[ 
https://issues.apache.org/jira/browse/PARQUET-1206?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Gabor Szadovszky resolved PARQUET-1206.
---------------------------------------
    Resolution: Invalid

There is a huge difference between Hive and Impala. Hive uses parquet-mr but 
Impala uses its own Parquet implementation. Moreover, TBLPROPERTIES are 
interpreted by the SQL engine and not parquet-mr directly.

You should create a Jira for Impala or Hive instead.

> Parquet properties are ignored at table creation time
> -----------------------------------------------------
>
>                 Key: PARQUET-1206
>                 URL: https://issues.apache.org/jira/browse/PARQUET-1206
>             Project: Parquet
>          Issue Type: Bug
>          Components: parquet-mr
>    Affects Versions: 1.6.0
>         Environment: Versions: Impalad version 2.6.0-cdh5.8.5 RELEASE (build 
> d460423934c1689082f6297101a45b28c49bd39f)
>  
> CDH 5.8.5 
>            Reporter: Istvan Szukacs
>            Priority: Major
>
> Create table ignores table properties.
>  
> {{CREATE EXTERNAL TABLE db.test (}}{{  month INT}}
> {{)}}
> {{STORED AS PARQUET}}
> {{LOCATION 'hdfs://nn/user/user/test'}}
> {{TBLPROPERTIES (}}
> {{  
> 'STATS_GENERATED_VIA_STATS_TASK'='true',}}{{'PARQUET.COMPRESSION'='GZIP',}}{{'PARQUET.VERSION'='PARQUET_2_0');}}
>  
> After inserting into this table with the following:
> {{INSERT INTO TABLE db.test}}
> {{SELECT * FROM db.source;}}
> The created Parquet files are SNAPPY compressed according to parquet-tools 
> meta. Is there a difference which query engine is used? Impala or Hive? 
> Should this be resulting GZIP compressed files or Parquet ignores 
> TBLPROPERTIES?
>  
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to