Mailing-List: contact dev-help@geronimo.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@geronimo.apache.org
Received-SPF: neutral (hermes.apache.org: local policy)
Message-ID: <41446E55.5040902@gluecode.com>
Date: Sun, 12 Sep 2004 08:42:13 -0700
From: Jeremy Boynes <jboynes@gluecode.com>
User-Agent: Mozilla Thunderbird 0.7.1 (Windows/20040626)
MIME-Version: 1.0
To: dev@geronimo.apache.org
Subject: Re: Deployment heirarchies
References: <FF7CE36846E01B439945B55FFB1DF5BEF77A@mail.toolazydogs.com>
In-Reply-To: <FF7CE36846E01B439945B55FFB1DF5BEF77A@mail.toolazydogs.com>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit

Alan D. Cabrera wrote:

> 
> The reason that I ask is that in itests I distribute and start app1A,
> app1B, and app2A.  I then try to stop and undeploy them in reverse
> order.  When I do this, app1A, app1B and app2A throw this exception when
> they are undeployed:
> 
> java.lang.IllegalStateException
>         at
> org.apache.geronimo.kernel.Kernel.stopConfiguration(Kernel.java:327)
> ...
> 

That looks like a bug.

> Does what I'm doing for these unit tests, undeploying them in reverse
> order, even make sense?  
> 

If they are peers, then you should be able to stop them in any order. If 
they form a hierarchy, then stopping just the parent would be sufficient 
as it would stop all the children as well.

> BTW, what configuration to I pass to tell the server to stop the whole
> tree all at once?  Do I have to stop the nodes that I started in reverse
> order?
> 

Stopping a configuration will stop all of its children first. So, from 
the tree above, stopping "Config1" will also stop "app1A" and "app2A". 
You could kill the whole tree by stopping "System" but that would stop 
all the management and communication GBeans leaving you with a zombied 
kernel.

For stopping groups of applications the intention was to have 
application defined units (like "Config1") so address this.

--
Jeremy