Hive 是一个基于 hadoop 的开源数据仓库工具,用于存储和处理海量结构化数据。它把海量数据存储于 hadoop 文件系统,而不是数据库,但提供了一套类数据库的数据存储和处理机制,并采用 HQL (类 SQL )语言对这些数据进行自动化管理和处理。我们可以把 Hive 中海量结构化数据看成一个个的表,而实际上这些数据是分布式存储在 HDFS 中的。 Hive 经过对语句进行解析和转换,最终生成一系列基于 hadoop 的 map/reduce 任务,通过执行这些任务完成数据处理。
Hive 诞生于 facebook 的日志分析需求,面对海量的结构化数据, Hive 以较低的成本完成了以往需要大规模数据库才能完成的任务,并且学习门槛相对较低,应用开发灵活而高效。
Hive 自 2009.4.29 发布第一个官方稳定版 0.3.0 至今,不过一年的时间,正在慢慢完善,网上能找到的相关资料相当少,尤其中文资料更少,本文结合业务对 Hive 的应用做了一些探索,并把这些经验做一个总结,所谓前车之鉴,希望读者能少走一些弯路。
JDK:1.8
Hadoop Release:2.7.4
centos:7.3
node1(master) 主机: 192.168.252.121
node2(slave1) 从机: 192.168.252.122
node3(slave2) 从机: 192.168.252.123
node4(mysql) 从机: 192.168.252.124
安装Apache Hive
前提是要先安装hadoop
集群,并且hive只需要在hadoop的namenode节点集群里安装即可(需要在有的namenode上安装),可以不在datanode节点的机器上安装。还需要说明的是,虽然修改配置文件并不需要把hadoop运行起来,但是本文中用到了hadoop的hdfs命令,在执行这些命令时你必须确保hadoop是正在运行着的,而且启动hive的前提也需要hadoop在正常运行着,所以建议先把hadoop集群启动起来。
安装MySQL
用于存储 Hive 的元数据(也可以用 Hive 自带的嵌入式数据库 Derby,但是 Hive 的生产环境一般不用 Derby),这里只需要安装 MySQL 单机版即可,如果想保证高可用的化,也可以部署 MySQL 主从模式;
Hadoop
Hadoop-2.7.4 集群快速搭建
MySQL 随意任选其一
CentOs7.3 安装 MySQL 5.7.19 二进制版本
搭建 MySQL 5.7.19 主从复制,以及复制实现细节分析
su hadoop
cd /home/hadoop/
wget https://mirrors.tuna.tsinghua.edu.cn/apache/hive/hive-2.3.0/apache-hive-2.3.0-bin.tar.gz
tar -zxvf apache-hive-2.3.0-bin.tar.gz
mv apache-hive-2.3.0-bin hive-2.3.0
如果是对所有的用户都生效就修改vi /etc/profile
文件
如果只针对当前用户生效就修改 vi ~/.bahsrc
文件
sudo vi /etc/profile
#hive
export PATH=${HIVE_HOME}/bin:$PATH
export HIVE_HOME=/home/hadoop/hive-2.3.0/
使环境变量生效,运行 source /etc/profile
使/etc/profile
文件生效
cd /home/hadoop/hive-2.3.0/conf
cp hive-default.xml.template hive-site.xml
使用 hadoop 新建 hdfs 目录,因为在 hive-site.xml 中有默认如下配置:
hive.metastore.warehouse.dir
/user/hive/warehouse
location of default database for the warehouse
进入 hadoop 安装目录 执行hadoop命令新建/user/hive/warehouse目录,并授权,用于存储文件
cd /home/hadoop/hadoop-2.7.4
bin/hadoop fs -mkdir -p /user/hive/warehouse
bin/hadoop fs -mkdir -p /user/hive/tmp
bin/hadoop fs -mkdir -p /user/hive/log
bin/hadoop fs -chmod -R 777 /user/hive/warehouse
bin/hadoop fs -chmod -R 777 /user/hive/tmp
bin/hadoop fs -chmod -R 777 /user/hive/log
用以下命令检查目录是否创建成功
bin/hadoop fs -ls /user/hive
搜索hive.exec.scratchdir,将该name对应的value修改为/user/hive/tmp
<property>
<name>hive.exec.scratchdirname>
<value>/user/hive/tmpvalue>
property>
搜索hive.querylog.location,将该name对应的value修改为/user/hive/log/hadoop
<property>
<name>hive.querylog.locationname>
<value>/user/hive/log/hadoopvalue>
<description>Location of Hive run time structured log filedescription>
property>
搜索javax.jdo.option.connectionURL,将该name对应的value修改为MySQL的地址
<property>
<name>javax.jdo.option.ConnectionURLname>
<value>jdbc:mysql://192.168.252.124:3306/hive?createDatabaseIfNotExist=truevalue>
<description>
JDBC connect string for a JDBC metastore.
To use SSL to encrypt/authenticate the connection, provide database-specific SSL flag in the connection URL.
