You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@qpid.apache.org by Aidan Skinner <ai...@gmail.com> on 2009/07/08 14:52:51 UTC

Multiple broker control in QpidTestCase (was Re: svn commit: r791954 - /qpid/trunk/qpid/java/systests/src/main/java/org/apache/qpid/test/utils/FailoverBaseCase.java)

On Wed, Jul 8, 2009 at 12:57 PM, Rafael Schloming<ra...@redhat.com> wrote:
> Martin Ritchie wrote:

>> Agreed, Though I think cleanBroker() needs to be improved as it would
>> be good to have the ablity to delete a data set from one of the
>> non-running brokers (Meaning we have other brokers running in the
>> test). I was thinking of a cluster test scenario where you stop broker
>> B and need to ensure it has no current data before you start it up
>> again and veify that it regains the cluster state. Not that we have
>> clustering in the Java code yet, but given that we can stop and start
>> any number of brokers via our tests we should think about how we can
>> clear a broker's data directories.
>
> I do agree it would be nice to have a base test case that provides tests
> with explicit control over starting, stoping, and cleaning brokers. The
> current cleanBroker() really wasn't intended to be used outside of the way
> it currently is inside of QpidTestCase. (In fact it should probably be
> private.)

Which nicely brings us back to the thing I was trying to address when
I started making these changes in the first place - Failover tests
need to be able to start and stop two brokers with stores.
PeristentStoreTest also needs to do something similar.

I'm not sure how I feel about the cleaning the store between each run
thing, having options for both is probably handy.

- Aidan

-- 
Apache Qpid - AMQP, JMS, other messaging love http://qpid.apache.org
"A witty saying proves nothing" - Voltaire

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

Re: Multiple broker control in QpidTestCase (was Re: svn commit: r791954 - /qpid/trunk/qpid/java/systests/src/main/java/org/apache/qpid/test/utils/FailoverBaseCase.java)

Posted by Gordon Sim <gs...@redhat.com>.

Aidan Skinner wrote:
> On Thu, Jul 9, 2009 at 9:03 AM, Gordon Sim<gs...@redhat.com> wrote:
> 
>> Rafael Schloming wrote:
>>> I don't know about this change, but it does seem like the configuration of
>>> persistent resources for the java broker needs some thinking through.
>>> Probably the cpp broker as well, last I checked it automatically creates a
>>> data-dir in your home directory if you don't specify one, and IMHO that's
>>> not particularly good form.
>> Do you think there should be a different default location (e.g. current
>> dir)? or that the user should be forced to specify a data-dir (or specify
>> --no-data-dir)?
> 
> /var/spool or maybe /var/lib would seem sensible choices on *nix to
> me. I have no idea where you'd put that sort of thing on Windows. I
> think defaulting to $HOME or cwd is a bit weird. Something to allow as
> an option at build time?

There is an init script (cpp/etc/qpidd) that sets data-dir to 
/var/lib/qpidd and that would be the common practice for an installed 
service. The defaulting to home or cwd is mainly for use by users 
without write permission to those directories.

I don't think having /var/lib/... as a built in default is in general 
the right thing. It could always be a build option, but I'm not at all 
convinced anyone would use that rather than the runtime option.

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

Re: Multiple broker control in QpidTestCase (was Re: svn commit: r791954 - /qpid/trunk/qpid/java/systests/src/main/java/org/apache/qpid/test/utils/FailoverBaseCase.java)

Posted by Aidan Skinner <ai...@gmail.com>.

On Thu, Jul 9, 2009 at 9:03 AM, Gordon Sim<gs...@redhat.com> wrote:

> Rafael Schloming wrote:
>>
>> I don't know about this change, but it does seem like the configuration of
>> persistent resources for the java broker needs some thinking through.
>> Probably the cpp broker as well, last I checked it automatically creates a
>> data-dir in your home directory if you don't specify one, and IMHO that's
>> not particularly good form.
>
> Do you think there should be a different default location (e.g. current
> dir)? or that the user should be forced to specify a data-dir (or specify
> --no-data-dir)?

/var/spool or maybe /var/lib would seem sensible choices on *nix to
me. I have no idea where you'd put that sort of thing on Windows. I
think defaulting to $HOME or cwd is a bit weird. Something to allow as
an option at build time?

- Aidan
-- 
Apache Qpid - AMQP, JMS, other messaging love http://qpid.apache.org
"A witty saying proves nothing" - Voltaire

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