112.HBase Endpoint类型的Coprocessor开发与部署

• 下载Protobuf2.5.0版本的安装包：

https://github.com/google/protobuf/releases/download/v2.5.0/protobuf-2.5.0.tar.gz

• 选择一台服务器安装Protobuf

[root@ip-186-31-5-38 ~]# wget https://github.com/google/protobuf/releases/download/v2.5.0/protobuf-2.5.0.tar.gz

• 安装Protobuf所需要的依赖包

yum install -y autoconf automake libtool curl make g++ unzip gcc-c++

• 解压protobuf-2.5.0.tar.gz

[root@ip-186-31-5-38 ~]# tar -zxvf protobuf-2.5.0.tar.gz
[root@ip-186-31-5-38 ~]# cd protobuf-2.5.0
[root@ip-186-31-5-38 protobuf-2.5.0]# ./configure --prefix=/usr/local/protobuf
[root@ip-186-31-5-38 protobuf-2.5.0]# make && make install

• Protobuf环境变量配置

export PROTOBUF_HOME=/usr/local/protobuf
export PATH=$PROTOBUF_HOME/bin:$PATH

• 环境变量生效

[root@ip-186-31-5-38 protobuf-2.5.0]# source /etc/profile

• HBase测试表

create 'fayson_coprocessor', {NAME => 'info'}
put 'fayson_coprocessor','001','info:sales',12.3
put 'fayson_coprocessor','002','info:sales',24.5
put 'fayson_coprocessor','003','info:sales',10.5
put 'fayson_coprocessor','004','info:sales',11.5
put 'fayson_coprocessor','005','info:sales',10.5
put 'fayson_coprocessor','001','info:age',22
put 'fayson_coprocessor','002','info:age',33
put 'fayson_coprocessor','003','info:age',26
put 'fayson_coprocessor','004','info:age',28
put 'fayson_coprocessor','005','info:age',56

• MyFirstCoprocessor.proto文件

[root@ip-186-31-5-171 hbase-coprocessor]# vim MyFirstCoprocessor.proto 
syntax = "proto2";
option java_package = "com.cloudera.hbase.coprocessor.server";
option java_outer_classname = "MyFirstCoprocessor";
option java_generic_services = true;
option java_generate_equals_and_hash = true;
option optimize_for = SPEED;
message MyCoprocessRequest {
    required string family = 1;
    required string columns = 2;
}
message MyCoprocessResponse {
    required int64 count = 1;
    required double maxnum = 3;
    required double minnum = 4;
    required double sumnum = 5;
}
service AggregationService {
  rpc getAggregation(MyCoprocessRequest)
    returns (MyCoprocessResponse);
}

• 生成Java类

[root@ip-186-31-5-38 hbase-coprocessor]# protoc --java_out=./ MyFirstCoprocessor.proto 
[root@ip-186-31-5-38 hbase-coprocessor]# ll
total 4
drwxr-xr-x 3 root root  22 May 14 16:34 com
-rw-r--r-- 1 root root 609 May 14 16:33 MyFirstCoprocessor.proto
[root@ip-186-31-5-38 hbase-coprocessor]# 

• pom.xml文件内容

    org.apache.hadoop
    hadoop-client
    2.6.0-cdh5.11.2


    org.apache.hadoop
    hadoop-common
    2.6.0-cdh5.11.2


    org.apache.hbase
    hbase-client
    1.2.0-cdh5.11.2


    org.apache.hbase
    hbase-examples
    1.2.0-cdh5.11.2


    com.google.protobuf
    protobuf-java
    2.5.0

• Protobuf生成的java类拷贝至指定的包目录下与MyFirstCoprocessor.proto文件指定的java_package包目录一致
• MyFirstCoprocessorEndpoint实现类

