在搭建了Hadoop和hive环境后,就可以使用hive来进行数据库相关操作了。Hive提供了hql(类sql)语句来操作,基本过程与mysql类似,区别的就是对于hive中的聚合操作,将使用hadoop底层的mapreduce进程来执行。
下面以一个游戏公司的游戏、用户等相关分析大数据业务为例,以Hive为工具来完成游戏活跃度、用户使用情况等的统计分析工作。
(1)数据的产生
因为获取游戏公司的实际数据还是比较困难的,我们直接自己来构建。使用python脚本就可以完成。由于hadoop和hive都是安装在centos系统上,centos默认安装了python2.7,所以可以直接编写脚本,然后在centos上运行得到结果。
用户数据的构建,模拟产生1000个用户
import random
def getUser():
location_List= ['BJ','SH','TJ','GZ','SZ']
fd= fopen('userinfo','w+')
for i in range(1000):
userid = str(1000+i)
age = str(random.randrange(10,40))
area = random.choice(location_List)
user_money = str(i)
str_tmp = userid + ','+ age+ ',' + area + ',' + usermoney + '\n'
fd.write(str_tmp)
fd.close()
if __name__=='__main__':
getUser()
游戏信息的构建,模拟共4个游戏:
def getGame():
Game_list= ['CHESS','LANDLORD','QGAME','ROYAL']
fd =fopen('gameinfo','w+')
for i in range(4):
gameid = str(i)
gamename = Game_list[i]
str_tmp = gameid+ ',' +gamename + '\n'
fd.write(str_tmp)
fd.close()
if __name__=='__main__':
getGame()
用户玩游戏时间数据构建,模拟一共10天的用户玩游戏时间的记录
import datetime
from datetime import timedelta
import random
def gameTime():
game_list = [0, 1, 2,3]
time_list= [10,15,20,20,50,60,90]
for j in range(10):
fdate = (datetime.datetime.now() + datetime.timedelta(days=j)).strftime('%Y-%m-%d')
fd = open('gametime\\{}\\gametime_{}.txt'.format(fdate,fdate), 'w+')
for i in range(1000):
userid = str(1000 + i)
gameid = str(random.choice(game_list))
gametime = str(random.choice(time_list))
str_tmp = fdate+ ','+ userid+ ','+ gameid + ',' + gametime + '\n' + '\n'
fd.write(str_tmp)
fd.close()
if __name__=='__main__':
gameTime()
用户玩游戏费用数据构建,模拟这10天里每天的费用:
import random,datetime
from datetime import timedelta
def userFee():
game_list = [0, 1, 2,3]
money_list= [10,15,30,27,55,66,90]
for j in range(10):
fdate = (datetime.datetime.now() + datetime.timedelta(days=j)).strftime('%Y-%m-%d')
os.mkdir('userfee\\{}'.format(fdate))
for j in range(10):
fdate = (datetime.datetime.now() + datetime.timedelta(days=j)).strftime('%Y-%m-%d')
fd = open('userfee\\{}\\userfee_{}.txt'.format(fdate,fdate), 'w+')
for i in range(1000):
userid = str(1000 + i)
gameid = str(random.choice(game_list))
gametime = str(random.choice(money_list))
str_tmp = fdate+ ',' + userid+ ','+ gameid + ',' + gametime + '\n' + '\n'
fd.write(str_tmp)
fd.close()
if __name__=='__main__':
userFee()
(2)数据的Hive存储
上述的用户数据都是以文件方式存储下来的,接下来我们将其存储到Hadoop上。使用的时候就是利用Hive以建表创建数据、插入数据等方式来实现。
首先在hadoop中新建如下文件夹,用于设置hive存储位置。
[[email protected] ~]$ hdfs dfs -mkdir /stat
[[email protected] ~]$ hdfs dfs -mkdir /stat/data/
[[email protected] ~]$ hdfs dfs -mkdir /stat/data/gameinfo
[[email protected] ~]$ hdfs dfs -mkdir /stat/data/gametime
[[email protected] ~]$ hdfs dfs -mkdir /stat/data/userinfo
[[email protected] ~]$ hdfs dfs -mkdir /stat/data/userfee
然后进入hive shell命令端,开始创建stat数据库,用于存放上述生产的数据,同时创建一个analysis数据库,用于存放hive统计分析数据。
