基于Hive及Sqoop的每日PV、UV、IP定时分析

[Author]: kwu

1、创建pvuvip的hive表

hive -e "
use stage;
CREATE  EXTERNAL  TABLE pvuvip(
	day string,
	pv int,
	uv int,
	ipcnt int

)
ROW FORMAT DELIMITED FIELDS TERMINATED BY ' '
LOCATION '/dw/stage/pvuvip/';"

2、创建mysql关系库的表

CREATE TABLE
    pvuvip
    (
        id INT NOT NULL AUTO_INCREMENT,
        DAY VARCHAR(50),
        pv INT,
        uv INT,
        ipcnt INT,
        PRIMARY KEY (id)
    )
    ENGINE=MyISAM DEFAULT CHARSET=latin1

3、每日定时执行的自动化脚本

#!/bin/sh
# upload logs to hdfs

yesterday=`date --date='1 days ago' +%Y%m%d`

hive -e "
use stage;
insert overwrite table pvuvip select day,count(*) pv, count(distinct cookieid) uv , count(distinct ip) ipcnt from ods.tracklog where day='${yesterday}' group by day;"

sqoop export  --connect jdbc:mysql://localhost:3306/charts  --username root --password 123456 --table pvuvip --fields-terminated-by ' ' --columns "day,pv,uv,ipcnt" --export-dir /dw/stage/pvuvip/;

4、在crontab中增加定时任务

18  06 * * * /opt/bin/sqoop_opt/pvuvip.opt

5、刷新定时任务

/sbin/service crond reload