[
https://issues.apache.org/jira/browse/HADOOP-8982?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Chris Nauroth updated HADOOP-8982:
----------------------------------
Attachment: HADOOP-8982.1.patch
After many, many failed attempts to simulate partial writes on Windows, I'm
submitting a patch that bypasses just the portion of the test that depends on
partial write when running on Windows.
The earlier analysis is correct that the test is not able to observe partial
write on Windows. At first, I thought the behavior was strictly due to
differences in implementation of pipes between Linux and Windows, so I tried
changing the test to use a real network socket. Then, I tried ever-increasing
write buffer sizes. None of my attempts successfully simulated partial write
on Windows. At this point, I think we need to skip this portion of the test on
Windows.
This page lists some interesting differences between Unix and Windows sockets:
http://itamarst.org/writings/win32sockets.html
The last point is particularly relevant.
> TestSocketIOWithTimeout fails on Windows
> ----------------------------------------
>
> Key: HADOOP-8982
> URL: https://issues.apache.org/jira/browse/HADOOP-8982
> Project: Hadoop Common
> Issue Type: Bug
> Components: net
> Affects Versions: 3.0.0
> Reporter: Chris Nauroth
> Assignee: Chris Nauroth
> Attachments: HADOOP-8982.1.patch
>
>
> This is a possible race condition or difference in socket handling on Windows.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