hive> create database stat;
hive> create database analysis;
接下来就可以使用hive创建表的命令创建4个表,gameinfo,gametime,userinfo,userfee:
hive > use stat;
hive > create table if not EXISTS gameinfo ( gameid int, gamename string)
> row format delimited fields terminated by ','
> location '/stat/data/gameinfo';
hive > create table if not EXISTS userinfo ( userid int, age int, area string , money int)
> row format delimited fields terminated by ','
> location '/stat/data/userinfo';
hive > create table if not EXISTS gametime ( date string, userid int, gameid int, gametime int)
> partitioned by (dt string )
> row format delimited fields terminated by ','
> location '/stat/data/gametime';
hive > create table if not EXISTS userfee ( date string, userid int, gameid int, fee int)
> partitioned by (st string)
> row format delimited fields terminated by ','
> location '/stat/data/userfee';
注意到建表时各个字段与第一步构建数据时要对应,这样保证后面数据能够正常导入。其中gametime和userfee都是涉及到分区,因为有每日的数据需要单独进行存储,所以在创建表时就设定好分区。
有了表名和字段定义后,可以导入数据了。
如下将userinfo和gameinfo的数据存入hive:
hive > load data local inpath 'userinfo.txt' into table stat.userinfo ;
hive > load data local inpath 'gameinfo.txt' into table stat.gameinfo ;
对于后续两个分区表,则采用alter table add方式来导入:
hive > alter table stat.gametime add if not EXISTS partition (dt = '2020-02-09')
> location '/stat/data/gametime/2020-02-09/';
hive > alter table stat.gametime add if not EXISTS partition (dt = '2020-02-10')
> location '/stat/data/gametime/2020-02-10/';
hive > alter table stat.gametime add if not EXISTS partition (dt = '2020-02-11')
> location '/stat/data/gametime/2020-02-11/';
hive > alter table stat.gametime add if not EXISTS partition (dt = '2020-02-12')
> location '/stat/data/gametime/2020-02-12/';
hive > alter table stat.gametime add if not EXISTS partition (dt = '2020-02-13')
> location '/stat/data/gametime/2020-02-13/';
hive > alter table stat.gametime add if not EXISTS partition (dt = '2020-02-14')
> location '/stat/data/gametime/2020-02-14/';
hive > alter table stat.gametime add if not EXISTS partition (dt = '2020-02-15')
> location '/stat/data/gametime/2020-02-15/';
hive > alter table stat.gametime add if not EXISTS partition (dt = '2020-02-16')
> location '/stat/data/gametime/2020-02-16/';
hive > alter table stat.gametime add if not EXISTS partition (dt = '2020-02-17')
> location '/stat/data/gametime/2020-02-17/';
hive > alter table stat.userfee add if not EXISTS partition (dt = '2020-02-09')
> location '/stat/data/userfee/2020-02-09/';
hive > alter table stat.userfee add if not EXISTS partition (dt = '2020-02-10')
> location '/stat/data/userfee/2020-02-10/';
hive > alter table stat.userfee add if not EXISTS partition (dt = '2020-02-11')
> location '/stat/data/userfee/2020-02-11/';
hive > alter table stat.userfee add if not EXISTS partition (dt = '2020-02-12')
