手动维护kafka的偏移量，各类方式整理总结

对sparkstreaming消费kafka消息进行逻辑判断，逻辑清晰明确，可是offset存在zookeeper当中，须要本身实现其余存储 http://www.javashuo.com/article/p-rqfcsjml-gd.htmlhtml 一样是存储到zookeeper当中，记录博文较多java https://www.cnblogs.com/junneyang/p/8267902.h

>>阅读原文<<