有一些時候,多個團隊須要共同完成一個任務,好比,A團隊將Hadoop集羣計算的結果交給B團隊繼續計算,B完成了本身任務再交給C團隊繼續作。這就有點像業務系統的工做流同樣,一環一環地傳下html
去,直到最後一部分完成。在業務系統中,咱們常常會用SOA的架構來解決這種問題,每一個團隊在ESB(企業服務股總線)服務器上部署本身的服務,而後經過消息中間件完成調度任務。對亍分步式的多個java
Hadoop集羣系統的協做,一樣能夠用這種架構來作只要把消息中間件引擎換成支持分步式的消息中間件的引擎就好了。node
本文樓主將使用zookeeper作爲分步式消息中間件構造一個大型超市的部分數據計算模型來完成各個區域利潤計算的業務需求。git
因爲採購和銷售分別是由不一樣廠商進行的軟件開發和維護,並且業務往來也在不一樣的城市和地區。 因此在每個月底結算時,工做量都特別大。 好比,計算利潤表: 當月利潤 = 當月銷售金額 - 當月採購github
額 - 當月其餘支出(樓主只是粗略計算)。若是採購系統是單獨的系統,銷售是另外單獨的系統,及以其餘幾十個大大小小的系統, 如何能讓多個系統,配合起來完成該需求?數據庫
樓主基於zookeeper來構建一個分步式隊列的應用,來解決上面的功能需求。排除了ESB的部分,只保留zookeeper進行實現。apache
咱們設計一個同步隊列,這個隊列有3個條件節點,分別對應採購(purchase),銷售 (sell),其餘費用(other)3個部分。當3個節點都被建立後,程序會自動觸發計算利潤, 幵建立利潤(profit)節點。上面3個節點的建立,無順序要求。每一個節點只能被建立一次 。
服務器
Hadoop mapreduce1,Hadoop mapreduce2 是2個獨立的Hadoop集羣應用。 Java App 是2個獨立的Java應用 。ZooKeeper集羣的有3個節點 。架構
當/qeueu/profit被建立後,利潤java app被啓動,全部zookeeper的鏈接通知同步程序(紅色線),隊列已完成,全部程序結束。app
1)hadoop集羣。樓主用的6個節點的hadoop2.7.3集羣,各位同窗能夠根據本身的實際狀況進行搭建,但至少須要1臺僞分佈式的。(參考http://www.cnblogs.com/qq503665965/p/6790580.html)
2)zookeeper集羣。至少三個節點。安裝參考樓主這篇文章(http://www.cnblogs.com/qq503665965/p/6790580.html)
3)java開發環境。
計算採購金額:
1 package zkqueue; 2 import java.io.IOException; 3 import java.util.HashMap; 4 import java.util.Map; 5 import java.util.regex.Pattern; 6 7 import org.apache.hadoop.fs.Path; 8 import org.apache.hadoop.io.IntWritable; 9 import org.apache.hadoop.io.LongWritable; 10 import org.apache.hadoop.io.Text; 11 import org.apache.hadoop.mapred.JobConf; 12 import org.apache.hadoop.mapreduce.Job; 13 import org.apache.hadoop.mapreduce.Mapper; 14 import org.apache.hadoop.mapreduce.Reducer; 15 import org.apache.hadoop.mapreduce.lib.input.FileInputFormat; 16 import org.apache.hadoop.mapreduce.lib.input.TextInputFormat; 17 import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat; 18 import org.apache.hadoop.mapreduce.lib.output.TextOutputFormat; 19 20 21 /** 22 * 採購金額計算 23 * @author Jon_China 24 * 25 */ 26 public class Purchase { 27 28 public static final String HDFS = "hdfs://192.168.8.101:9000"; 29 public static final Pattern DELIMITER = Pattern.compile("[\t,]"); 30 31 public static class PurchaseMapper extends Mapper<LongWritable, Text, Text, IntWritable> { 32 33 private String month = "2017-01"; 34 private Text k = new Text(month); 35 private IntWritable v = new IntWritable(); 36 private int money = 0; 37 38 public void map(LongWritable key, Text values, Context context) throws IOException, InterruptedException { 39 System.out.println(values.toString()); 40 String[] tokens = DELIMITER.split(values.toString());//拆分源數據 41 if (tokens[3].startsWith(month)) {// 過濾1月份數據 42 money = Integer.parseInt(tokens[1]) * Integer.parseInt(tokens[2]);//計算 43 v.set(money); 44 context.write(k, v); 45 } 46 } 47 } 48 49 public static class PurchaseReducer extends Reducer<Text, IntWritable, Text, IntWritable> { 50 private IntWritable v = new IntWritable(); 51 private int money = 0; 52 53 @Override 54 public void reduce(Text key, Iterable<IntWritable> values, Context context) throws IOException, InterruptedException { 55 for (IntWritable