[
https://issues.apache.org/jira/browse/IMPALA-1706?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Tim Armstrong updated IMPALA-1706:
----------------------------------
Component/s: (was: Perf Investigation)
Frontend
> Join returning single distinct column unnecessarily computes cross product
> --------------------------------------------------------------------------
>
> Key: IMPALA-1706
> URL: https://issues.apache.org/jira/browse/IMPALA-1706
> Project: IMPALA
> Issue Type: Bug
> Components: Frontend
> Affects Versions: Impala 2.0
> Reporter: Dileep Kumar
> Priority: Minor
> Labels: performance
> Attachments: profile, summary
>
>
> While joining two simple table's on a columns with many duplicate values
> never completes. The query page show 100% progress but no result returned.
> The query is :
> {{select distinct ca_country from customer_address, customer where
> c_birth_country = upper(ca_country);}}
> The table stats are as belowe:
> customer:
> {code}
> Query: show table stats customer
> +---------+--------+----------+--------------+---------+-------------------+
> | #Rows | #Files | Size | Bytes Cached | Format | Incremental stats |
> +---------+--------+----------+--------------+---------+-------------------+
> | 3850000 | 21 | 214.42MB | NOT CACHED | PARQUET | false |
> +---------+--------+----------+--------------+---------+-------------------+
> customer_address:
> Query: show table stats customer_address
> +---------+--------+----------+--------------+--------+-------------------+
> | #Rows | #Files | Size | Bytes Cached | Format | Incremental stats |
> +---------+--------+----------+--------------+--------+-------------------+
> | 1925000 | 21 | 208.55MB | NOT CACHED | TEXT | false |
> +---------+--------+----------+--------------+--------+-------------------+
> {code}
> The query profile and execution summary is attached.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