package com.cloudera.hbase.coprocessor.server;
import com.google.protobuf.RpcCallback;
import com.google.protobuf.RpcController;
import com.google.protobuf.Service;
import org.apache.commons.collections.map.HashedMap;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.hbase.Cell;
import org.apache.hadoop.hbase.CellUtil;
import org.apache.hadoop.hbase.Coprocessor;
import org.apache.hadoop.hbase.CoprocessorEnvironment;
import org.apache.hadoop.hbase.client.Scan;
import org.apache.hadoop.hbase.coprocessor.CoprocessorException;
import org.apache.hadoop.hbase.coprocessor.CoprocessorService;
import org.apache.hadoop.hbase.coprocessor.RegionCoprocessorEnvironment;
import org.apache.hadoop.hbase.protobuf.ResponseConverter;
import org.apache.hadoop.hbase.regionserver.InternalScanner;
import org.apache.hadoop.hbase.util.Bytes;
import java.io.IOException;
import java.util.ArrayList;
import java.util.List;
import java.util.Map;

public class MyFirstCoprocessorEndPoint extends MyFirstCoprocessor.AggregationService implements Coprocessor, CoprocessorService {
    protected static final Log log = LogFactory.getLog(MyFirstCoprocessorEndPoint.class);
    private RegionCoprocessorEnvironment env;
    @Override
    public void getAggregation(RpcController controller, MyFirstCoprocessor.MyCoprocessRequest request, RpcCallback done) {
        Scan scan = new Scan();
        scan.addFamily(Bytes.toBytes(request.getFamily()));
        //传入列的方式   sales:MAX,sales:MIN,sales:AVG,slaes:SUM,sales:COUNT
        String colums = request.getColumns();
        //记录所有要扫描的列
        Map> columnMaps = new HashedMap();
        for (String columnAndType : colums.split(",")) {
            String column = columnAndType.split(":")[0];
            String type = columnAndType.split(":")[1];
            List typeList = null;
            if (columnMaps.containsKey(column)) {
                typeList = columnMaps.get(column);
            } else {
                typeList = new ArrayList<>();
                //将column添加到Scan中
                scan.addColumn(Bytes.toBytes(request.getFamily()), Bytes.toBytes(column));
            }
            typeList.add(type);
            columnMaps.put(column, typeList);
        }
        InternalScanner scanner = null;
        MyFirstCoprocessor.MyCoprocessResponse response = null;
        Double max = null;
        Double min = null;
        Double sumVal = null;
        long counter = 0L;
        try {
            scanner = this.env.getRegion().getScanner(scan);
            List results = new ArrayList<>();
            boolean hasMore = false;
            scanner = env.getRegion().getScanner(scan);
            do {
                hasMore = scanner.next(results);
                if (results.size() > 0) {
                    ++counter;
                }
                log.info("counter:" + counter);
                log.info("results size:" + results.size());
                for (Cell cell : results) {
                    String column = Bytes.toString(CellUtil.cloneQualifier(cell));
                    log.info("Column Name: " + column);
                    log.info("Cell Value:" + new String(CellUtil.cloneValue(cell)));
                    Double temp = Double.parseDouble(new String(CellUtil.cloneValue(cell)));
                    if (columnMaps.containsKey(column)) {
                        List types = columnMaps.get(column);
                        for (String type : types) {
                            switch (type.toUpperCase()) {
                                case "MIN":
                                    min = min != null && (temp == null || compare(temp, min) >= 0) ? min : temp;
                                    log.info("MIN Value: " + min.doubleValue());
                                    break;
                                case "MAX":
                                    max = max != null && (temp == null || compare(temp, max) <= 0) ? max : temp;
                                    break;
                                case "SUM":
                                    if (temp != null) {
                                        sumVal = add(sumVal, temp);
                                    }
                                    break;
                                default:
                                    break;
                            }
                        }
                    }
                }
                results.clear();
            } while (hasMore);
            response = MyFirstCoprocessor.MyCoprocessResponse.newBuilder()
                    .setMaxnum(max!=null?max.doubleValue():Double.MAX_VALUE)
                    .setMinnum(min!=null?min.doubleValue():Double.MIN_NORMAL)
                    .setCount(counter)
                    .setSumnum(sumVal!=null?sumVal.doubleValue():Double.MIN_NORMAL).build();
        } catch (IOException e) {
            e.printStackTrace();
            ResponseConverter.setControllerException(controller, e);
        } finally {
            if (scanner != null) {
                try {
                    scanner.close();
                } catch (IOException e) {
                    e.printStackTrace();
                }
            }
        }
        done.run(response);
    }
    public static int compare(Double l1, Double l2) {
        if (l1 == null ^ l2 == null) {
            return l1 == null ? -1 : 1; // either of one is null.
        } else if (l1 == null)
            return 0; // both are null
        return l1.compareTo(l2); // natural ordering.
    }
    public double divideForAvg(Double d1, Long l2) {
        return l2 != null && d1 != null?d1.doubleValue() / l2.doubleValue():0.0D / 0.0;
    }
    public Double add(Double d1, Double d2) {
        return d1 != null && d2 != null ? Double.valueOf(d1.doubleValue() + d2.doubleValue()) : (d1 == null ? d2 : d1);
    }
    @Override
    public void start(CoprocessorEnvironment coprocessorEnvironment) throws IOException {
        if (coprocessorEnvironment instanceof RegionCoprocessorEnvironment) {
            this.env = (RegionCoprocessorEnvironment) coprocessorEnvironment;
        } else {
            throw new CoprocessorException("Must be loaded on a table region!");
        }
    }
    @Override
    public void stop(CoprocessorEnvironment coprocessorEnvironment) throws IOException {
    }
    @Override
    public Service getService() {
        return this;
    }
}