line : values) { 56 money += line.get(); 57 } 58 v.set(money); 59 context.write(null, v); 60 System.out.println("Output:" + key + "," + money); 61 } 62 63 } 64 65 public static void run(Map<String, String> path) throws IOException, InterruptedException, ClassNotFoundException { 66 JobConf conf = config(); 67 String local_data = path.get("purchase"); 68 String input = path.get("input"); 69 String output = path.get("output"); 70 71 72 HdfsDAO hdfs = new HdfsDAO(HDFS, conf); 73 hdfs.rmr(input); 74 hdfs.mkdirs(input); 75 hdfs.copyFile(local_data, input); 76 77 Job job = Job.getInstance(conf); 78 job.setJarByClass(Purchase.class); 79 80 job.setOutputKeyClass(Text.class); 81 job.setOutputValueClass(IntWritable.class); 82 83 job.setMapperClass(PurchaseMapper.class); 84 job.setReducerClass(PurchaseReducer.class); 85 86 job.setInputFormatClass(TextInputFormat.class); 87 job.setOutputFormatClass(TextOutputFormat.class); 88 89 FileInputFormat.setInputPaths(job, new Path(input)); 90 FileOutputFormat.setOutputPath(job, new Path(output)); 91 92 job.waitForCompletion(true); 93 } 94 95 public static JobConf config() { 96 JobConf conf = new JobConf(Purchase.class); 97 conf.setJobName("purchase"); 98 conf.addResource("classpath:/hadoop/core-site.xml"); 99 conf.addResource("classpath:/hadoop/hdfs-site.xml"); 100 conf.addResource("classpath:/hadoop/mapred-site.xml"); 101 conf.addResource("classpath:/hadoop/yarn-site.xml"); 102 return conf; 103 } 104 105 public static Map<String,String> path(){ 106 Map<String, String> path = new HashMap<String, String>(); 107 path.put("purchase", Purchase.class.getClassLoader().getResource("logfile/biz/purchase.csv").getPath());// 源文件數據 108 path.put("input", HDFS + "/user/hdfs/biz/purchase");//hdfs存儲路徑 109 path.put("output", HDFS + "/user/hdfs/biz/purchase/output"); //hdfs輸出路徑 110 return path; 111 } 112 113 public static void main(String[] args) throws Exception { 114 run(path()); 115 } 116 117 }
銷售數據計算:
1 package zkqueue; 2 3 import java.io.IOException; 4 import java.util.HashMap; 5 import java.util.Map; 6 import java.util.regex.Pattern; 7 8 import org.apache.hadoop.fs.Path; 9 import org.apache.hadoop.io.IntWritable; 10 import org.apache.hadoop.io.LongWritable; 11 import org.apache.hadoop.io.Text; 12 import org.apache.hadoop.mapred.JobConf; 13 import org.apache.hadoop.mapreduce.Job; 14 import org.apache.hadoop.mapreduce.Mapper; 15 import org.apache.hadoop.mapreduce.Reducer; 16 import org.apache.hadoop.mapreduce.lib.input.FileInputFormat; 17 import org.apache.hadoop.mapreduce.lib.input.TextInputFormat; 18 import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat; 19 import org.apache.hadoop.mapreduce.lib.output.TextOutputFormat; 20 21 /** 22 * 銷售數據計算 23 * @author Jon_China 24 * 25 */ 26 public class Sell { 27 28 public static final String HDFS = "hdfs://192.168.8.101:9000"; 29 public static final Pattern DELIMITER = Pattern.compile("[\t,]"); 30 31 public static class SellMapper extends Mapper<LongWritable, Text, Text, IntWritable> { 32 33 private String month = "2013-01"; 