Phoenix 使用技巧 | 易学教程

Phoenix shell:

1、启动：phoenix/bin/sqlline.py hadoop:2181
（说明：phoenix也是基于zookeeper的，启动时后面需要跟zookeeper集群信息，进入后可通过 !list查看jdbc连接地址）

2、帮助：help 查看具体的使用命令

3、查看元数据：

select TENANT_ID,TABLE_SCHEM,TABLE_NAME,COLUMN_NAME,COLUMN_FAMILY,DATA_TYPE,COLUMN_SIZE,DECIMAL_DIGITS from SYSTEM.CATALOG;

(说明：通过查看元数据可以看到表字段名、字段类型、映射hbase的列族名、字段大小)

4、查看指定表元数据：

select TENANT_ID,TABLE_SCHEM,TABLE_NAME,COLUMN_NAME,COLUMN_FAMILY,DATA_TYPE,COLUMN_SIZE,DECIMAL_DIGITS from SYSTEM.CATALOG where TABLE_NAME='ADS_WTCHK_SALES_DASHBOARD_H_REALTIME_TMP';

select元数据
（说明：表名大写时，需用 ‘table_name’ 单引号，phoenix对大小写敏感，当hbase和phoenix表名为小写时，select需用 “table_name” ）

5、添加数据：

UPSERT INTO "yul_test"("id","BU","StoreNo","TicketNo","PosID","DepartmentNumber","ItemNo","CardNumber","Qty","Price","Amount","TransTime") 
VALUES('000001','yulin',27,'etl',334,34342324,'gwet',4554,1323,4543.45,34.1,'dgdg');

(说明：当字段类型为char时需用 '单引号)

6、CREATE INDEX IDX 创建索引：

CREATE INDEX IDX_INSERT_TEST_USERID ON "Insert_test"("UserID");//全局二级索引
CREATE LOCAL INDEX IDX_INSERT_TEST_USERID ON "Insert_Test"("UserID");//本地索引

（说明：同一张表，索引数量不得超过10，索引表越多，插入数据越慢！）

explain //执行计划，在sql语句前加，查看sql准备做些什么。

SELECT /*+ INDEX(my_table my_index) */ v2 FROM my_table WHERE v1 = 'foo';//强制指定索引
DROP INDEX IDX_INSERT_TEST_USERID ON "Insert_test";//删除索引

（注：SELECT字段含有非索引的字段，则索引不会被使用）

7、删除

DELETE FROM SYSTEM.CATALOG where TABLE_NAME='ads_idmapping_paltform_ids'; //删除表元数据，删除时要把hbase表数据清空
ALTER TABLE ADS_WTCHK_SALES_DASHBOARD_D_REALTIME_TMP DROP COLUMN "store_target";  //删除指定字段

(说明：删除表数据和字段时，需要清空当前表数据或者所在字段数据)

8、添加字段

ALTER TABLE my_table ADD dept_name char(50)	//添加指定字段
ALTER TABLE ADS_WTCHK_SALES_DASHBOARD_D_TOTAL_REALTIME_TMP ADD "info"."flow_count" DOUBLE VERSIONS=10 //添加字段并加上列族

9、建表实例：

--总total每小时销售
create table ADS_WTCHK_SALES_DASHBOARD_H_TOTAL_REALTIME_TMP(
"info"."total_sales_h" DOUBLE,
"info"."total_txn_h" DOUBLE,
"stat_date" VARCHAR(20) NOT NULL,
CONSTRAINT pk PRIMARY KEY ("stat_date")
) SALT_BUCKETS = 48;
create local index WTCHK_SALES_DASHBOARD_H_TOTAL_REALTIME_TMP ON ADS_WTCHK_SALES_DASHBOARD_H_TOTAL_REALTIME_TMP("stat_date");//本地索引

Phoenix scala连接：

package cn.com.pnshk.sink

import java.sql.{Connection, DriverManager, Statement}
import java.text.SimpleDateFormat
import java.util.{Date, Properties, TimeZone}

import cn.com.pnshk.pojo.SalesDetails
import cn.com.pnshk.utils.SqlUtils
import org.apache.flink.configuration.Configuration
import org.apache.flink.streaming.api.functions.sink.RichSinkFunction

class SinkPhoenix extends RichSinkFunction[SalesDetails]{


  private var connection:Connection = null
  private var statement:Statement = null

  override def open(parameters: Configuration): Unit = {

    val prop = new Properties()
    val stream = this.getClass.getClassLoader.getResourceAsStream("prop.properties")
    prop.load(stream)

    val driver = prop.getProperty("org.apache.phoenix.jdbc.PhoenixDriver")
    val url = prop.getProperty("jdbc:phoenix:hadoop:2181:/hbase-unsecure")

    Class.forName(driver)
    connection = DriverManager.getConnection(url)
    statement = connection.createStatement()
    statement.setQueryTimeout(6000)
  }

  //每个元素的插入，都要触发一次invoke，这里主要进行invoke插入
  override def invoke(sales:SalesDetails): Unit = {
    try {
      val sqlList = SqlUtils.getSql(sales)
      for (sql <- sqlList) {
        statement.execute(sql)
        connection.commit()
      }
  val dateFormat: SimpleDateFormat = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss")
      val today = dateFormat.format(new Date())
      println("success" + today)

    }
    catch {
      case e: Exception => println(e.getMessage)
    }
  }
  
	//释放资源
  override def close(): Unit = {
    if(statement != null){
      statement.close()
    }
    if(connection != null){
      connection.close()
    }
  }
}

（注：上面的RichSinkFunction我是配合Flink的addSink使用）

来源：CSDN

作者：Lone--Wolf

链接：https://blog.csdn.net/qq_39140816/article/details/103909946

标签

phoenix