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 torsten@fusesource.com
 designates 74.125.245.78 as permitted sender)
Content-Type: text/plain; charset=us-ascii
Mime-Version: 1.0 (Apple Message framework v1084)
Subject: Re: Ordering of messages with a network of brokers ...
From: Torsten Mielke <torsten@fusesource.com>
In-Reply-To: 
 <CA+GjePVXTVJZEeRrvu4rY2ZDxnZmew4=8WSJBTHFmfc+L3C9-g@mail.gmail.com>
Date: Fri, 16 Mar 2012 10:15:52 +0100
Content-Transfer-Encoding: quoted-printable
Message-Id: <699408C8-E48D-4BFF-810B-D35A2D84C7CB@fusesource.com>
References: 
 <CA+GjePVXTVJZEeRrvu4rY2ZDxnZmew4=8WSJBTHFmfc+L3C9-g@mail.gmail.com>
To: users@activemq.apache.org

Hello Greg,

Both statements are correct. The first link refers to running a single =
broker instance. In that case messages will be dispatched in order to =
all topic consumers and in round-robin fashion to all queue consumers.

Your second quote refers to the fact that when running broker instances =
in a broker network (each node in the network runs a single broker, not =
a master/slave pair) messages may be trapped in a broker if that broker =
goes down and hence potentially cannot be delivered in order (i.e. if =
another producer connected to a different broker in the network =
continues to produce msgs on the same destination, these msgs will be =
dispatched). The messages on the crashed broker can only be delivered =
whenever that crashed broker is restarted. This could potentially take =
some time, depending on how quickly you can recover from that crash.=20
One possible solution is to create a broker network with master/slave =
pairs on each node. In that case the slave broker will be able to take =
over after a master broker crash within a few seconds. As that failover =
still takes a couple of seconds, you still have the potential that for =
that period of time messages may be delivered out of order.=20


> or with a network of brokers will the other
> brokers store up the messages from the producer until the failed
> broker recovers, thus preserving the order during delivery?


No, that will not happen. Other brokers will continue to accept and =
dispatch messages and hence the order of messages being dispatched may =
not be preserved.=20


Hope this helps.

Torsten Mielke
torsten@fusesource.com
tmielke@blogspot.com

On Mar 15, 2012, at 6:08 PM, Greg Thomas wrote:

> I note at =
http://activemq.apache.org/how-do-i-preserve-order-of-messages.html
> that is says "ActiveMQ will preserve the order of messages sent by a
> single producer to all consumers on a topic".
>=20
> However, at http://activemq.apache.org/clustering.html it says:
>=20
> The problem with running lots of stand alone brokers or brokers in a
> network is that messages are owned by a single physical broker at any
> point in time. If that broker goes down, you have to wait for it to be
> restarted before the message can be delivered.
>=20
> Now does that mean that the FAQ in the first link only applies to
> non-clustered solutions; or with a network of brokers will the other
> brokers store up the messages from the producer until the failed
> broker recovers, thus preserving the order during delivery?
>=20
> Thanks,
>=20
> Greg