34 private Text k = new Text(month); 35 private IntWritable v = new IntWritable(); 36 private int money = 0; 37 38 public void map(LongWritable key, Text values, Context context) throws IOException, InterruptedException { 39 System.out.println(values.toString()); 40 String[] tokens = DELIMITER.split(values.toString()); 41 if (tokens[3].startsWith(month)) {// 1月的數據 42 money = Integer.parseInt(tokens[1]) * Integer.parseInt(tokens[2]);//單價*數量 43 v.set(money); 44 context.write(k, v); 45 } 46 } 47 } 48 49 public static class SellReducer extends Reducer<Text, IntWritable, Text, IntWritable> { 50 private IntWritable v = new IntWritable(); 51 private int money = 0; 52 53 @Override 54 public void reduce(Text key, Iterable<IntWritable> values, Context context) throws IOException, InterruptedException { 55 for (IntWritable line : values) { 56 money += line.get(); 57 } 58 v.set(money); 59 context.write(null, v); 60 System.out.println("Output:" + key + "," + money); 61 } 62 63 } 64 65 public static void run(Map<String, String> path) throws IOException, InterruptedException, ClassNotFoundException { 66 JobConf conf = config(); 67 String local_data = path.get("sell"); 68 String input = path.get("input"); 69 String output = path.get("output"); 70 71 // 初始化sell 72 HdfsDAO hdfs = new HdfsDAO(HDFS, conf); 73 hdfs.rmr(input); 74 hdfs.mkdirs(input); 75 hdfs.copyFile(local_data, input); 76 77 Job job = Job.getInstance(conf); 78 job.setJarByClass(Sell.class); 79 80 job.setOutputKeyClass(Text.class); 81 job.setOutputValueClass(IntWritable.class); 82 83 job.setMapperClass(SellMapper.class); 84 job.setReducerClass(SellReducer.class); 85 86 job.setInputFormatClass(TextInputFormat.class); 87 job.setOutputFormatClass(TextOutputFormat.class); 88 89 FileInputFormat.setInputPaths(job, new Path(input)); 90 FileOutputFormat.setOutputPath(job, new Path(output)); 91 92 job.waitForCompletion(true); 93 } 94 95 public static JobConf config() {// Hadoop集羣的遠程配置信息 96 JobConf conf = new JobConf(Purchase.class); 97 conf.setJobName("purchase"); 98 conf.addResource("classpath:/hadoop/core-site.xml"); 99 conf.addResource("classpath:/hadoop/hdfs-site.xml"); 100 conf.addResource("classpath:/hadoop/mapred-site.xml"); 101 conf.addResource("classpath:/hadoop/yarn-site.xml"); 102 return conf; 103 } 104 105 public static Map<String,String> path(){ 106 Map<String, String> path = new HashMap<String, String>(); 107 path.put("sell", Sell.class.getClassLoader().getResource("logfile/biz/sell.csv").getPath());// 本地的數據文件 108 path.put("input", HDFS + "/user/hdfs/biz/sell");// HDFS的目錄 109 path.put("output", HDFS + "/user/hdfs/biz/sell/output"); // 輸出目錄 110 return path; 111 } 112 113 public static void main(String[] args) throws Exception { 114 run(path()); 115 } 116 117 }
其餘金額計算:
