首页 > 数据库技术 > 详细

hive的sql语句小节

时间:2019-02-13 15:17:36      阅读:189      评论:0      收藏:0      [点我收藏+]
20180615:

Hive是基于Hadoop的一个数据仓库工具,是一种逻辑上的数据库,依赖hdfs文件系统;它把实际的数据文件映射为一张数据库表来作为元数据来管理hdfs上的数据。Hive存在的意义并非数据存储,而是被用来处理数据的,它计算的实质是用sql调用了底层的mapreduce。因为hive的sql的学习成本比较低,几乎和mysql这些数据库近似,所以使用极其广泛。很多公司使用hive几乎都能够撑起整个项目。

 

cd bigdata-Guanxiyingxiao/Guanxiyingxiao/2018

 

get Guanxiyingxiao.zip

 

 

 

 

 

cd  dawen.chi/product_city_staypoint_data/2018

 

get  product_city_staypoint_data.zip

 

 

 

cd   ctyun-product-tianyuan/city-plan-paltform/2018

 

get  city-plan-paltform.zip

 

 

 

 

 

清空hive表:

 

insert overwrite table dal_bdcsc_bad_black_list_m select * from t_table1 where 1=0;

 

 

 

md5加密:

 

 

 

 

愛癡 2018/6/25 15:25:21

 

 

 

yarn queue -status pro2

 

愛癡 2018/6/25 15:25:32

 

 

 

你用这个看看队列资源

 

 

 

秦文杰 2018/6/25 15:26:02

 

 

 

技术分享图片

 

愛癡 2018/6/25 15:26:06

 

 

 

hadoop job -list | 队列名   用这个命令看看你现在的队列上面有几个任务

 

15:26:35

 

愛癡 2018/6/25 15:26:35

 

 

 

hadoop job -list | grep test1

 

 

 

秦文杰 2018/6/25 15:27:05

 

 

 

技术分享图片

 

 

 

INSERT OVERWRITE table csqinwj.dal_bdcsc_bad_black_list_m PARTITION(PROV_ID, MONTH_ID)

 

select g.accs_nbr,

 

       g.cert_nbr,

 

       g.cust_name,

 

       g.prov_id,

 

       201705 MONTH_ID  

 

from (select accs_nbr,

 

             cert_nbr,

 

                         cust_name,

 

                         uninstall_date,

 

                         prov_id

 

           from CSQINWJ.DWI_SEV_USER_MAIN_INFO_PRO_MSK_M

 

           where month_id=‘201804‘ and cert_type=‘1‘)g --证件类型为身份证

 

INNER join

 

        (select MD5(t.accs_nbr) accs_nbr,

 

                substr(t.biz_event_time,1,8) time

 

          from(select b.accs_nbr,

 

                  b.biz_event_time,

 

                  max(case when a.rn=b.rn+1 then a.order_type end) type_a,

 

                  max(case when a.rn=b.rn+2 then a.order_type end) type_b

 

            from (select a.accs_nbr,

 

                          a.user_status,

 

                          a.order_type,

 

                          a.biz_event_time,

 

                          row_number() over(partition by a.accs_nbr order by a.biz_event_time desc) as rn

 

                    from (select  accs_nbr,

 

                                  user_status,

 

                                  order_type,

 

                                  biz_event_time,

 

                                  row_number() over(partition by accs_nbr order by biz_event_time desc) as rn

 

                                  from csqinwj.dwi_sev_user_vsop_decode_d_mid where MONTH_ID<=201705) a

 

                    left join (select accs_nbr,

 

                                      user_status,

 

                                      order_type,

 

                                      biz_event_time,

 

                                      row_number() over(partition by accs_nbr order by biz_event_time desc) as rn

 

                                from csqinwj.dwi_sev_user_vsop_decode_d_mid where MONTH_ID<=201705) b

 

                    on a.accs_nbr = b.accs_nbr and a.rn = b.rn + 1

 

                    where a.order_type <> nvl(b.order_type, ‘0‘)) a

 

            inner join (select * from (select a.accs_nbr,

 

                                       a.user_status,

 

                                       a.order_type,

 

                                       a.biz_event_time,

 

                                       row_number() over(partition by a.accs_nbr order by a.biz_event_time desc) as rn

 

                                       from (select accs_nbr,

 

                                                    user_status,

 

                                                    order_type,

 

                                                    biz_event_time,

 

                                                    row_number() over(partition by accs_nbr order by biz_event_time desc) as rn

 

                                                    from csqinwj.dwi_sev_user_vsop_decode_d_mid where MONTH_ID<=201705) a

 

                                              left join (select accs_nbr,

 

                                                               user_status,

 

                                                               order_type,

 

                                                               biz_event_time,

 

                                                               row_number() over(partition by accs_nbr order by biz_event_time desc) as rn

 

                                                               from csqinwj.dwi_sev_user_vsop_decode_d_mid where MONTH_ID<=201705) b

 

                                              on a.accs_nbr = b.accs_nbr

 

                                              and a.rn = b.rn + 1

 

                                              where a.order_type <> nvl(b.order_type, ‘0‘))c where c.order_type=‘300‘ --销户/拆机

 

                                                                                          ) b

 

                                on a.accs_nbr=b.accs_nbr

 

                                group by b.accs_nbr,b.biz_event_time,b.order_type)t

 

                where t.type_a=‘48‘ or  --欠费双停

 

              (t.type_a=‘301‘ --预约销户

 

                      and t.type_b=‘48‘)) h

 

        where g.accs_nbr=h.accs_nbr and substr(g.uninstall_date,1,8)=h.time;

 

 

 

hive的sql语句小节

原文:https://www.cnblogs.com/jie-2019/p/10369634.html

(0)
(0)
   
举报
评论 一句话评论(0
关于我们 - 联系我们 - 留言反馈 - 联系我们:wmxa8@hotmail.com
© 2014 bubuko.com 版权所有
打开技术之扣,分享程序人生!