> location '/stat/data/userfee/2020-02-12/';
hive > alter table stat.userfee add if not EXISTS partition (dt = '2020-02-13')
> location '/stat/data/userfee/2020-02-13/';
hive > alter table stat.userfee add if not EXISTS partition (dt = '2020-02-14')
> location '/stat/data/userfee/2020-02-14/';
hive > alter table stat.userfee add if not EXISTS partition (dt = '2020-02-15')
> location '/stat/data/userfee/2020-02-15/';
hive > alter table stat.userfee add if not EXISTS partition (dt = '2020-02-16')
> location '/stat/data/userfee/2020-02-16/';
hive > alter table stat.userfee add if not EXISTS partition (dt = '2020-02-17')
> location '/stat/data/userfee/2020-02-17/';
这样就把生产的游戏时间、用户费用数据存入了hdfs中。其实可以观察到,对于后面的两个用户游戏时间、用户游戏费用数据的存储,直接使用hive来操作还是比较笨重的,毕竟语句之间差别的就是日期变量。如果能使用脚本来传参,将日期作为变量传入执行脚本,执行效率会高很多。这里就需要启动hiveserver2进程,使用jdbc或者beeline来进行操作,尤其可以使用javaAPI或者python来远程访问,编写脚本来实现hive数据库的管理效率会跟多。
如下为hdfs数据结构:
进入hive,查询一下导入数据的情况:
hive> select * from stat.gameinfo;
OK
0 LandLord
1 Buffle
2 Farm
3 PuQ
Time taken: 0.212 seconds, Fetched: 4 row(s)
hive> select * from stat.gametime where dt='2020-02-09' limit 1;
OK
2020-02-09 1000 1 60 2020-02-09
Time taken: 0.498 seconds, Fetched: 1 row(s)
(3)数据的hive统计分析
前面的4个业务相关数据表,我们需要经过统计分析来获得相应的规律信息。如通过游戏信息表、用户玩游戏时间表关联分析可获得每个游戏每天的活跃度(统计每个游戏每天有多少用户在玩,进而分析最喜爱游戏进行推荐);通过用户信息表、用户游戏消费表关联分析来获得基于年龄的付费信息,即看游戏对哪个年龄段最有吸引力;通过游戏信息表、用户游戏消费表关联分析获得每个游戏获得的收入金额等。在hive shell输入hql语句,用于分析统计类的,包括各种聚合、关联条件等,执行时系统自动将任务交给底层的mapreduce来执行。此时我们将使用另外一个数据库analysis。
首先来写一下hql语句,实现游戏活跃度的统计:
hql= ''' insert overwrite table analysis.gameactive partition (dt='2020-02-17')
select gt.fdate as fdate,gi.fgamename as fgamename,count(gt.fuserid) as fcount from stat.gametime gt,stat.gameinfo as gi
where gt.fgameid=gi.fgameid and fdate='2020-02-17' group by gt.fdate,gi.fgamename ''';
如果表比较多的时候,还可以这样来使用:
insert overwrite table analysis.gameactive partition (dt='2020-02-17')
select gt.fdate as fdate,gi.fgamename as fgamename,count(gt.fuserid) as fcount
from stat.gametime gt
left join stat.gameinfo gi
on gt.fgameid = gi.fgameid
where fdate ='2020-02-17'
group by gt.fdate,gi.fgamename;
语句中有left join,左连接的方式,left join... on...,on后面为关联方式。然后接where语句,查询条件。最后在使用group by分组时,注意到分组所用的字段,一定是需要前面select语句里出现过的属性。
同样直接在hive shell命令行窗口输入:
hive > insert overwrite table analysis.gameactive partition (dt='2020-02-17')
hive > select gt.fdate as fdate,gi.fgamename as fgamename,count(gt.fuserid) as fcount
hive > from stat.gametime gt,stat.gameinfo as gi
hive > where gt.fgameid=gi.fgameid and fdate='2020-02-17'
hive > group by gt.fdate,gi.fgamename;
执行结束后,依次将时间修改成产生数据的10天时间,然后可以查询结果如下:
hive> select * from analysis.gameactive;
OK
2020-02-09 Buffle 268 2020-02-09
2020-02-09 Farm 221 2020-02-09
2020-02-09 LandLord 256 2020-02-09
2020-02-09 PuQ 255 2020-02-09
2020-02-10 Buffle 223 2020-02-10