For example, jdbc:postgresql://myhost/db?ssl=true for postgres database.
description>
property>
搜索javax.jdo.option.ConnectionDriverName,将该name对应的value修改为MySQL驱动类路径
<property>
<name>javax.jdo.option.ConnectionDriverNamename>
<value>com.mysql.jdbc.Drivervalue>
<description>Driver class name for a JDBC metastoredescription>
property>
搜索javax.jdo.option.ConnectionUserName,将对应的value修改为MySQL数据库登录名
<property>
<name>javax.jdo.option.ConnectionUserNamename>
<value>rootvalue>
<description>Username to use against metastore databasedescription>
property>
搜索javax.jdo.option.ConnectionPassword,将对应的value修改为MySQL数据库的登录密码
<property>
<name>javax.jdo.option.ConnectionPasswordname>
<value>mimavalue>
<description>password to use against metastore databasedescription>
property>
mkdir /home/hadoop/hive-2.3.0/tmp
并在 hive-site.xml
中修改
把{system:java.io.tmpdir}
改成 /home/hadoop/hive-2.3.0/tmp
把 {system:user.name}
改成 {user.name}
cp hive-env.sh.template hive-env.sh
vi hive-env.sh
HADOOP_HOME=/home/hadoop/hadoop-2.7.4/
export HIVE_CONF_DIR=/home/hadoop/hive-2.3.0/conf
export HIVE_AUX_JARS_PATH=/home/hadoop/hive-2.3.0/lib
cd /home/hadoop/hive-2.3.0/lib
wget http://central.maven.org/maven2/mysql/mysql-connector-java/5.1.38/mysql-connector-java-5.1.38.jar
首先确保 mysql 中已经创建 hive
库
cd /home/hadoop/hive-2.3.0/bin
./schematool -initSchema -dbType mysql
如果看到如下,表示初始化成功
Starting metastore schema initialization to 2.3.0
Initialization script hive-schema-2.3.0.mysql.sql
Initialization script completed
schemaTool completed
/usr/local/mysql/bin/mysql -uroot -p
mysql> show databases;
+--------------------+
| Database |
+--------------------+
| information_schema |
| hive |
| mysql |
| performance_schema |
| sys |
+--------------------+
5 rows in set (0.00 sec)
mysql> use hive;
Reading table information for completion of table and column names
You can turn off this feature to get a quicker startup with -A
Database changed
mysql> show tables;
+---------------------------+
| Tables_in_hive |
+---------------------------+
| AUX_TABLE |
| BUCKETING_COLS |
| CDS |
| COLUMNS_V2 |
| COMPACTION_QUEUE |
| COMPLETED_COMPACTIONS |
| COMPLETED_TXN_COMPONENTS |
| DATABASE_PARAMS |
| DBS |
| DB_PRIVS |
| DELEGATION_TOKENS |
| FUNCS |
| FUNC_RU |
| GLOBAL_PRIVS |
| HIVE_LOCKS |
| IDXS |
| INDEX_PARAMS |
| KEY_CONSTRAINTS |
| MASTER_KEYS |
| NEXT_COMPACTION_QUEUE_ID |
| NEXT_LOCK_ID |
| NEXT_TXN_ID |
| NOTIFICATION_LOG |
| NOTIFICATION_SEQUENCE |
| NUCLEUS_TABLES |
| PARTITIONS |
| PARTITION_EVENTS |
| PARTITION_KEYS |
| PARTITION_KEY_VALS |
| PARTITION_PARAMS |
| PART_COL_PRIVS |
| PART_COL_STATS |
| PART_PRIVS |
| ROLES |
| ROLE_MAP |
| SDS |
| SD_PARAMS |
| SEQUENCE_TABLE |
| SERDES |
| SERDE_PARAMS |
| SKEWED_COL_NAMES |
| SKEWED_COL_VALUE_LOC_MAP |
| SKEWED_STRING_LIST |
| SKEWED_STRING_LIST_VALUES |
| SKEWED_VALUES |
| SORT_COLS |
| TABLE_PARAMS |
| TAB_COL_STATS |
| TBLS |
| TBL_COL_PRIVS |
| TBL_PRIVS |
| TXNS |
| TXN_COMPONENTS |
| TYPES |
| TYPE_FIELDS |
| VERSION |
| WRITE_SET |
+---------------------------+
57 rows in set (0.00 sec)
启动Hive
cd /home/hadoop/hive-2.3.0/bin
./hive
创建 hive 库
hive> create database ymq;
OK
Time taken: 0.742 seconds
选择库
hive> use ymq;
OK
Time taken: 0.036 seconds
创建表
hive> create table test (mykey string,myval string);
OK
Time taken: 0.569 seconds
插入数据
hive> insert into test values("1","www.ymq.io");
WARNING: Hive-on-MR is deprecated in Hive 2 and may not be available in the future versions. Consider using a different execution engine (i.e. spark, tez) or using Hive 1.X releases.
Query ID = hadoop_20170922011126_abadfa44-8ebe-4ffc-9615-4241707b3c03
Total jobs = 3
Launching Job 1 out of 3
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_1506006892375_0001, Tracking URL = http://node1:8088/proxy/application_1506006892375_0001/
Kill Command = /home/hadoop/hadoop-2.7.4//bin/hadoop job -kill job_1506006892375_0001
Hadoop job information for Stage-1: number of mappers: 1; number of reducers: 0
2017-09-22 01:12:12,763 Stage-1 map = 0%, reduce = 0%
2017-09-22 01:12:20,751 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 1.24 sec
MapReduce Total cumulative CPU time: 1 seconds 240 msec
Ended Job = job_1506006892375_0001
Stage-4 is selected by condition resolver.
Stage-3 is filtered out by condition resolver.
Stage-5 is filtered out by condition resolver.
Moving data to directory hdfs://node1:9000/user/hive/warehouse/ymq.db/test/.hive-staging_hive_2017-09-22_01-11-26_242_8022847052615616955-1/-ext-10000
Loading data to table ymq.test
MapReduce Jobs Launched:
Stage-Stage-1: Map: 1 Cumulative CPU: 1.24 sec HDFS Read: 4056 HDFS Write: 77 SUCCESS
Total MapReduce CPU Time Spent: 1 seconds 240 msec
OK
Time taken: 56.642 seconds
查询数据
hive> select * from test;
OK
1 www.ymq.io
Time taken: 0.253 seconds, Fetched: 1 row(s)
在界面上查看刚刚写入的hdfs数据