• MyFirstCoprocessExample.java类

package com.cloudera.hbase.coprocessor.client;
import com.cloudera.hbase.coprocessor.server.MyFirstCoprocessor;
import com.cloudera.hbase.coprocessor.server.MyFirstCoprocessorEndPoint;
import com.google.common.util.concurrent.AtomicDouble;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hbase.Coprocessor;
import org.apache.hadoop.hbase.HBaseConfiguration;
import org.apache.hadoop.hbase.HTableDescriptor;
import org.apache.hadoop.hbase.TableName;
import org.apache.hadoop.hbase.client.Connection;
import org.apache.hadoop.hbase.client.ConnectionFactory;
import org.apache.hadoop.hbase.client.Table;
import org.apache.hadoop.hbase.client.coprocessor.Batch;
import org.apache.hadoop.hbase.client.coprocessor.DoubleColumnInterpreter;
import org.apache.hadoop.hbase.ipc.BlockingRpcCallback;
import java.io.IOException;
import java.util.concurrent.atomic.AtomicLong;

public class MyFirstCoprocessExample {
    public static void main(String[] args) {
        String table_name = "fayson_coprocessor";
        //初始化HBase配置
        Configuration configuration = HBaseConfiguration.create();
        configuration.set("hbase.zookeeper.property.clientPort", "2181");
        configuration.setStrings("hbase.zookeeper.quorum", "ip-186-31-5-38.ap-southeast-1.compute.internal,ip-186-31-8-230.ap-southeast-1.compute.internal,ip-186-31-5-171.ap-southeast-1.compute.internal");
        try {
            //创建一个HBase的Connection
            Connection connection = ConnectionFactory.createConnection(configuration);
            TableName tableName = TableName.valueOf(table_name);
            if(!connection.getAdmin().tableExists(tableName)) {
                System.out.println(table_name + "does not exist....");
                System.exit(0);
            }
            Table table = connection.getTable(tableName);
            //删除表上的协处理器
            deleteCoprocessor(connection, table, MyFirstCoprocessorEndPoint.class);
            //为指定的表添加协处理器
            String hdfspath = "hdfs://nameservice3/hbase/coprocessor/hbase-demo-1.0-SNAPSHOT.jar";
            setupToExistTable(connection, table, hdfspath, MyFirstCoprocessorEndPoint.class);
            //客户端调用Region端的协处理器
            execFastEndpointCoprocessor(table, "info", "sales:MAX,sales:MIN,sales:AVG,sales:SUM,sales:COUNT");
            //关闭连接
            connection.close();
        } catch (IOException e) {
            e.printStackTrace();
        }
    }
    
    public static void deleteCoprocessor(Connection connection, Table table, Class... cls) {
        System.out.println("begin delete " + table.getName().toString() + " Coprocessor......");
        try {
            HTableDescriptor hTableDescriptor = table.getTableDescriptor();
            for(Class cass : cls) {
                hTableDescriptor.removeCoprocessor(cass.getCanonicalName());
            }
            connection.getAdmin().modifyTable(table.getName(), hTableDescriptor);
        } catch (IOException e) {
            e.printStackTrace();
        }
        System.out.println("end delete " + table.getName().toString() + " Coprocessor......");
    }
    