2020-02-10 Farm 258 2020-02-10
2020-02-10 LandLord 255 2020-02-10
2020-02-10 PuQ 264 2020-02-10
2020-02-11 Buffle 235 2020-02-11
2020-02-11 Farm 259 2020-02-11
2020-02-11 LandLord 246 2020-02-11
2020-02-11 PuQ 260 2020-02-11
2020-02-12 Buffle 253 2020-02-12
2020-02-12 Farm 241 2020-02-12
2020-02-12 LandLord 266 2020-02-12
2020-02-12 PuQ 240 2020-02-12
2020-02-13 Buffle 222 2020-02-13
2020-02-13 Farm 273 2020-02-13
2020-02-13 LandLord 252 2020-02-13
2020-02-13 PuQ 253 2020-02-13
2020-02-14 Buffle 253 2020-02-14
2020-02-14 Farm 257 2020-02-14
2020-02-14 LandLord 245 2020-02-14
2020-02-14 PuQ 245 2020-02-14
2020-02-15 Buffle 251 2020-02-15
2020-02-15 Farm 239 2020-02-15
2020-02-15 LandLord 250 2020-02-15
2020-02-15 PuQ 260 2020-02-15
2020-02-16 Buffle 261 2020-02-16
2020-02-16 Farm 263 2020-02-16
2020-02-16 LandLord 231 2020-02-16
2020-02-16 PuQ 245 2020-02-16
2020-02-17 Buffle 242 2020-02-17
2020-02-17 Farm 223 2020-02-17
2020-02-17 LandLord 261 2020-02-17
2020-02-17 PuQ 274 2020-02-17
2020-02-18 Buffle 271 2020-02-18
2020-02-18 Farm 253 2020-02-18
2020-02-18 LandLord 226 2020-02-18
2020-02-18 PuQ 250 2020-02-18
Time taken: 0.523 seconds, Fetched: 40 row(s)
由此我们看到整个10天里,每天4个游戏的参与人数统计就出来了。
同样接下来实现游戏用户年龄段情况统计,使用的hql语句为:
hive > insert overwrite table analysis.gameuserage partition (dt='2020-02-11')
hive > select gt.fdate as fdate,sum(gt.fgametime) as fgametime,ui.fage as fage
hive > from stat.gametime gt,stat.userinfo as ui
hive > where gt.fuserid=ui.fuserid and fdate='2020-02-11'
hive > group by ui.fage,gt.fdate;
执行后,依次对每天数据进行统计处理,查询结果如下(第一列为日期,第二列为用户玩游戏的时间,第三列为用户年龄,因为在产生数据的时候我们设定了年龄段从10到39岁,所以结果就是对这30个年龄组玩游戏的时间进行了总计分析):
2020-02-18 1695 10 2020-02-18
2020-02-18 1350 11 2020-02-18
2020-02-18 1325 12 2020-02-18
2020-02-18 990 13 2020-02-18
2020-02-18 1355 14 2020-02-18
2020-02-18 1420 15 2020-02-18
2020-02-18 905 16 2020-02-18
2020-02-18 1140 17 2020-02-18
2020-02-18 1580 18 2020-02-18
2020-02-18 1085 19 2020-02-18
2020-02-18 1350 20 2020-02-18
2020-02-18 1525 21 2020-02-18
2020-02-18 1285 22 2020-02-18
2020-02-18 1105 23 2020-02-18
2020-02-18 1035 24 2020-02-18
2020-02-18 1185 25 2020-02-18
2020-02-18 975 26 2020-02-18
2020-02-18 1625 27 2020-02-18
2020-02-18 1370 28 2020-02-18
2020-02-18 1485 29 2020-02-18
2020-02-18 930 30 2020-02-18
2020-02-18 1390 31 2020-02-18
2020-02-18 1250 32 2020-02-18
2020-02-18 1005 33 2020-02-18
2020-02-18 1250 34 2020-02-18
2020-02-18 1280 35 2020-02-18
2020-02-18 970 36 2020-02-18
2020-02-18 1170 37 2020-02-18
2020-02-18 1015 38 2020-02-18
2020-02-18 1015 39 2020-02-18
Time taken: 0.446 seconds, Fetched: 270 row(s)
(4)sqoop工具的使用
sqoop是一个实现RDMS与HDFS数据交换处理的桥梁工具。实践过程中安装相对较为简单,不过本人在实践时下载sqoop1.99版本时在验证一直通不过,各种环境变量、依赖包都设置了还是不行。无奈只好选择sqoop1.4版本。
1. 安装配置
sqoop1.4tar包可以直接从各种镜像上下载下来,然后解压到centos,由于解压后名字较长,可以将其重命名一下。
