Mailing-List: contact users-help@activemq.apache.org; run by ezmlm
Precedence: bulk
Reply-To: users@activemq.apache.org
Received-SPF: pass (nike.apache.org: domain of gary.tully@gmail.com designates
 209.85.218.215 as permitted sender)
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=mime-version:in-reply-to:references:date:message-id:subject:from:to
         :content-type;
        b=Z8AIEuXYNNTvrFyWOC+L1KNpHRbu+XUg68ZxuzTqBRt7SUvSX8dWBcZbSryCqxQrWd
         YVjafbppkgBQAskO6qFsqvY/o19HMKB6iu31Uqbr7MKKmDFSqXWibbTLAlUY7tEEB5kX
         AdwxlBOuWEkPBPkgRa9bv14+eDMM6VfjxPWfs=
MIME-Version: 1.0
In-Reply-To: <71e1b5740912110720j345fd147xe41898ad967d9e30@mail.gmail.com>
References: <71e1b5740912110411n51b69310td8ef75cacb358db0@mail.gmail.com>
	 <3a73c17c0912110440kf8e028fqb6b3dde6ec7bd64d@mail.gmail.com>
	 <71e1b5740912110449h358f7a44jda213c5878f748b9@mail.gmail.com>
	 <3a73c17c0912110522p42061cfdk1a4f88ee40600ebd@mail.gmail.com>
	 <71e1b5740912110557o4bffa2eai980fb34ef371b2ff@mail.gmail.com>
	 <3a73c17c0912110653v5932e7du22cc60c8d5ae99cf@mail.gmail.com>
	 <71e1b5740912110720j345fd147xe41898ad967d9e30@mail.gmail.com>
Date: Fri, 11 Dec 2009 15:33:57 +0000
Message-ID: <3a73c17c0912110733q73dd99d5re5b55f434049ba2a@mail.gmail.com>
Subject: Re: Browsing a queue in a distributed broker network sometimes not
	seeing messages
From: Gary Tully <gary.tully@gmail.com>
To: users@activemq.apache.org, antelder@apache.org
Content-Type: multipart/alternative; boundary=0015175d0a3cf89add047a75a62c

--0015175d0a3cf89add047a75a62c
Content-Type: text/plain; charset=ISO-8859-1

2009/12/11 ant elder <antelder@apache.org>

> Ok, i'll have to think about that option. Going back to what you said
> earlier about statically configured destinations on the network
> connector, is it right that the static destination only needs to be
> configured at the remote end of the network connector?


yea. a network connector is one way, unless configured as duplex.


> Eg without
> changing the put code i tried replacing the addNetworkConnector in
> that browse code from the first email with this:
>
>        NetworkConnector nc =
> broker2.addNetworkConnector("multicast://default");
>        ActiveMQDestination destiantion = new ActiveMQQueue("d1");
>        nc.addStaticallyIncludedDestination(destiantion);
>
> and that does then get the first browse working.
>
> Looking at the doc on addStaticallyIncludedDestination at
> http://activemq.apache.org/networks-of-brokers.html near the bottom of
> the page its says "N.B. You can use wildcards in inclusive , exclusive
> destination properties" but trying a destination of "d*" doesn't seem
> to work, are wild cards not applicable in the statically included
> destinations?
>

correct, they are not applicable as it is not possible to know up front what
they would need to expand to.
I tidied up that little NB in the doc to add an 'only' clause


