[
https://issues.apache.org/jira/browse/HIVE-21574?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Panda Song updated HIVE-21574:
------------------------------
Description:
when I use a table instead of the sub select,I get the right result,much more
rows are joined together(metrics old_uv is bigger!!!)
Is there some bugs here?
Please help me ,thanks a lot!!!!!!
{code:java}
select
a.event_date,
count(distinct a.device_id) as uv,
count(distinct case when b.device_id is not null then b.device_id end) as
old_uv,
count(distinct a.device_id) - count(distinct case when b.device_id is not null
then b.device_id end) as new_uv
from
(
select
event_date,
device_id,
qingting_id
from datacenter.bl_page_chain_day
where event_date = '2019-03-31'
and (current_content like 'https://a.qingting.fm/membership5%'
or current_content like 'https://m.qingting.fm/vips/members%'
or current_content like 'https://sss.qingting.fm/vips/members/v2/%')
)a
left join
(select
b.device_id
from
lzq_test.first_buy_vip a
inner join datacenter.device_qingting b on a.qingting_id = b.qingting_id
where a.first_buy < '2019-03-31'
group by b.device_id
)b
on a.device_id = b.device_id
group by a.event_date;
{code}
plan:
{code:java}
Plan optimized by CBO.
Vertex dependency in root stage
Map 1 <- Map 3 (BROADCAST_EDGE)
Reducer 2 <- Map 1 (SIMPLE_EDGE)
Reducer 5 <- Map 4 (CUSTOM_SIMPLE_EDGE), Reducer 2 (ONE_TO_ONE_EDGE)
Reducer 6 <- Reducer 5 (SIMPLE_EDGE)
Stage-0
Fetch Operator
limit:-1
Stage-1
Reducer 6
File Output Operator [FS_26]
Select Operator [SEL_25] (rows=35527639 width=349)
Output:["_col0","_col1","_col2","_col3"]
Group By Operator [GBY_24] (rows=35527639 width=349)
Output:["_col0","_col1","_col2"],aggregations:["count(DISTINCT
KEY._col1:0._col0)","count(DISTINCT KEY._col1:1._col0)"],keys:KEY._col0
<-Reducer 5 [SIMPLE_EDGE]
SHUFFLE [RS_23]
PartitionCols:_col0
Group By Operator [GBY_22] (rows=71055278 width=349)
Output:["_col0","_col1","_col2","_col3","_col4"],aggregations:["count(DISTINCT
_col1)","count(DISTINCT _col2)"],keys:true, _col1, _col2
Select Operator [SEL_20] (rows=71055278 width=349)
Output:["_col1","_col2"]
Map Join Operator [MAPJOIN_45] (rows=71055278 width=349)
Conds:RS_17.KEY.reducesinkkey0=RS_18.KEY.reducesinkkey0(Right
Outer),Output:["_col0","_col1"]
<-Reducer 2 [ONE_TO_ONE_EDGE]
FORWARD [RS_17]
PartitionCols:_col0
Group By Operator [GBY_12] (rows=21738609 width=235)
Output:["_col0"],keys:KEY._col0
<-Map 1 [SIMPLE_EDGE]
SHUFFLE [RS_11]
PartitionCols:_col0
Group By Operator [GBY_10] (rows=43477219 width=235)
Output:["_col0"],keys:_col0
Map Join Operator [MAPJOIN_44] (rows=43477219
width=235)
Conds:SEL_2._col1=RS_7._col0(Inner),Output:["_col0"]
<-Map 3 [BROADCAST_EDGE]
BROADCAST [RS_7]
PartitionCols:_col0
Select Operator [SEL_5] (rows=301013
width=228)
Output:["_col0"]
Filter Operator [FIL_32] (rows=301013
width=228)
predicate:((first_buy < DATE'2019-03-31')
and qingting_id is not null)
TableScan [TS_3] (rows=1062401 width=228)
lzq_test@first_buy_vip,a, transactional
table,Tbl:COMPLETE,Col:NONE,Output:["qingting_id","first_buy"]