[[email protected] ~]$ tar -xvf sqoop-1.4.6-cdh5.14.0.tar.gz
[[email protected] ~]$ mv sqoop-1.4.6-cdh5.14.0.tar.gz sqoop1.4.6
然后设置一下环境变量:
[[email protected] ~]$ vi ~/.bashrc
输入SQOOP_HOME路径:
export SQOOP_HOME=/home/hadoop/sqoop1.4.6
export PATH=$PATH:$SQOOP_HOME/bin
接下来进入sqoop文件夹的conf配置目录中,配置sqoop-env.sh:
[[email protected] conf]$ mv sqoop-env-template.sh sqoop-env.sh
[[email protected] conf]$ vi sqoop-env.sh
在该文件中给定hadoop相关路径:
#Set path to where bin/hadoop is available
export HADOOP_COMMON_HOME=/home/hadoop/hadoop-3.1.2
#Set path to where hadoop-*-core.jar is available
export HADOOP_MAPRED_HOME=/home/hadoop/hadoop-3.1.2
#set the path to where bin/hbase is available
#export HBASE_HOME=
#Set the path to where bin/hive is available
export HIVE_HOME=/home/hadoop/hive-3.1.2-bin
#Set the path for where zookeper config dir is
#export ZOOCFGDIR=
由于本次没有使用hbase和zookeeper,所以没有设置两者的路径。
如此环境变量就配置完毕,还有两个jar包需要导入,一个是mysql的jdbc工具包,一个是java的json工具包,这两个都可以从网上下载下来,然后导入到sqoop的lib目录中:
[[email protected] lib]$ ll java-json.jar
-rw-r--r--. 1 hadoop hadoop 84697 Feb 13 14:35 java-json.jar
[[email protected] lib]$ ll mysql-connector-java-8.0.16.jar
-rw-r--r--. 1 hadoop hadoop 2293144 Feb 13 12:05 mysql-connector-java-8.0.16.jar
2. 开始使用
sqoop的语法较为复杂,不过感觉也都是模板化的,按照规则就可以正常执行。
首先来测试一下,在当前用户目录下输入sqoop help:
[[email protected] ~]$ sqoop help
Warning: /home/hadoop/sqoop1.4.6/../hbase does not exist! HBase imports will fail.
Please set $HBASE_HOME to the root of your HBase installation.
Warning: /home/hadoop/sqoop1.4.6/../hcatalog does not exist! HCatalog jobs will fail.
Please set $HCAT_HOME to the root of your HCatalog installation.
Warning: /home/hadoop/sqoop1.4.6/../accumulo does not exist! Accumulo imports will fail.
Please set $ACCUMULO_HOME to the root of your Accumulo installation.
Warning: /home/hadoop/sqoop1.4.6/../zookeeper does not exist! Accumulo imports will fail.
Please set $ZOOKEEPER_HOME to the root of your Zookeeper installation.
2020-02-13 14:56:14,436 INFO sqoop.Sqoop: Running Sqoop version: 1.4.6-cdh5.14.0
usage: sqoop COMMAND [ARGS]
Available commands:
codegen Generate code to interact with database records
create-hive-table Import a table definition into Hive
eval Evaluate a SQL statement and display the results
export Export an HDFS directory to a database table
help List available commands
import Import a table from a database to HDFS
import-all-tables Import tables from a database to HDFS
import-mainframe Import datasets from a mainframe server to HDFS
job Work with saved jobs
list-databases List available databases on a server
list-tables List available tables in a database
merge Merge results of incremental imports
metastore Run a standalone Sqoop metastore
version Display version information
See 'sqoop help COMMAND' for information on a specific command.
