADOOP-3992) Synthetic Load Generator for NameNode testing

["Tsz Wo (Nicholas), SZE (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HADOOP-3992?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12627859#action_12627859 ] 

Tsz Wo (Nicholas), SZE commented on HADOOP-3992:
------------------------------------------------

Quick feedback:
- need license headers
- better use double instead of float for the probabilities
- Some of the USAGE string is too long (e.g. StructureGenerator.USAGE).  It is better to manually divide it into several lines.  Otherwise, it may not easy to parse it if it is wrapped arbitrarily.  For example,
{noformat}
java LoadGenerator -readProbability <read probability> -writeProbabili
ty <write probability> -root <root> -maxDelayBetweenOps <maxDelayBetwe
enOpsInMS> -numOfThreads <numOfThreads> -elapsedTime <elapsedTimeInMin
s>-startTime <startTimeInMS>
{noformat}
- need unit tests?

I will take a deeper look at the codes.

> Synthetic Load Generator for NameNode testing
> ---------------------------------------------
>
>                 Key: HADOOP-3992
>                 URL: https://issues.apache.org/jira/browse/HADOOP-3992
>             Project: Hadoop Core
>          Issue Type: New Feature
>          Components: dfs
>            Reporter: Robert Chansler
>            Assignee: Hairong Kuang
>             Fix For: 0.19.0
>
>         Attachments: loadGenerator.patch, LoadGeneratorUserGuide.pdf
>
>
> For performance and correctness testing. The goal is to faithfully simulate the load of a 5000-node cluster with a (comparatively) tiny number of nodes. 

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.