<-Select Operator [SEL_2] (rows=39524744
width=235)
Output:["_col0","_col1"]
Filter Operator [FIL_31] (rows=39524744
width=235)
predicate:qingting_id is not null
TableScan [TS_0] (rows=39524744 width=235)
datacenter@device_qingting,b, ACID
table,Tbl:COMPLETE,Col:COMPLETE,Output:["device_id","qingting_id"]
<-Map 4 [CUSTOM_SIMPLE_EDGE]
PARTITION_ONLY_SHUFFLE [RS_18]
PartitionCols:_col0
Select Operator [SEL_16] (rows=64595706 width=349)
Output:["_col0"]
Filter Operator [FIL_33] (rows=64595706 width=349)
predicate:((current_content like
'https://a.qingting.fm/membership5%') or (current_content like
'https://m.qingting.fm/vips/members%') or (current_content like
'https://sss.qingting.fm/vips/members/v2/%'))
TableScan [TS_14] (rows=64595706 width=349)
datacenter@bl_page_chain_day,bl_page_chain_day,Tbl:COMPLETE,Col:NONE,Output:["device_id","current_content"]
{code}
was:
when I use a table instead of the sub select,I get the right result,much more
rows are joined together(metrics old_uv is bigger!!!)
Is there some bugs here?
Please help me ,thanks a lot!!!!!!
{code:java}
select
a.event_date,
count(distinct a.device_id) as uv,
count(distinct case when b.device_id is not null then b.device_id end) as
old_uv,
count(distinct a.device_id) - count(distinct case when b.device_id is not null
then b.device_id end) as new_uv
from
(
select
event_date,
device_id,
qingting_id
from datacenter.bl_page_chain_day
where event_date = '2019-03-31'
and (current_content like 'https://a.qingting.fm/membership5%'
or current_content like 'https://m.qingting.fm/vips/members%'
or current_content like 'https://sss.qingting.fm/vips/members/v2/%')
)a
left join
(select
b.device_id
from
lzq_test.first_buy_vip a
inner join datacenter.device_qingting b on a.qingting_id = b.qingting_id
where a.first_buy < '2019-03-31'
group by b.device_id
)b
on a.device_id = b.device_id
group by a.event_date;
{code}
plan:
{code:java}
Plan optimized by CBO. |
Vertex dependency in root stage
Map 1 <- Map 3 (BROADCAST_EDGE)
Reducer 2 <- Map 1 (SIMPLE_EDGE)
Reducer 5 <- Map 4 (CUSTOM_SIMPLE_EDGE), Reducer 2 (ONE_TO_ONE_EDGE)
Reducer 6 <- Reducer 5 (SIMPLE_EDGE)
Stage-0
Fetch Operator
limit:-1
Stage-1
Reducer 6
File Output Operator [FS_26]
Select Operator [SEL_25] (rows=35527639 width=349)
Output:["_col0","_col1","_col2","_col3"]
Group By Operator [GBY_24] (rows=35527639 width=349)
Output:["_col0","_col1","_col2"],aggregations:["count(DISTINCT
KEY._col1:0._col0)","count(DISTINCT KEY._col1:1._col0)"],keys:KEY._col0
<-Reducer 5 [SIMPLE_EDGE]
SHUFFLE [RS_23]
PartitionCols:_col0
Group By Operator [GBY_22] (rows=71055278 width=349)
Output:["_col0","_col1","_col2","_col3","_col4"],aggregations:["count(DISTINCT
_col1)","count(DISTINCT _col2)"],keys:true, _col1, _col2
Select Operator [SEL_20] (rows=71055278 width=349)
Output:["_col1","_col2"]
Map Join Operator [MAPJOIN_45] (rows=71055278 width=349)
Conds:RS_17.KEY.reducesinkkey0=RS_18.KEY.reducesinkkey0(Right
Outer),Output:["_col0","_col1"]
<-Reducer 2 [ONE_TO_ONE_EDGE]
FORWARD [RS_17]
PartitionCols:_col0