1 package zkqueue; 2 3 import java.io.BufferedReader; 4 import java.io.File; 5 import java.io.FileReader; 6 import java.io.IOException; 7 import java.util.regex.Pattern; 8 9 public class Other { 10 11 public static String file = "/logfile/biz/other.csv"; 12 public static final Pattern DELIMITER = Pattern.compile("[\t,]"); 13 private static String month = "2017-01"; 14 15 public static void main(String[] args) throws IOException { 16 calcOther(file); 17 } 18 19 public static int calcOther(String file) throws IOException { 20 int money = 0; 21 BufferedReader br = new BufferedReader(new FileReader(new File(file))); 22 23 String s = null; 24 while ((s = br.readLine()) != null) { 25 String[] tokens = DELIMITER.split(s); 26 if (tokens[0].startsWith(month)) {// 1月的數據 27 money += Integer.parseInt(tokens[1]); 28 } 29 } 30 br.close(); 31 System.out.println("Output:" + month + "," + money); 32 return money; 33 } 34 }
計算利潤:
1 package zkqueue; 2 3 import java.io.IOException; 4 5 6 /** 7 * 利潤計算 8 * @author Jon_China 9 * 10 */ 11 public class Profit { 12 13 public static void main(String[] args) throws Exception { 14 profit(); 15 } 16 17 public static void profit() throws Exception { 18 int sell = getSell(); 19 int purchase = getPurchase(); 20 int other = getOther(); 21 int profit = sell - purchase - other; 22 System.out.printf("profit = sell - purchase - other = %d - %d - %d = %d\n", sell, purchase, other, profit); 23 } 24 25 public static int getPurchase() throws Exception { 26 HdfsDAO hdfs = new HdfsDAO(Purchase.HDFS, Purchase.config()); 27 return Integer.parseInt(hdfs.cat(Purchase.path().get("output") + "/part-r-00000").trim()); 28 } 29 30 public static int getSell() throws Exception { 31 HdfsDAO hdfs = new HdfsDAO(Sell.HDFS, Sell.config()); 32 return Integer.parseInt(hdfs.cat(Sell.path().get("output") + "/part-r-00000").trim()); 33 } 34 35 public static int getOther() throws IOException { 36 return Other.calcOther(Other.file); 37 } 38 39 }
zookeeper任務調度:
1 package zkqueue; 2 3 import java.io.IOException; 4 import java.util.List; 5 6 import org.apache.zookeeper.CreateMode; 7 import org.apache.zookeeper.KeeperException; 8 import org.apache.zookeeper.WatchedEvent; 9 import org.apache.zookeeper.Watcher; 10 import org.apache.zookeeper.ZooDefs.Ids; 11 import org.apache.zookeeper.ZooKeeper; 12 /** 13 * 分佈式隊列zookeeper調度 14 * @author Jon_China 15 * 16 */ 17 public class QueueZookeeper { 18 //設置隊列目錄樹 19 final public static String QUEUE = "/queue"; 20 final public static String PROFIT = "/queue/profit"; 21 final public static String PURCHASE = "/queue/purchase"; 22 final public static String SELL = "/queue/sell"; 23 final public static String OTHER = "/queue/other"; 24 25 public static void main(String[] args) throws Exception { 26 if (args.length == 0) { 27 System.out.println("Please start a task:"); 28 } else { 29 doAction(Integer.parseInt(args[0])); 30 } 31 } 32 public static void doAction(int client) throws Exception { 33 //zookeeper地址 34 String host1 = "192.168.8.104:2181"; 35 String host2 = "192.168.8.105:2181"; 36 String host3 = "192.168.8.106:2181"; 37 38 ZooKeeper zk = null; 39 switch (client) {//1,2,3分別將不一樣任務加入隊列 40 case 1: 41 zk = connection(host1); 42 initQueue(zk); 43 doPurchase(zk); 44 break; 45 case 2: 46 zk = connection(host2); 47 initQueue(zk); 48 doSell(zk); 49 break; 50 case 