qpid-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Martin Ritchie <ritch...@apache.org>
Subject Re: svn commit: r791954 - /qpid/trunk/qpid/java/systests/src/main/java/org/apache/qpid/test/utils/FailoverBaseCase.java
Date Wed, 08 Jul 2009 09:55:04 GMT
2009/7/8 Rafael Schloming <rafaels@redhat.com>:
> Martin Ritchie wrote:
>>
>> 2009/7/8 Rafael Schloming <rafaels@redhat.com>:
>>>
>>> Aidan Skinner wrote:
>>>>
>>>> On Tue, Jul 7, 2009 at 8:45 PM, <rhs@apache.org> wrote:
>>>>
>>>>> Author: rhs
>>>>> Date: Tue Jul  7 19:45:04 2009
>>>>> New Revision: 791954
>>>>>
>>>>> URL: http://svn.apache.org/viewvc?rev=791954&view=rev
>>>>> Log:
>>>>> Reverted 787626 as it was causing the regular profiles to fail.
>>>>
>>>> Uh, it is? They passed fine for me. What problem were you seeing?
>>>
>>> The cleanBroker() ends up deleting the data directory out from underneath
>>> the running broker causing it to die.
>>>
>>> I suspect the problem you're seeing with QPID-1935 is actually a real
>>> broker
>>> bug where it's failing to update the persistent store when messages are
>>> dequeued causing one test to interfere with the next.
>>>
>>> --Rafael
>>
>> If the cleanBoker/delete was done before the super.setUp() call then
>> the broker would not be running so it should't be a problem.
>
> Possibly, but cleanBroker() isn't really meant to be used that way, and it
> invalidates the test anyways. You really shouldn't have to clean for the
> test to pass. As I said, I'm pretty sure if the test is failing when
> persistence is enabled it's pointing to an actual bug.

I don't recall how the cpp broker tests run but the transient java
tests have a fresh broker for each run. So if I put messages on the
queue and test they are they as part of the test then stop the broker
the messages go away. If use persistence then they won't. So the fact
that most of our tests use the queue 'queue' will result in previous
tests affecting the current test. Running clean between each test case
should not cause a failure in the broker as the broker is stopped and
started between each test. As long as the order is stop, clean ,start
then there should be no problem.

Or am I missing something?

Martin

> --Rafael
>
>
> ---------------------------------------------------------------------
> Apache Qpid - AMQP Messaging Implementation
> Project:      http://qpid.apache.org
> Use/Interact: mailto:dev-subscribe@qpid.apache.org
>
>



-- 
Martin Ritchie

---------------------------------------------------------------------
Apache Qpid - AMQP Messaging Implementation
Project:      http://qpid.apache.org
Use/Interact: mailto:dev-subscribe@qpid.apache.org


Mime
View raw message