Group By Operator [GBY_12] (rows=21738609 width=235)
Output:["_col0"],keys:KEY._col0
<-Map 1 [SIMPLE_EDGE]
SHUFFLE [RS_11]
PartitionCols:_col0
Group By Operator [GBY_10] (rows=43477219 width=235)
Output:["_col0"],keys:_col0
Map Join Operator [MAPJOIN_44] (rows=43477219
width=235)
Conds:SEL_2._col1=RS_7._col0(Inner),Output:["_col0"]
<-Map 3 [BROADCAST_EDGE]
BROADCAST [RS_7]
PartitionCols:_col0
Select Operator [SEL_5] (rows=301013
width=228)
Output:["_col0"]
Filter Operator [FIL_32] (rows=301013
width=228)
predicate:((first_buy < DATE'2019-03-31')
and qingting_id is not null)
TableScan [TS_3] (rows=1062401 width=228)
lzq_test@first_buy_vip,a, transactional
table,Tbl:COMPLETE,Col:NONE,Output:["qingting_id","first_buy"]
<-Select Operator [SEL_2] (rows=39524744
width=235)
Output:["_col0","_col1"]
Filter Operator [FIL_31] (rows=39524744
width=235)
predicate:qingting_id is not null
TableScan [TS_0] (rows=39524744 width=235)
datacenter@device_qingting,b, ACID
table,Tbl:COMPLETE,Col:COMPLETE,Output:["device_id","qingting_id"]
<-Map 4 [CUSTOM_SIMPLE_EDGE]
PARTITION_ONLY_SHUFFLE [RS_18]
PartitionCols:_col0
Select Operator [SEL_16] (rows=64595706 width=349)
Output:["_col0"]
Filter Operator [FIL_33] (rows=64595706 width=349)
predicate:((current_content like
'https://a.qingting.fm/membership5%') or (current_content like
'https://m.qingting.fm/vips/members%') or (current_content like
'https://sss.qingting.fm/vips/members/v2/%'))
TableScan [TS_14] (rows=64595706 width=349)
datacenter@bl_page_chain_day,bl_page_chain_day,Tbl:COMPLETE,Col:NONE,Output:["device_id","current_content"]
{code}
> return wrong result when excuting left join sql
> -----------------------------------------------
>
> Key: HIVE-21574
> URL: https://issues.apache.org/jira/browse/HIVE-21574
> Project: Hive
> Issue Type: Bug
> Affects Versions: 3.1.0
> Environment: hive 3.1.0 hdfs 3.1.1
> Reporter: Panda Song
> Priority: Blocker
>
> when I use a table instead of the sub select,I get the right result,much more
> rows are joined together(metrics old_uv is bigger!!!)
> Is there some bugs here?
> Please help me ,thanks a lot!!!!!!
> {code:java}
> select
> a.event_date,
> count(distinct a.device_id) as uv,
> count(distinct case when b.device_id is not null then b.device_id end) as
> old_uv,
> count(distinct a.device_id) - count(distinct case when b.device_id is not
> null then b.device_id end) as new_uv
> from
> (
> select
> event_date,
> device_id,
> qingting_id
> from datacenter.bl_page_chain_day
> where event_date = '2019-03-31'
> and (current_content like 'https://a.qingting.fm/membership5%'
> or current_content like 'https://m.qingting.fm/vips/members%'
> or current_content like 'https://sss.qingting.fm/vips/members/v2/%')
> )a
> left join
> (select
> b.device_id
> from
> lzq_test.first_buy_vip a
> inner join datacenter.device_qingting b on a.qingting_id = b.qingting_id
> where a.first_buy < '2019-03-31'
> group by b.device_id
> )b
> on a.device_id = b.device_id
> group by a.event_date;
> {code}
> plan:
> {code:java}
> Plan optimized by CBO.
