@犀牛
--变化数据
SELECT
cod_cust_id
,amt_credit_new
,dat_create
FROM (
SELECT
cod_cust_id
,amt_credit_new
,dat_create
,ROW_NUMBER()
OVER (PARTITION BY cod_cust_id,SUBSTRING(dat_create,1,10) ORDER BY dat_create DESC ) AS rn--如果客户在同一天多次申请通过，则取日期最大那次
FROM thai_dw.snap_cmisdb_cmis_card_limit_change
WHERE flg_status_apply =8
)
WHERE rn =1
;

--没有发生变化的数据
SELECT
p2.cod_cust_id
,p2.amt_credit
,p2.dat_create
FROM (
SELECT
cod_cust_id
,amt_credit_new
FROM (
SELECT
cod_cust_id
,amt_credit_new
,dat_create
,ROW_NUMBER()
OVER (PARTITION BY cod_cust_id,SUBSTRING(dat_create,1,10) ORDER BY dat_create DESC ) AS rn--如果客户在同一天多次申请通过，则取日期最大那次
FROM thai_dw.snap_cmisdb_cmis_card_limit_change
WHERE flg_status_apply =8
)
WHERE rn =1
) p1
RIGHT JOIN (
SELECT
cod_cust_id
,amt_credit
,dat_create
FROM thai_dw.snap_cmisdb_cmis_card_apply
WHERE flg_status_apply =8
) p2
ON p1.cod_cust_id = p2.cod_cust_id
WHERE p1.cod_cust_id IS NULL
;

我以数据创建时间dat_create作为starttime,什么为endtime呢？没有发生变化的数据mark都是i，发生变化的数据mark都是u可以吗？因为不会有数据被删除

@犀牛 感谢 我尝试实现一下

同样是格式类型转换。orc转为hfile也不用reduce聚合，为何不设置reduce数量为0也不报错呢？

已解决，reduce数量设置为0

@苏牛

本地跑也是同样错误

通过-D参数传递zk地址已解决问题

已解决 host文件出错