Hi All I'm looking at ZeroMQ for a particular application and it looks terrific. However, I have some quick (newbie) questions:
My app is a publish/subscribe client server app. Basically the master process operates on Computer 1 and every so often (asynchronously) sends out information to some number (say 1-100) slave processes each running on their own computers, Computer 2, 3,4 etc. This looks like a standard Pub Sub model where one publisher has to publish new data to some number of data subscribers. The publishing frequency is not that high, say a few times a day, again at unpredictable times, but the subscribers must receive and react to this information as quickly as possible. Think online stock trading app and you get the idea. I understand that each client/slave must connect to the master at the known master's IP address/port (eg tcp://123.101.55.27:5555). But is it also the case that the master/server must also bind to the client at the client's IP address/port? Suppose the client computer is someone's home PC behind a firewall on a DSL connection without a fixed IP. How can I set up and use ZeroMQ such that a client can just register an interest in receiving messages from a server but the server does not have to always know the IP address of every client that issues such requests? Or is this even possible - perhaps to use ZeroMQ both sides must always know the IP address of the other side? If this app runs in real time continuously and perpetually, must I always leave this socket open? How can the receiving socket (the slave/client) tell if the socket connection has died for whatever reason and try to establish a new one? Any suggestions would be welcome. Thanks. ---------------------------------------------------------------------------- ------ Marc Ilgen [email protected]
_______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
