hive不分区增量更新

insert overwrite table ods.zeg_so 
select 
*,
case when zsm.id is not null then cast(current_timestamp as string) else zs.etl_update end etl_update 
from ods.zeg_so_mid zsm 
full join ods.zeg_so zs on zsm.id=zs.id
----------------------------------------------------------
insert overwrite table data_center.test_no_partition 
select tmp.id,tmp.name,tmp.age
from tmp.temp_test_no_partition tmp
full join data_center.test_no_partition org
on tmp.id=org.id

#hive全连接

insert overwrite table data_center.test_no_partition 
select tmp.id,tmp.name,tmp.age from tmp.temp_test_no_partition tmp 
union all 
select org.id,org.name,org.age from data_center.test_no_partition org 
left outer join tmp.temp_test_no_partition b 
on org.id=b.id
-------------------------------------------------------------- 
insert overwrite table data_center.test_no_partition 
select tmp.id,tmp.name,tmp.age from tmp.temp_test_no_partition tmp 
union all 
select b.id,b.name,b.age from data_center.test_no_partition org 
left outer join tmp.temp_test_no_partition b 
on org.id=b.id
----------------------------------------------------------------
insert overwrite table data_center.test_no_partition 
select tmp.id,tmp.name,tmp.age from tmp.temp_test_no_partition tmp
full join data_center.test_no_partition org on tmp.id=org.id
--------------------------------------------------------------------
insert overwrite table data_center.test_no_partition 
select org.id,org.name,org.age from tmp.temp_test_no_partition tmp
full join data_center.test_no_partition org on tmp.id=org.id
-------------------------------------------------------------------------------- 
不分区增量更新 OK
insert overwrite table data_center.test_no_partition 
select tmp.id,tmp.name,tmp.age from tmp.temp_test_no_partition tmp 
union all 
select org.id,org.name,org.age from data_center.test_no_partition org 
left outer join tmp.temp_test_no_partition b 
on org.id=b.id
where b.id is null;

分区：增量更新
INSERT OVERWRITE TABLE %s.%s PARTITION(%s) select %s from %s''' % (job_info_map["w_database"], job_info_map["w_table"], job_info_map["w_partition_name"],
colums_str, "temp." + job_info_map["temp_table"])

hive_sql = "set hive.exec.dynamic.partition.mode=nonstrict;"
hive_sql += "set hive.exec.max.dynamic.partitions.pernode=1000;"
hive_sql += "set mapreduce.reduce.shuffle.input.buffer.percent=0.5;"
hive_sql += "INSERT OVERWRITE TABLE " + hive_table + " PARTITION(" + job_info_map["w_partition_name"] + ")" + "\nSELECT " + colums_str + " FROM " + temp_table + " UNION ALL SELECT a.* FROM " + hive_table + " a LEFT OUTER JOIN " + temp_table + " b on "

#增量取数据条件
update_time>=subdate(current_date,1)
COALESCE(update_time,create_Time)>=subdate(current_date,1)

-------------------------------------------------------------------
#分区增量更新语句：
insert overwrite table data_center.test_partition partition(date_id='2017-06-13')
select tmp.id,tmp.name,tmp.age from tmp.temp_test_partition tmp
union all select org.id,org.name,org.age from data_center.test_partition org
left outer join tmp.temp_test_partition b
on org.id=b.id
where b.id is null;
报错：SemanticException [Error 10044]: Line 1:23 Cannot insert into target table because column number/types are different ''2017-06-13'': Table insclause-0 has 3 columns, 
but query has 4 columns

CREATE TABLE `tmp.temp_test_partition`( 
`id` int, 
`name` 
string, 
`age` int) 
ROW FORMAT SERDE 
'org.apache.hadoop.hive.serde2.l 
azy.LazySimpleSerDe' 
WITH SERDEPROPERTIES ( 'field.delim'=',', 'serialization.format'=',') 
STORED AS INPUTFORMAT 
'org.apache.hadoop.mapred.TextInputF 
ormat' 
OUTPUTFORMAT 
'org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFor 
mat' 
LOCATION 
'hdfs://master:9000/user/hive/warehouse/tmp.db/temp_test_partiti 
on' TBLPROPERTIES ( 'transient_lastDdlTime'='1497492633')

alter table tmp.temp_test_partition set serdeproperties ('field.delim'=',','serialization.format'=',')

#分区增量更新语句：OK
set hive.exec.dynamic.partition.mode=nonstrict;
set hive.exec.max.dynamic.partitions.pernode=1000;
set mapreduce.reduce.shuffle.input.buffer.percent=0.5;
insert overwrite table data_center.test_partition partition(date_id) 
select tmp.id,tmp.name,tmp.age,'2017-06-15' as date_id from tmp.temp_test_partition tmp 
union all select a.* from data_center.test_partition a
left outer join tmp.temp_test_partition b on a.id=b.id where b.id is null and a.date_id in ('2017-06-14','2017-06-15','2017-06-16');

#删除部分数据
set hive.exec.dynamic.partition.mode=nonstrict;
set hive.exec.max.dynamic.partitions.pernode=1000;
set mapreduce.reduce.shuffle.input.buffer.percent=0.5;
insert overwrite table data_center.test_partition partition (date_id) select id,name,age,date_id from data_center.test_partition limit 2;

set hive.exec.dynamic.partition.mode=nonstrict;
set hive.exec.max.dynamic.partitions.pernode=1000;
set mapreduce.reduce.shuffle.input.buffer.percent=0.5;
insert overwrite table data_center.test_partition partition (date_id) select id,name,age,date_id from data_center.test_partition where name in ('lisi','ccc')