行数不匹配

Question

我的查询

insert into table hist
Select subs2.SUBS_KEY
, subs2.FACT_DATE
, subs2.TYPE
, subs2.BALANCE
, '${DATE_1D}' as TIME_KEY
from subs2
where CNT < 15 and AMT > 0;

其中'${DATE_1D}'为hivevar DATE_1D=2017-02-28

执行后，行数从

开始

db.hist stats: [numFiles=22168, numRows=254582570, totalSize=19188669803, rawDataSize=19613070967]

到

db.hist stats: [numFiles=22268, numRows=257376901, totalSize=19386949977, rawDataSize=19808556810]

差异= 2794331行

但是当我做的时候

insert into table db.new_hist
select *
from db.hist
where time_key = '2017-02-28';

我得到了

db.new_hist stats: [numFiles=201, numRows=178798, totalSize=2227208, rawDataSize=12510497]

我使用new_hist

创建了create table like

show create

的结果

CREATE TABLE `db.hist`(
  `subs_key` string,
  `fact_date` string,
  `type` int,
  `balance` double,
  `time_key` string)
ROW FORMAT SERDE
  'org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe'
STORED AS INPUTFORMAT
  'org.apache.hadoop.mapred.TextInputFormat'
OUTPUTFORMAT
  'org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat'

为什么行数不匹配？

我做了

Select count(*)
from hist

有没有

set hive.compute.query.using.stats=true;

导致两种情况= 246210052

由于某种原因，插入日志显示错误的行数。