[rabbitmq-discuss] active/active HA and publishers/consumers

• Previous message: [rabbitmq-discuss] Clustering and scaling
• Next message: [rabbitmq-discuss] active/active HA and publishers/consumers
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

What is the recommended way for consumers/publishers to use mirrored
queues for high availability? Should they be aware of all the nodes
that a queue is mirrored to and handle reconnecting to another node in
a failure scenario?

My thought was that I could setup a 2 rabbit node cluster using
mirrored queues on all and use a layer 4 load balancer with only the
master active to make the failure of a node transparent to publishers/
consumers. Is this a valid use case for mirrored queues and if so what
is the best indicator of which node is the master that can be used by
the load balancer? ie. a simple port check to determine rabbit is
running or a specific HTTP API call?

Thanks,
John

• Previous message: [rabbitmq-discuss] Clustering and scaling
• Next message: [rabbitmq-discuss] active/active HA and publishers/consumers
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]