Mailing-List: contact issues-help@cxf.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@cxf.apache.org
Date: Thu, 21 Apr 2011 22:08:05 +0000 (UTC)
From: "Aki Yoshida (JIRA)" <jira@apache.org>
To: issues@cxf.apache.org
Message-ID: 
 <218068062.74852.1303423685764.JavaMail.tomcat@hel.zones.apache.org>
In-Reply-To: <14862722.36171291900862401.JavaMail.jira@thor>
Subject: [jira] [Issue Comment Edited] (CXF-3179) ServerPersistenceTest fail
 intermittently
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/CXF-3179?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13022929#comment-13022929 ] 

Aki Yoshida edited comment on CXF-3179 at 4/21/11 10:08 PM:
------------------------------------------------------------

Hi,
I also thought the cause to be some timing issue.
But sometimes with a large waiting value, the first part of the test already failed.
So, I looked into it and found another problem.

This test inserts the message loss simulator on the server's outbound chain. That means, among the three greeter replies from the server, the second one (not necessarily the response for TWO, but the the second response that is asynchronously sent out from the server) should get lost. However, as the server is sending out responses concurrently and the message loss simulator's counter isn't synchronized, I observed that sometimes two messages were getting lost. 

When I made the MessageLossSimulator's handleMessage method synchronized, this problem disappeared and the waiting time can be made even shorter.

Regards, Aki

      was (Author: ay):
    Hi,
I also thought the cause to be some timing issue.
But sometimes with a large waiting value, the first part of the test already failed.
So, I looked into it and found another problem.

This test inserts the message loss simulator on the server's outbound chain. That means, among the three greeter replies from the server, the second one (not necessarily the response for TWO, but the the second response that is asynchronously sent out from the server). However, as the server is sending out responses concurrently, the message loss simulator's counter isn't synchronized, I observed that sometimes two messages were getting lost. 

When I made the MessageLossSimulator's handleMessage method synchronized, this problem disappeared and the waiting time can be made even shorter.

Regards, Aki
  
> ServerPersistenceTest fail intermittently
> -----------------------------------------
>
>                 Key: CXF-3179
>                 URL: https://issues.apache.org/jira/browse/CXF-3179
>             Project: CXF
>          Issue Type: Test
>            Reporter: Freeman Fang
>            Assignee: Freeman Fang
>             Fix For: 2.3.2, 2.4
>
>


--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira