[rabbitmq-discuss] High availability questions

[Emile Joubert]

Hi Chris,

On 27/11/12 21:54, Chris Toomey wrote:
> 1) Given the clustered server redundancy and mirrored queues, is there
> any reason to still make exchanges/queues/messages durable?  Is it just
> to protect against the case when all nodes in the cluster fail?

You might still want to declare queues and exchanges as durable so that 
they persist broker restarts. Without that clients would need to 
redeclare them after a broker restart.

You might still want to publish messages with delivery mode 2 in case 
all the nodes crash. A common disaster scenario is loss of power to a 
cluster of nodes.

> 2) In order to update the server configuration, it's necessary to
> restart, correct?  If so, what's the best way to accomplish config.
> updates across a cluster while minimizing downtime and loss of redundancy?

If by config you mean exchanges, users, vhosts, queues bindings, etc. 
then a restart is not required. These can be modified while the broker 
is running.

Since v3.0.0 you can also modify runtime parameters and policies while 
the broker is running to reconfigure federation and HA:

http://www.rabbitmq.com/parameters.html

Modifications to the broker configuration file and the broker 
environment will require a restart.

> 3) Same question for upgrading to a newer version of RabbitMQ?

See the notes on upgrading a cluster:

http://www.rabbitmq.com/clustering.html#upgrading



-Emile