>
> Vertex dependency in root stage
> Map 1 <- Map 3 (BROADCAST_EDGE)
> Reducer 2 <- Map 1 (SIMPLE_EDGE)
> Reducer 5 <- Map 4 (CUSTOM_SIMPLE_EDGE), Reducer 2 (ONE_TO_ONE_EDGE)
> Reducer 6 <- Reducer 5 (SIMPLE_EDGE)
>
> Stage-0
> Fetch Operator
> limit:-1
> Stage-1
> Reducer 6
> File Output Operator [FS_26]
> Select Operator [SEL_25] (rows=35527639 width=349)
> Output:["_col0","_col1","_col2","_col3"]
> Group By Operator [GBY_24] (rows=35527639 width=349)
> Output:["_col0","_col1","_col2"],aggregations:["count(DISTINCT
> KEY._col1:0._col0)","count(DISTINCT KEY._col1:1._col0)"],keys:KEY._col0
> <-Reducer 5 [SIMPLE_EDGE]
> SHUFFLE [RS_23]
> PartitionCols:_col0
> Group By Operator [GBY_22] (rows=71055278 width=349)
>
> Output:["_col0","_col1","_col2","_col3","_col4"],aggregations:["count(DISTINCT
> _col1)","count(DISTINCT _col2)"],keys:true, _col1, _col2
> Select Operator [SEL_20] (rows=71055278 width=349)
> Output:["_col1","_col2"]
> Map Join Operator [MAPJOIN_45] (rows=71055278 width=349)
>
> Conds:RS_17.KEY.reducesinkkey0=RS_18.KEY.reducesinkkey0(Right
> Outer),Output:["_col0","_col1"]
> <-Reducer 2 [ONE_TO_ONE_EDGE]
> FORWARD [RS_17]
> PartitionCols:_col0
> Group By Operator [GBY_12] (rows=21738609 width=235)
> Output:["_col0"],keys:KEY._col0
> <-Map 1 [SIMPLE_EDGE]
> SHUFFLE [RS_11]
> PartitionCols:_col0
> Group By Operator [GBY_10] (rows=43477219
> width=235)
> Output:["_col0"],keys:_col0
> Map Join Operator [MAPJOIN_44] (rows=43477219
> width=235)
>
> Conds:SEL_2._col1=RS_7._col0(Inner),Output:["_col0"]
> <-Map 3 [BROADCAST_EDGE]
> BROADCAST [RS_7]
> PartitionCols:_col0
> Select Operator [SEL_5] (rows=301013
> width=228)
> Output:["_col0"]
> Filter Operator [FIL_32] (rows=301013
> width=228)
> predicate:((first_buy <
> DATE'2019-03-31') and qingting_id is not null)
> TableScan [TS_3] (rows=1062401
> width=228)
> lzq_test@first_buy_vip,a,
> transactional table,Tbl:COMPLETE,Col:NONE,Output:["qingting_id","first_buy"]
> <-Select Operator [SEL_2] (rows=39524744
> width=235)
> Output:["_col0","_col1"]
> Filter Operator [FIL_31] (rows=39524744
> width=235)
> predicate:qingting_id is not null
> TableScan [TS_0] (rows=39524744 width=235)
> datacenter@device_qingting,b, ACID
> table,Tbl:COMPLETE,Col:COMPLETE,Output:["device_id","qingting_id"]
> <-Map 4 [CUSTOM_SIMPLE_EDGE]
> PARTITION_ONLY_SHUFFLE [RS_18]
> PartitionCols:_col0
> Select Operator [SEL_16] (rows=64595706 width=349)
> Output:["_col0"]
> Filter Operator [FIL_33] (rows=64595706 width=349)
> predicate:((current_content like
> 'https://a.qingting.fm/membership5%') or (current_content like
> 'https://m.qingting.fm/vips/members%') or (current_content like
> 'https://sss.qingting.fm/vips/members/v2/%'))
> TableScan [TS_14] (rows=64595706 width=349)
>
> datacenter@bl_page_chain_day,bl_page_chain_day,Tbl:COMPLETE,Col:NONE,Output:["device_id","current_content"]
>
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