我们看到help中提示的,也是sqoop的主要使用方法,import 和export,可以将mysql中的表导入到HDFS系统中,也可以将HDFS系统目录导入到mysql中以表的形式存储。还可以使用create-hive-table来直接创建hive的表。
例如先将mysql中创建一个数据库名为sqoop,然后新建一个数据表为gameactive,插入两行记录如下:
mysql> select * from gameactive;
Empty set (0.01 sec)
mysql> insert into gameactive values('2020-02-13','puke',100),('2020-02-13','dizhu',150);
Query OK, 2 rows affected (0.14 sec)
Records: 2 Duplicates: 0 Warnings: 0
mysql> select * from gameactive;
+------------+-----------+--------+
| fdate | fgamename | fcount |
+------------+-----------+--------+
| 2020-02-13 | puke | 100 |
| 2020-02-13 | dizhu | 150 |
+------------+-----------+--------+
2 rows in set (0.00 sec)
然后使用sqoop脚本连接mysql和hadoop,将mysql中的这个数据表存入hdfs系统中:
[[email protected] ~]$ sqoop import --connect jdbc:mysql://master:3306/sqoop --username root --password Root-123 --table gameactive --target-dir /stat/test --delete-target-dir --num-mappers 1 --fields-terminated-by ','
上述脚本分为几个部分:
sqoop import \ 采用import命令
--connect jdbc:mysql://master:3306/sqoop 使用jdbc连接mysql数据库,主机名为master,端口为3306,sqoop为mysql中的数据库名
--username root 连接时使用的mysql用户名为root
--password Root-123 连接时使用的mysql用户密码
--table gameactive 访问mysql中sqoop数据库里的gameactive数据表
--target-dir /stat/test 将数据表中内容存入hdfs中的stat/test目录下
--deleter-target-dir 如果该目录已存在就删除后再存入
--num-mappers 1 使用mapreduce中的mapper进程,数量为1
--fields-terminated-by ',' 字段之间间隔采用逗号
执行上述脚本,系统会开启hadoop中的mapreduce任务进程处理:
[[email protected] ~]$ sqoop import --connect jdbc:mysql://master:3306/sqoop --username root --password Root-123 --table gameactive --target-dir /stat/test --delete-target-dir --num-mappers 1 --fields-terminated-by ','
Warning: /home/hadoop/sqoop1.4.6/../hbase does not exist! HBase imports will fail.
Please set $HBASE_HOME to the root of your HBase installation.
Warning: /home/hadoop/sqoop1.4.6/../hcatalog does not exist! HCatalog jobs will fail.
Please set $HCAT_HOME to the root of your HCatalog installation.
Warning: /home/hadoop/sqoop1.4.6/../accumulo does not exist! Accumulo imports will fail.
Please set $ACCUMULO_HOME to the root of your Accumulo installation.
Warning: /home/hadoop/sqoop1.4.6/../zookeeper does not exist! Accumulo imports will fail.
Please set $ZOOKEEPER_HOME to the root of your Zookeeper installation.
2020-02-13 14:38:17,349 INFO sqoop.Sqoop: Running Sqoop version: 1.4.6-cdh5.14.0
2020-02-13 14:38:17,386 WARN tool.BaseSqoopTool: Setting your password on the command-line is insecure. Consider using -P instead.
2020-02-13 14:38:17,526 INFO manager.MySQLManager: Preparing to use a MySQL streaming resultset.
2020-02-13 14:38:17,529 INFO tool.CodeGenTool: Beginning code generation
Loading class `com.mysql.jdbc.Driver'. This is deprecated. The new driver class is `com.mysql.cj.jdbc.Driver'. The driver is automatically registered via the SPI and manual loading of the driver class is generally unnecessary.
