Do Clients Need Access to Servers?

Hello,

I’ve been doing some reading on Sensu and have a question. Do Sensu clients need direct access to servers or just to RabbitMQ (for any comms, including keep alives)? In my scenario the server is on network A, which has access to network B but network B does not have reachback to A. Can I put the clients and rabbitmq on B and run the server on A, or do I need to run everything on B and then use some other method to poll from a B server to A? In this last scenario is it possible to tier servers?

Thanks for any help on this.

Although not exactly explicit in the FAQ:
http://sensuapp.org/docs/0.12/faq
And in the general flow diagram that floats around:


Yes, only RabbitMQ connectivity is required for the clients. You can
have all your stuff in B and have a sensu-server in A that just reads
things off the queue and operates on them. (assuming other components
like redis + api + dashboard go with the server. I guess they could go
in B too if you wanted)

···

On Thu, Jun 26, 2014 at 1:42 PM, Ichabod <klumpba@gmail.com> wrote:

Hello,
    I've been doing some reading on Sensu and have a question. Do Sensu
clients need direct access to servers or just to RabbitMQ (for any comms,
including keep alives)? In my scenario the server is on network A, which
has access to network B but network B does not have reachback to A. Can I
put the clients and rabbitmq on B and run the server on A, or do I need to
run everything on B and then use some other method to poll from a B server
to A? In this last scenario is it possible to tier servers?

Thanks for any help on this.