1.修改ZipkinSpanCollector中collect方法，队列满后直接丢弃，不再阻塞等待。
2.修改ZooKeeperSamplingTraceFilter，增加连接zookeeper等待超时时间；增加sampleRate为double处理。
3.对调用做了简单的封装，保证调用服务在zipkin相关服务出现问题时不影响调用服务的正常功能。

//Init zipkin
ZipKinConfig zkConfig = new ZipKinConfig();
zkConfig.setTraceServerIp("127.0.0.1");
zkConfig.setTraceServerPort(8080);
zkConfig.setTraceServerName("ServiceName");
zkConfig.setCollectorServerIp("127.0.0.1");
zkConfig.setCollectorServerPort(9410);
zkConfig.setSpanQueueSize(150);
zkConfig.setZkServerConnectString("127.0.0.1:2181");
zkConfig.setSampleRateNode("/zipkin/config/samplerat");
ZipkinCollectorClient.getInstance().init(zkConfig);

//Begin tracer
ZipkinCollectorClient tracer = ZipkinCollectorClient.getInstance();
tracer.startNewSpan("timeline");
tracer.setClientSent();

//Your code here
tracer.setClientReceived();

//Shutdown
ZipkinCollectorClient.getInstance().shutdown();