    public static void setupToExistTable(Connection connection, Table table, String jarPath, Class... cls) {
        try {
            if(jarPath != null && !jarPath.isEmpty()) {
                Path path = new Path(jarPath);
                HTableDescriptor hTableDescriptor = table.getTableDescriptor();
                for(Class cass : cls) {
                    hTableDescriptor.addCoprocessor(cass.getCanonicalName(), path, Coprocessor.PRIORITY_USER, null);
                }
                connection.getAdmin().modifyTable(table.getName(), hTableDescriptor);
            }
        } catch (IOException e) {
            e.printStackTrace();
        }
    }
    
    public static long execFastEndpointCoprocessor(Table table, String family, String columns) {
        long start_t = System.currentTimeMillis();
        //定义总的 rowCount 变量
        AtomicLong totalRowCount = new AtomicLong();
        AtomicDouble maxValue = new AtomicDouble(Double.MIN_VALUE);
        AtomicDouble minValue = new AtomicDouble(Double.MAX_VALUE);
        AtomicDouble sumValue = new AtomicDouble();
        try {
            Batch.Callback callback = new Batch.Callback() {
                @Override
                public void update(byte[] bytes, byte[] bytes1, MyFirstCoprocessor.MyCoprocessResponse myCoprocessResponse) {
                    //更新Count值
                    totalRowCount.getAndAdd(myCoprocessResponse.getCount());
                    //更新最大值
                    if(myCoprocessResponse.getMaxnum() > maxValue.doubleValue()) {
                        maxValue.compareAndSet(maxValue.doubleValue(), myCoprocessResponse.getMaxnum());
                    }
                    //更新最小值
                    if(myCoprocessResponse.getMinnum() < minValue.doubleValue()) {
                        minValue.compareAndSet(minValue.doubleValue(), myCoprocessResponse.getMinnum());
                    }
                    //更新求和
                    sumValue.getAndAdd(myCoprocessResponse.getSumnum());
                }
            };
            table.coprocessorService(MyFirstCoprocessor.AggregationService.class, null, null, new Batch.Call() {
                @Override
                public MyFirstCoprocessor.MyCoprocessResponse call(MyFirstCoprocessor.AggregationService aggregationService) throws IOException {
                    MyFirstCoprocessor.MyCoprocessRequest requet = MyFirstCoprocessor.MyCoprocessRequest.newBuilder().setFamily(family).setColumns(columns).build();
                    BlockingRpcCallback rpcCallback = new BlockingRpcCallback<>();
                    aggregationService.getAggregation(null, requet, rpcCallback);
                    MyFirstCoprocessor.MyCoprocessResponse response = rpcCallback.get();
                    return response;
                }
            }, callback);
        } catch (Throwable throwable) {
            throwable.printStackTrace();
        }
        System.out.println("耗时：" + (System.currentTimeMillis() - start_t));
        System.out.println("totalRowCount：" + totalRowCount.longValue());
        System.out.println("maxValue：" + maxValue.doubleValue());
        System.out.println("minValue：" + minValue.doubleValue());
        System.out.println("sumValue：" + sumValue.doubleValue());
        System.out.println("avg：" + new DoubleColumnInterpreter().divideForAvg(sumValue.doubleValue(), totalRowCount.longValue()));
        return totalRowCount.longValue();
    }
}

• mvn编译

mvn clean package

• 上传HDFS的/hbase/coprocessor目录

[root@ip-186-31-5-38 ~]# export HADOOP_USER_NAME=hbase         
[root@ip-186-31-5-38 ~]# hadoop fs -mkdir -p /hbase/coprocessor
[root@ip-186-31-5-38 ~]# hadoop fs -put hbase-demo-1.0-SNAPSHOT.jar /hbase/coprocessor
[root@ip-186-31-5-38 ~]# hadoop fs -ls /hbase/coprocessor

大数据视频推荐：
CSDN
大数据语音推荐：
ELK7 stack开发运维
企业级大数据技术应用
大数据机器学习案例之推荐系统
自然语言处理
大数据基础
人工智能：深度学习入门到精通