2020-02-13 14:38:18,426 INFO manager.SqlManager: Executing SQL statement: SELECT t.* FROM `gameactive` AS t LIMIT 1
2020-02-13 14:38:18,525 INFO manager.SqlManager: Executing SQL statement: SELECT t.* FROM `gameactive` AS t LIMIT 1
2020-02-13 14:38:18,534 INFO orm.CompilationManager: HADOOP_MAPRED_HOME is /home/hadoop/hadoop-3.1.2
Note: /tmp/sqoop-hadoop/compile/3acb0490ad7cb85df80adb8d2b955e47/gameactive.java uses or overrides a deprecated API.
Note: Recompile with -Xlint:deprecation for details.
2020-02-13 14:38:20,418 INFO orm.CompilationManager: Writing jar file: /tmp/sqoop-hadoop/compile/3acb0490ad7cb85df80adb8d2b955e47/gameactive.jar
2020-02-13 14:38:21,378 INFO tool.ImportTool: Destination directory /stat/test is not present, hence not deleting.
2020-02-13 14:38:21,378 WARN manager.MySQLManager: It looks like you are importing from mysql.
2020-02-13 14:38:21,378 WARN manager.MySQLManager: This transfer can be faster! Use the --direct
2020-02-13 14:38:21,378 WARN manager.MySQLManager: option to exercise a MySQL-specific fast path.
2020-02-13 14:38:21,378 INFO manager.MySQLManager: Setting zero DATETIME behavior to convertToNull (mysql)
2020-02-13 14:38:21,386 INFO mapreduce.ImportJobBase: Beginning import of gameactive
2020-02-13 14:38:21,387 INFO Configuration.deprecation: mapred.job.tracker is deprecated. Instead, use mapreduce.jobtracker.address
2020-02-13 14:38:21,456 INFO Configuration.deprecation: mapred.jar is deprecated. Instead, use mapreduce.job.jar
2020-02-13 14:38:21,491 INFO Configuration.deprecation: mapred.map.tasks is deprecated. Instead, use mapreduce.job.maps
2020-02-13 14:38:21,768 INFO client.RMProxy: Connecting to ResourceManager at master/192.168.58.159:8032
2020-02-13 14:38:22,841 INFO mapreduce.JobResourceUploader: Disabling Erasure Coding for path: /tmp/hadoop-yarn/staging/hadoop/.staging/job_1581564243812_0001
2020-02-13 14:38:54,794 INFO db.DBInputFormat: Using read commited transaction isolation
2020-02-13 14:38:55,013 INFO mapreduce.JobSubmitter: number of splits:1
2020-02-13 14:38:55,632 INFO mapreduce.JobSubmitter: Submitting tokens for job: job_1581564243812_0001
2020-02-13 14:38:55,634 INFO mapreduce.JobSubmitter: Executing with tokens: []
2020-02-13 14:38:56,023 INFO conf.Configuration: resource-types.xml not found
2020-02-13 14:38:56,023 INFO resource.ResourceUtils: Unable to find 'resource-types.xml'.
