zeroMQ初体验-23.可靠性-懒惰的海盗模式
相较于通常的阻塞模式，这里只是做了一点简单的动作来加强系统的可靠性（不再是通常性质的阻塞了）。
这种模式增加/变更了以下几点：
轮询已经发出的请求直到得到响应在一定时间内未得到相应则重新发出请求在一定次数的重试不成功之后，停止请求
相较于传统的阻塞模式，好处显而易见：在未得到答复时可以继续发出请求而不是碰到预料之外的报错之类信息。（zeromq还默默地为你做了这么件事：当链路出现问题，它会悄悄的重新建立一个～）
client:
require?'rubygems'require?'zmq'class?LPClient??def?initialize(connect,?retries?=?nil,?timeout?=?nil)????@connect?=?connect????@retries?=?(retries?||?3).to_i????@timeout?=?(timeout?||?3).to_i????@ctx?=?ZMQ::Context.new(1)????client_sock????at_exit?do??????@socket.close????end??end????def?client_sock????@socket?=?@ctx.socket(ZMQ::REQ)????@socket.setsockopt(ZMQ::LINGER,?0)????@socket.connect(@connect)??end??def?send(message)????@retries.times?do?|tries|??????raise("Send:?#{message}?failed")?unless?@socket.send(message)??????if?ZMQ.select(?[@socket],?nil,?nil,?@timeout)????????yield?@socket.recv????????return??????else????????@socket.close????????client_sock??????end????end????raise?'Server?down'??end??????endif?$0?==?FILE????server?=?LPClient.new(ARGV[0]?||?"tcp://localhost:5555",?ARGV[1],?ARGV[2])????count?=?0????loop?do??????request?=?"#{count}"??????count?+=?1??????server.send(request)?do?|reply|????????if?reply?==?request??????????puts("I:?server?replied?OK?(#{reply})")????????else??????????puts("E:?malformed?reply?from?server:?#{reply}")????????end??????end????end????puts?'success'endrequire?'rubygems'require?'zmq'class?LPServer??def?initialize(connect)????@ctx?=?ZMQ::Context.new(1)????@socket?=?@ctx.socket(ZMQ::REP)????@socket.bind(connect)??end??def?run????begin??????loop?do????????rsl?=?yield?@socket.recv????????@socket.send?rsl??????end????ensure??????@socket.close??????@ctx.close????end??end??????endif?$0?==?FILE??cycles?=?0??srand??LPServer.new(ARGV[0]?||?"tcp://*:5555").run?do?|request|????cycles?+=?1????if?cycles?>?3??????if?rand(3)?==?0????????puts?"I:?simulating?a?crash"????????break??????elsif?rand(3)?==?0????????puts?"I:?simulating?CPU?overload"????????sleep(3)??????end????end????puts?"I:?normal?request?(#{request})"????sleep(1)????request??end????end