大数据入门第十六天——流式计算之storm详解(三)集群相关进阶
一、集群提交任务流程分析
1.集群提交操作
参考:https://www.jianshu.com/p/6783f1ec2da0
2.任务分配与启动流程
参考:https://www.cnblogs.com/heitaok/p/5531535.html
二、相关目录树
1.组件本地目录树
2.storm zk目录树
三、集群通信
Worker间的通信经常需要通过网络跨节点进行,Storm使用ZeroMQ或Netty(0.9以后默认使用)作为进程间通信的消息框架。
Worker进程内部通信:不同worker的thread通信使用LMAX Disruptor来完成。
不同topologey之间的通信,Storm不负责,需要自己想办法实现,例如使用kafka等;
通信图解:
相关博文参考:http://blog.csdn.net/bbaiggey/article/details/55510010?locationNum=10&fps=1
四、消息容错机制ack-fail
1.概述
l 在storm中,可靠的信息处理机制是从spout开始的。
l 一个提供了可靠的处理机制的spout需要记录他发射出去的tuple,当下游bolt处理tuple或者子tuple失败时spout能够重新发射。
l Storm通过调用Spout的nextTuple()发送一个tuple。为实现可靠的消息处理,首先要给每个发出的tuple带上唯一的ID,并且将ID作为参数传递给 SoputOutputCollector的emit()方法:collector.emit(new Values("value1","value2"), msgId); messageid就是用来标示唯一的tuple的,而rootid是随机生成的
给每个tuple指定ID告诉Storm系统,无论处理成功还是失败,spout都要接收tuple树上所有节点返回的通知。
如果处理成功,spout的ack()方法将会对编号是 msgId的消息应答确认;如果处理失败或者超时,会调用fail()方法。
2.基本实现
Storm 系统中有一组叫做"acker"的特殊的任务,它们负责跟踪DAG(有向无环图)中的每个消息。
acker任务保存了spout id到一对值的映射。第一个值就是spout的任务id,通过这个id,acker就知道消息处理完成时该通知哪个spout任务。第二个值是一个64bit的数字,我们称之为"ack val", 它是树中所有消息的随机id的异或计算结果。
ack val表示了整棵树的的状态,无论这棵树多大,只需要这个固定大小的数字就可以跟踪整棵树。当消息被创建和被应答的时候都会有相同的消息id发送过来做异或。 每当acker发现一棵树的ack val值为0的时候,它就知道这棵树已经被完全处理了
要实现ack机制:
1,spout发射tuple的时候指定messageId
2,spout要重写BaseRichSpout的fail和ack方法
3,spout对发射的tuple进行缓存(否则spout的fail方法收到acker发来的messsageId,
spout也无法获取到发送失败的数据进行重发),看看系统提供的接口,
只有msgId这个参数,这里的设计不合理,其实在系统里是有cache整个msg的,只给用户一个messageid,
用户如何取得原来的msg貌似需要自己cache,然后用这个msgId去查询,太坑爹了
3,spout根据messageId对于ack的tuple则从缓存队列中删除,对于fail的tuple可以选择重发。
4,设置acker数至少大于0;Config.setNumAckers(conf, ackerParal);
3.代码示例
package ackfail; import backtype.storm.Config; import backtype.storm.LocalCluster; import backtype.storm.StormSubmitter; import backtype.storm.topology.TopologyBuilder; /** * Created by maoxiangyi on 2016/4/25. */ public class MyAckFailTopology { public static void main(String[] args) throws Exception { TopologyBuilder topologyBuilder = new TopologyBuilder(); topologyBuilder.setSpout("mySpout", new MySpout(), 1); topologyBuilder.setBolt("mybolt1", new MyBolt1(), 1).shuffleGrouping("mySpout"); Config conf = new Config(); String name = MyAckFailTopology.class.getSimpleName(); if (args != null && args.length > 0) { String nimbus = args[0]; conf.put(Config.NIMBUS_HOST, nimbus); conf.setNumWorkers(1); StormSubmitter.submitTopologyWithProgressBar(name, conf, topologyBuilder.createTopology()); } else { LocalCluster cluster = new LocalCluster(); cluster.submitTopology(name, conf, topologyBuilder.createTopology()); Thread.sleep(60 * 60 * 1000); cluster.shutdown(); } } }
package ackfail; import backtype.storm.spout.SpoutOutputCollector; import backtype.storm.task.TopologyContext; import backtype.storm.topology.IControlSpout; import backtype.storm.topology.IRichSpout; import backtype.storm.topology.OutputFieldsDeclarer; import backtype.storm.topology.base.BaseRichSpout; import backtype.storm.tuple.Fields; import backtype.storm.tuple.Values; import java.util.HashMap; import java.util.Map; import java.util.Random; import java.util.UUID; /** * Created by maoxiangyi on 2016/4/25. */ public class MySpout extends BaseRichSpout { private SpoutOutputCollector collector; private Random rand; private Map<String,Values> buffer = new HashMap<>(); @Override public void declareOutputFields(OutputFieldsDeclarer declarer) { declarer.declare(new Fields("sentence")); rand = new Random(); } @Override public void open(Map conf, TopologyContext context, SpoutOutputCollector collector) { this.collector = collector; } @Override public void nextTuple() { String[] sentences = new String[]{"the cow jumped over the moon", "the cow jumped over the moon", "the cow jumped over the moon", "the cow jumped over the moon", "the cow jumped over the moon"}; String sentence = sentences[rand.nextInt(sentences.length)]; String messageId = UUID.randomUUID().toString().replace("-", ""); Values tuple = new Values(sentence); collector.emit(tuple, messageId); buffer.put(messageId,tuple); try { Thread.sleep(20000); } catch (InterruptedException e) { e.printStackTrace(); } } @Override public void ack(Object msgId) { System.out.println("消息处理成功,id= " + msgId); buffer.remove(msgId); } @Override public void fail(Object msgId) { System.out.println("消息处理失败,id= " + msgId); Values tuple = buffer.get(msgId); collector.emit(tuple,msgId); } }
package ackfail; import backtype.storm.task.OutputCollector; import backtype.storm.task.TopologyContext; import backtype.storm.topology.*; import backtype.storm.topology.base.BaseBasicBolt; import backtype.storm.topology.base.BaseRichBolt; import backtype.storm.tuple.Fields; import backtype.storm.tuple.Tuple; import backtype.storm.tuple.Values; import java.util.Map; /** * Created by maoxiangyi on 2016/4/25. */ public class MyBolt1 extends BaseRichBolt { private OutputCollector collector; @Override public void prepare(Map stormConf, TopologyContext context, OutputCollector collector) { this.collector = collector; } @Override public void execute(Tuple input) { String sentence = input.getString(0); String[] words = sentence.split(" "); for (String word : words) { word = word.trim(); if (!word.isEmpty()) { word = word.toLowerCase(); collector.emit(input, new Values(word)); } } collector.ack(input); } @Override public void declareOutputFields(OutputFieldsDeclarer declarer) { declarer.declare(new Fields("word")); } }
4.参考阅读