[
https://issues.apache.org/jira/browse/HDFS-3738?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Aaron T. Myers updated HDFS-3738:
---------------------------------
Attachment: HDFS-3738.patch
I took a look into this and concluded that the performance of this test is
indeed a regression, inadvertently caused by HDFS-3401. Before that change, the
test was using the broadcast address as its fake address for DNs, which is
resolvable but will return an error immediately upon attempting a connection.
HDFS-3401 changed this to the IP address "1.2.3.4" which actually requires the
socket timeout be hit, which slowed down the test. So, setting the timeout low
(as this patch does) is a perfectly reasonable thing to do, it just so happens
that it wasn't necessary before.
I also checked to see if any of the other tests touched by HDFS-3401 had been
inadvertently slowed down, and found that one had. This updated patch improves
the running time of TestPBHelper from ~11.5 seconds to ~0.5 seconds, which is
similar to the running time before HDFS-3401.
> TestDFSClientRetries#testFailuresArePerOperation sets incorrect timeout config
> ------------------------------------------------------------------------------
>
> Key: HDFS-3738
> URL: https://issues.apache.org/jira/browse/HDFS-3738
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: test
> Affects Versions: 2.0.0-alpha
> Reporter: Aaron T. Myers
> Assignee: Aaron T. Myers
> Priority: Minor
> Attachments: HDFS-3738.patch, HDFS-3738.patch
>
>
> TestDFSClientRetries#testFailuresArePerOperation involves testing retries by
> making use of expected timeouts. However, this test sets the wrong config to
> lower the timeout, and thus takes far longer than it should.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