2020-02-13 14:38:56,742 INFO impl.YarnClientImpl: Submitted application application_1581564243812_0001
2020-02-13 14:38:56,923 INFO mapreduce.Job: The url to track the job: http://master:8088/proxy/application_1581564243812_0001/
2020-02-13 14:38:56,924 INFO mapreduce.Job: Running job: job_1581564243812_0001
2020-02-13 14:39:34,699 INFO mapreduce.Job: Job job_1581564243812_0001 running in uber mode : false
2020-02-13 14:39:34,702 INFO mapreduce.Job: map 0% reduce 0%
2020-02-13 14:39:48,428 INFO mapreduce.Job: map 100% reduce 0%
2020-02-13 14:39:49,482 INFO mapreduce.Job: Job job_1581564243812_0001 completed successfully
2020-02-13 14:39:49,615 INFO mapreduce.Job: Counters: 32
File System Counters
FILE: Number of bytes read=0
FILE: Number of bytes written=234015
FILE: Number of read operations=0
FILE: Number of large read operations=0
FILE: Number of write operations=0
HDFS: Number of bytes read=87
HDFS: Number of bytes written=41
HDFS: Number of read operations=6
HDFS: Number of large read operations=0
HDFS: Number of write operations=2
Job Counters
Launched map tasks=1
Other local map tasks=1
Total time spent by all maps in occupied slots (ms)=21786
Total time spent by all reduces in occupied slots (ms)=0
Total time spent by all map tasks (ms)=10893
Total vcore-milliseconds taken by all map tasks=10893
Total megabyte-milliseconds taken by all map tasks=22308864
Map-Reduce Framework
Map input records=2
Map output records=2
Input split bytes=87
Spilled Records=0
Failed Shuffles=0
Merged Map outputs=0
GC time elapsed (ms)=404
CPU time spent (ms)=2520
Physical memory (bytes) snapshot=124497920
Virtual memory (bytes) snapshot=3604873216
Total committed heap usage (bytes)=40763392
Peak Map Physical memory (bytes)=124497920
Peak Map Virtual memory (bytes)=3604873216
File Input Format Counters
Bytes Read=0
File Output Format Counters
Bytes Written=41
2020-02-13 14:39:49,631 INFO mapreduce.ImportJobBase: Transferred 41 bytes in 88.1291 seconds (0.4652 bytes/sec)
2020-02-13 14:39:49,646 INFO mapreduce.ImportJobBase: Retrieved 2 records.
处理结束后,可以使用web界面来查看,也可以直接采用hdfs命令来访问/stat/test目录:
[[email protected] ~]$ hdfs dfs -ls /stat/test
Found 2 items
-rw-r--r-- 1 hadoop supergroup 0 2020-02-13 14:39 /stat/test/_SUCCESS
-rw-r--r-- 1 hadoop supergroup 41 2020-02-13 14:39 /stat/test/part-m-00000
[[email protected] ~]$ hdfs dfs -cat /stat/test/part-m-00000
2020-02-13,puke,100
2020-02-13,dizhu,150
结果与mysql中建立的数据完全一致,这样就实现了mysql与hdfs之间的数据导入。
反过来如果将hdfs中数据导入到mysql中,执行脚本语言就得使用export方式。
首先在mysql端建立一个表getData,
mysql> use sqoop
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> create table getData(fdate varchar(64),fgamename varchar(64),fcount int);
Query OK, 0 rows affected (0.78 sec)
mysql> show tables;
+-----------------+
| Tables_in_sqoop |
+-----------------+
| gameactive |
| getData |
+-----------------+
2 rows in set (0.00 sec)
然后回到sqoop端,我们将前面从mysql中导出存储到HDFS中的数据再导回至mysql中,上述数据在hdfs系统存储位置为/stat/test,因此sqoop执行脚本写为:
[[email protected] ~]$ sqoop export --connect jdbc:mysql://master:3306/sqoop --username root --password Root-123 --table getData -m 1 --export-dir '/stat/test' --fields-terminated-by ','
详细看的话:
sqoop export \ 采用export命令
--connect jdbc:mysql://master:3306/sqoop 使用jdbc连接mysql数据库,主机名为master,端口为3306,sqoop为mysql中的数据库名
--username root 连接时使用的mysql用户名为root
--password Root-123 连接时使用的mysql用户密码
--table gameactive 访问mysql中sqoop数据库里的gameactive数据表
--target-dir /stat/test 将hdfs中的stat/test目录下的内容导出到mysql中
--num-mappers 1 使用mapreduce中的mapper进程,数量为1
--fields-terminated-by ',' 字段之间间隔采用逗号
执行结束后,可以去mysql数据中查询:
mysql> select * from getData;
+------------+-----------+--------+
| fdate | fgamename | fcount |
+------------+-----------+--------+
| 2020-02-13 | puke | 100 |
| 2020-02-13 | dizhu | 150 |
+------------+-----------+--------+
2 rows in set (0.04 sec)
可以看到,数据已经存入mysql中了。