3: 51 zk = connection(host3); 52 initQueue(zk); 53 doOther(zk); 54 break; 55 } 56 } 57 58 // 建立一個與服務器的鏈接 59 public static ZooKeeper connection(String host) throws IOException { 60 ZooKeeper zk = new ZooKeeper(host, 60000, new Watcher() { 61 // 監控全部被觸發的事件 62 public void process(WatchedEvent event) { 63 if (event.getType() == Event.EventType.NodeCreated && event.getPath().equals(PROFIT)) { 64 System.out.println("Queue has Completed!!!"); 65 } 66 } 67 }); 68 return zk; 69 } 70 /** 71 * 初始化隊列 72 * @param zk 73 * @throws KeeperException 74 * @throws InterruptedException 75 */ 76 public static void initQueue(ZooKeeper zk) throws KeeperException, InterruptedException { 77 System.out.println("WATCH => " + PROFIT); 78 zk.exists(PROFIT, true); 79 80 if (zk.exists(QUEUE, false) == null) { 81 System.out.println("create " + QUEUE); 82 zk.create(QUEUE, QUEUE.getBytes(), Ids.OPEN_ACL_UNSAFE, CreateMode.PERSISTENT); 83 } else { 84 System.out.println(QUEUE + " is exist!"); 85 } 86 } 87 /** 88 * 採購任務 89 * @param zk 90 * @throws Exception 91 */ 92 public static void doPurchase(ZooKeeper zk) throws Exception { 93 if (zk.exists(PURCHASE, false) == null) { 94 95 Purchase.run(Purchase.path()); 96 97 System.out.println("create " + PURCHASE); 98 zk.create(PURCHASE, PURCHASE.getBytes(), Ids.OPEN_ACL_UNSAFE, CreateMode.PERSISTENT); 99 } else { 100 System.out.println(PURCHASE + " is exist!"); 101 } 102 isCompleted(zk); 103 } 104 /** 105 * 銷售任務 106 * @param zk 107 * @throws Exception 108 */ 109 public static void doSell(ZooKeeper zk) throws Exception { 110 if (zk.exists(SELL, false) == null) { 111 112 Sell.run(Sell.path()); 113 114 System.out.println("create " + SELL); 115 zk.create(SELL, SELL.getBytes(), Ids.OPEN_ACL_UNSAFE, CreateMode.PERSISTENT); 116 } else { 117 System.out.println(SELL + " is exist!"); 118 } 119 isCompleted(zk); 120 } 121 /** 122 * 其餘計算任務 123 * @param zk 124 * @throws Exception 125 */ 126 public static void doOther(ZooKeeper zk) throws Exception { 127 if (zk.exists(OTHER, false) == null) { 128 129 Other.calcOther(Other.file); 130 131 System.out.println("create " + OTHER); 132 zk.create(OTHER, OTHER.getBytes(), Ids.OPEN_ACL_UNSAFE, CreateMode.PERSISTENT); 133 } else { 134 System.out.println(OTHER + " is exist!"); 135 } 136 isCompleted(zk); 137 } 138 /** 139 * 檢測完成狀況 140 * @param zk 141 * @throws Exception 142 */ 143 public static void isCompleted(ZooKeeper zk) throws Exception { 144 int size = 3; 145 List<String> children = zk.getChildren(QUEUE, true); 146 int length = children.size(); 147 148 System.out.println("Queue Complete:" + length + "/" + size); 149 if (length >= size) { 150 System.out.println("create " + PROFIT); 151 Profit.profit(); 152 zk.create(PROFIT, PROFIT.getBytes(), Ids.OPEN_ACL_UNSAFE, CreateMode.EPHEMERAL); 153 154 for (String child : children) {// 清空節點 155 zk.delete(QUEUE + "/" + child, -1); 156 } 157 158 } 159 } 160 }
在最後一步,統計其餘費用數據程序運行後,從日誌中看到3個條件節點都已知足要求 。而後,經過同步的分步式隊列自動啓動了計算利潤的程序,幵在日誌中打印了2017 年1月的利潤爲-6693765。
示例代碼地址:https://github.com/LJunChina/hadoop/tree/master/distributed_mq