[
https://issues.apache.org/jira/browse/HIVE-16509?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16000056#comment-16000056
]
Sahil Takiar commented on HIVE-16509:
-------------------------------------
Going to close this as not an issue. The issue described happens with regular
HoS explain plans, and in Hive-on-MR. It's confusing to me that there would be
a section in the explain plan called {{Local Work}} and it would be empty, but
maybe thats by design.
Hive-on-Tez doesn't have this issue since it doesn't use the
{{MapredLocalTask}}.
> Improve HoS user-level explain plan for Local Work and Map Reduce Local
> -----------------------------------------------------------------------
>
> Key: HIVE-16509
> URL: https://issues.apache.org/jira/browse/HIVE-16509
> Project: Hive
> Issue Type: Bug
> Components: Spark
> Reporter: Sahil Takiar
> Assignee: Sahil Takiar
>
> The current user-level explain plans (this may affect regular explain plans
> too, haven't checked) for HoS create empty entries for {{Local Work}} and
> {{Map Reduce Local}}
> The sections for local work are largely empty. We should add more info this
> section (or remove if it isn't needed).
> Example:
> {code}
> Vertex dependency in root stage
> Reducer 2 <- Map 1 (GROUP)
> Reducer 3 <- Reducer 2 (GROUP)
> Stage-0
> Fetch Operator
> limit:-1
> Stage-1
> Reducer 3
> File Output Operator [FS_14]
> Group By Operator [GBY_12] (rows=1 width=16)
>
> Output:["_col0","_col1"],aggregations:["sum(VALUE._col0)","sum(VALUE._col1)"]
> <-Reducer 2 [GROUP]
> GROUP [RS_11]
> Group By Operator [GBY_10] (rows=1 width=16)
>
> Output:["_col0","_col1"],aggregations:["sum(VALUE._col0)","sum(VALUE._col1)"]
> <-Map 1 [GROUP]
> GROUP [RS_9]
> PartitionCols:rand()
> Select Operator [SEL_7] (rows=1 width=33)
> Output:["_col0","_col1"]
> Map Join Operator [MAPJOIN_15] (rows=1 width=33)
> Conds:SEL_1.(UDFToDouble(_col0) +
> 1.0)=SEL_1.UDFToDouble(_col0)(Left Outer),Output:["_col0","_col2"]
> <-Select Operator [SEL_1] (rows=1 width=30)
> Output:["_col0"]
> TableScan [TS_0] (rows=1 width=30)
> default@t1,k,Tbl:COMPLETE,Col:NONE,Output:["key"]
> Map Reduce Local Work
> Stage-2
> Map 4
> keys: [HASHTABLESINK_17]
> 0(UDFToDouble(_col0) + 1.0),1UDFToDouble(_col0)
> Select Operator [SEL_3] (rows=1 width=30)
> Output:["_col0","_col1"]
> TableScan [TS_2] (rows=1 width=30)
> default@t1,v,Tbl:COMPLETE,Col:NONE,Output:["key","val"]
> Map Reduce Local Work
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
