[
https://issues.apache.org/jira/browse/HADOOP-5358?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jakob Homan updated HADOOP-5358:
--------------------------------
Attachment: HADOOP-5358.patch
Patch:
+ Provides the option to supply the name of a script file to the load
generator. A script is a text file (with lines beginning with # as comments)
of the form (for example):
{noformat}# duration in secods read probability write probability
10 .22 .44
12 .5 .25
{noformat}
which would have the load generator run for 10 seconds using a .22 probability
of reads, and .44 probability of writes. After 10 seconds, it would run for 12
seconds using a .5 and .25 probability of reads and writes, respectively.
+ Other operation of the load generator is not affected
+ The load generator unit test is updated to test the scripting ability
> Provide scripting functionality to the synthetic load generator
> ---------------------------------------------------------------
>
> Key: HADOOP-5358
> URL: https://issues.apache.org/jira/browse/HADOOP-5358
> Project: Hadoop Core
> Issue Type: Improvement
> Components: test
> Reporter: Jakob Homan
> Assignee: Jakob Homan
> Priority: Minor
> Attachments: HADOOP-5358.patch
>
>
> Currently the load generator accepts parameters at start time as to the read
> and write probabilities to apply against the namenode. It would be useful to
> be able to provide it with a script of these values, so that they can be
> varied over time. This would allow the namenode to be tested with varying
> loads over time so its behavior to changing loads can be examined.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