> Even without wild cards this may be enough as i might be able to
> organize the clients to workout the set of destination they're
> interested in before starting the client broker.
>
>   ...ant
>
> On Fri, Dec 11, 2009 at 2:53 PM, Gary Tully <gary.tully@gmail.com> wrote:
> > interesting... this could work but would have some limitations. If you
> use a
> > regular consumer, client ack with prefetch=0 and don't ack, by using a
> > blocking receive(timeout) - the demand will have time to propagate and
> the
> > messages will be visible.
> > However, because the network is store and forward, the message will only
> be
> > visible from one broker at a time and following a forward, only from
> another
> > broker. Currently the network does not allow forwarding back to an
> > originator as this can lead to out of control message bouncing.
> >
> > You will also have to, at some stage, delete the message which may
> require
> > locating the broker that currently holds the message.
> >
> > 2009/12/11 ant elder <antelder@apache.org>
> >
> >> Heh, well its probably not a very valid use case of ActiveMQ - I was
> >> toying with using an ActiveMQ broker network for a sort of simple
> >> distributed storage, with clients able to put a message on a queue and
> >> have other clients read the message nondestructively. None of the
> >> other technologies for that type of thing that I've looked at seemed
> >> do quite what i wanted in a simple way with a good license. The
> >> ActiveMQ network facilities look pretty powerful and easy to use to
> >> setup a distributed network so it would be perfect if there was a way
> >> to get the messages replicated to any client.
> >>
> >>   ...ant
> >>
> >> On Fri, Dec 11, 2009 at 1:22 PM, Gary Tully <gary.tully@gmail.com>
> wrote:
> >> > fraid not. What is your use case?
> >> >
> >> > 2009/12/11 ant elder <antelder@apache.org>
> >> >
> >> >> I see, thanks for the explanation.
> >> >>
> >> >> The statically configured destination isn't really an option as the
> >> >> queue names aren't known till runtime. Are there any other options
> >> >> that might help get the messages forwarded irrespective of demand?
> >> >>
> >> >>   ...ant
> >> >>
> >> >> On Fri, Dec 11, 2009 at 12:40 PM, Gary Tully <gary.tully@gmail.com>
> >> wrote:
> >> >> > This is expected. A queue browser takes a shapshot of the queue
> >> through a
> >> >> > short lived consumer. The network bridge is by default a demand
> >> >> forwarder,
> >> >> > such that messages are only forwarded when there are consumers on
> >> other
> >> >> > brokers.
> >> >> >
> >> >> > With the browser, the consumer (and hence demand) is transient, so
> it
> >> may
> >> >> > take a snapshot before the forwarding has a chance to kick in. And
> the
> >> >> > forwarding stops when the consumer goes away once the browse
> shapshot
> >> >> > completes.
> >> >> >
> >> >> > One way to make this more deterministic would be to have a
> statically
> >> >> > configured destination in the network connector configuration such
> >> that
> >> >> > messages for that destination are forwarded irrespective of demand.
> >> >> >
> >> >> > 2009/12/11 ant elder <ant.elder@gmail.com>
> >> >> >
> >> >> >> I'm trying to track down an problem where a queue browser on a VM
> >> >> >> connection to one AMQ broker is sometimes not seeing messages put
> on
> >> a
> >> >> >> queue from a VM connection to another broker in the same network.
> The
> >> >> >> code below can recreate the problem, running the put() on JVM and
> the
> >> >> >> browse() on another JVM the first QueueBrowser never sees the
> message
> >> >> >> but the second QueueBrowser does though sometimes it doesn't
> without
> >> >> >> the Thread.sleep for a few seconds.
> >> >> >>
> >> >> >> Is there anything obviously wrong with this setup or are there any
> >> >> >> config settings that could help?
> >> >> >>
> >> >> >>   ...ant
> >> >> >>
> >> >> >> The code to put a message on a queue:
> >> >> >>
> >> >> >>    public void put() throws Exception {
> >> >> >>        BrokerService broker1 = new BrokerService();
> >> >> >>        broker1.setBrokerName("default");
> >> >> >>        broker1.setPersistent(false);
> >> >> >>        broker1.setUseJmx(false);
> >> >> >>        TransportConnector tc =
> >> >> broker1.addConnector("tcp://localhost:0");
> >> >> >>        tc.setDiscoveryUri(URI.create("multicast://default"));
> >> >> >>        broker1.addNetworkConnector("multicast://default");
> >> >> >>        broker1.start();
> >> >> >>
> >> >> >>        Connection conn1 = new
> >> >> >>
> >> >>
> >>
> ActiveMQConnectionFactory("vm://default?create=false").createConnection();
> >> >> >>        conn1.start();
> >> >> >>        Session sess1 = conn1.createSession(false, 1);
> >> >> >>        Queue d1 = sess1.createQueue("d1");
> >> >> >>        MessageProducer p1 = sess1.createProducer(d1);
> >> >> >>        TextMessage m1 = sess1.createTextMessage("test");
> >> >> >>        p1.send(m1);
> >> >> >>   }
> >> >> >>
> >> >> >>
> >> >> >> The code to browse the messages on the queue:
> >> >> >>
> >> >> >>    public void browse() throws Exception {
> >> >> >>
> >> >> >>        BrokerService broker2 = new BrokerService();
> >> >> >>        broker2.setBrokerName("default");
> >> >> >>        broker2.setPersistent(false);
> >> >> >>        broker2.setUseJmx(false);
> >> >> >>        TransportConnector tc =
> >> >> broker2.addConnector("tcp://localhost:0");
> >> >> >>        tc.setDiscoveryUri(URI.create("multicast://default"));
> >> >> >>        broker2.addNetworkConnector("multicast://default");
> >> >> >>        broker2.start();
> >> >> >>
> >> >> >>        Connection conn2 = new
> >> >> >>
> >> >>
> >>
> ActiveMQConnectionFactory("vm://default?create=false").createConnection();
> >> >> >>        conn2.start();
> >> >> >>        Session sess2 = conn2.createSession(false, 1);
> >> >> >>        Queue d1 = sess2.createQueue("d1");
> >> >> >>        QueueBrowser b = sess2.createBrowser(d1);
> >> >> >>
> >> >> >>        Enumeration x2 = b.getEnumeration();
> >> >> >>        while (x2.hasMoreElements()) {
> >> >> >>            System.out.println(x2.nextElement()); // this never
> gets
> >> >> >> the message
> >> >> >>        }
> >> >> >>
> >> >> >>        Thread.sleep(5000);
> >> >> >>
> >> >> >>        b = sess2.createBrowser(d1);
> >> >> >>        Enumeration x2a = b.getEnumeration();
> >> >> >>        while (x2a.hasMoreElements()) {
> >> >> >>            System.out.println(x2a.nextElement()); // this does get
> >> the
> >> >> >> message
> >> >> >>        }
> >> >> >>
> >> >> >>    }
> >> >> >>
> >> >> >
> >> >> >
> >> >> >
> >> >> > --
> >> >> > http://blog.garytully.com
> >> >> >
> >> >> > Open Source Integration
> >> >> > http://fusesource.com
> >> >> >
> >> >>
> >> >
> >> >
> >> >
> >> > --
> >> > http://blog.garytully.com
> >> >
> >> > Open Source Integration
> >> > http://fusesource.com
> >> >
> >>
> >
> >
> >
> > --
> > http://blog.garytully.com
> >
> > Open Source Integration
> > http://fusesource.com
> >
>


-- 
http://blog.garytully.com

Open Source Integration
http://fusesource.com

--0015175d0a3cf89add047a75a62c--