MaxCompute：多任務樣本

前提條件

已通過快速入門完成測試所需環境配置。

測試準備

1. 準備好測試程式的JAR包，假設名字為mapreduce-examples.jar，本地存放路徑為MaxCompute用戶端bin目錄下data\resources。

2. 準備好MultiJobs測試表和資源。

   1. 建立測試表。

      CREATE TABLE mr_empty (key STRING, value STRING);
      CREATE TABLE mr_multijobs_out (value BIGINT);

   2. 添加測試資源。

      add table mr_multijobs_out as multijobs_res_table -f;
      
      -- 首次添加忽略-f覆蓋指令。
      add jar data\resources\mapreduce-examples.jar -f;

測試步驟

jar -resources mapreduce-examples.jar,multijobs_res_table -classpath data\resources\mapreduce-examples.jar 
 com.aliyun.odps.mapred.open.example.MultiJobs mr_multijobs_out;

預期結果

+------------+
| value      |
+------------+
| 0          |
+------------+

程式碼範例

Pom依賴資訊，請參見注意事項。

package com.aliyun.odps.mapred.open.example;
import java.io.IOException;
import java.util.Iterator;
import com.aliyun.odps.data.Record;
import com.aliyun.odps.data.TableInfo;
import com.aliyun.odps.mapred.JobClient;
import com.aliyun.odps.mapred.MapperBase;
import com.aliyun.odps.mapred.RunningJob;
import com.aliyun.odps.mapred.TaskContext;
import com.aliyun.odps.mapred.conf.JobConf;
import com.aliyun.odps.mapred.utils.InputUtils;
import com.aliyun.odps.mapred.utils.OutputUtils;
import com.aliyun.odps.mapred.utils.SchemaUtils;
/**
     * MultiJobs
     *
     * Running multiple job
     *
     **/
public class MultiJobs {
    public static class InitMapper extends MapperBase {
        @Override
            public void setup(TaskContext context) throws IOException {
            Record record = context.createOutputRecord();
            long v = context.getJobConf().getLong("multijobs.value", 2);
            record.set(0, v);
            context.write(record);
        }
    }
    public static class DecreaseMapper extends MapperBase {
        @Override
            public void cleanup(TaskContext context) throws IOException {
            /**從JobConf中擷取main函數中定義的變數值。*/
            long expect = context.getJobConf().getLong("multijobs.expect.value", -1);
            long v = -1;
            int count = 0;
            /**讀取資源表裡面的資料，這個表是上一個job的輸出表。*/
            Iterator<Record> iter = context.readResourceTable("multijobs_res_table");
            while (iter.hasNext()) {
                Record r = iter.next();
                v = (Long) r.get(0);
                if (expect != v) {
                    throw new IOException("expect: " + expect + ", but: " + v);
                }
                count++;
            }
            if (count != 1) {
                throw new IOException("res_table should have 1 record, but: " + count);
            }
            Record record = context.createOutputRecord();
            v--;
            record.set(0, v);
            context.write(record);
            /**設定counter，counter在作業成功結束後，可以在main函數中擷取到。*/
            context.getCounter("multijobs", "value").setValue(v);
        }
    }
    public static void main(String[] args) throws Exception {
        if (args.length != 1) {
            System.err.println("Usage: TestMultiJobs <table>");
            System.exit(1);
        }
        String tbl = args[0];
        long iterCount = 2;
        System.err.println("Start to run init job.");
        JobConf initJob = new JobConf();
        initJob.setLong("multijobs.value", iterCount);
        initJob.setMapperClass(InitMapper.class);
        InputUtils.addTable(TableInfo.builder().tableName("mr_empty").build(), initJob);
        OutputUtils.addTable(TableInfo.builder().tableName(tbl).build(), initJob);
        initJob.setMapOutputKeySchema(SchemaUtils.fromString("key:string"));
        initJob.setMapOutputValueSchema(SchemaUtils.fromString("value:string"));
        /**MapOnly作業需要顯式設定reducer的數目為0。*/
        initJob.setNumReduceTasks(0);
        JobClient.runJob(initJob);
        while (true) {
            System.err.println("Start to run iter job, count: " + iterCount);
            JobConf decJob = new JobConf();
            decJob.setLong("multijobs.expect.value", iterCount);
            decJob.setMapperClass(DecreaseMapper.class);
            InputUtils.addTable(TableInfo.builder().tableName("mr_empty").build(), decJob);
            OutputUtils.addTable(TableInfo.builder().tableName(tbl).build(), decJob);
            /**MapOnly作業需要顯式設定reducer的數目為0。*/
            decJob.setNumReduceTasks(0);
            RunningJob rJob = JobClient.runJob(decJob);
            iterCount--;
            /**如果迭代次數已經達到，則退出迴圈。*/
            if (rJob.getCounters().findCounter("multijobs", "value").getValue() == 0) {
                break;
            }
        }
        if (iterCount != 0) {
            throw new IOException("Job failed.");
        }
    }
}