kylin-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alberto Ramón <a.ramonporto...@gmail.com>
Subject Re: Codec of decimal (10,6)
Date Sat, 26 Nov 2016 13:14:30 GMT
@ShaoFeng,  You're right
Queries have same result in Hive and Kylin.

*I only have has a problem with MAX* (don't work. But sum, min, avg works
OK) (apache-kylin-1.6.0-SNAPSHOT-bin RC1)

SELECT
    SUM (FACT_VALORACIONES.VALORACION) as SUM_Valoracion
    ,MIN (FACT_VALORACIONES.VALORACION) as MIN_Valoracion
--    ,MAX (FACT_VALORACIONES.VALORACION) as MAX_Valoracion
    ,AVG (FACT_VALORACIONES.VALORACION) as AVG_Valoracion

FROM HERR_BANK.FACT_VALORACIONES as FACT_VALORACIONES
    INNER JOIN HERR_BANK.DIM_FECHAS as DIM_FECHAS
        ON FACT_VALORACIONES.IDFECHAVALORACION = DIM_FECHAS.IDFECHA
Group by DIM_FECHAS.ANYO


ERROR:
 *Can't find any realization. Please confirm with providers*. SQL digest:
fact table HERR_BANK.FACT_VALORACIONES,group by
[HERR_BANK.DIM_FECHAS.ANYO],filter on [],with aggregates[FunctionDesc
[expression=SUM, parameter=ParameterDesc [type=column, value=VALORACION,
nextParam=null], returnType=null], FunctionDesc [expression=COUNT,
parameter=ParameterDesc [type=column, value=VALORACION, nextParam=null],
returnType=null], FunctionDesc [expression=MIN, parameter=ParameterDesc
[type=column, value=VALORACION, nextParam=null], returnType=null],
FunctionDesc [expression=MAX, parameter=ParameterDesc [type=column,
value=VALORACION, nextParam=null], returnType=null]].


The result must be:
[image: Imágenes integradas 1]

Measure definition:
[image: Imágenes integradas 2]

2016-11-26 8:20 GMT+01:00 ShaoFeng Shi <shaofengshi@apache.org>:

> Hi Alberto,
>
> User need aware that Cube only has aggregated data, no raw data; at the
> very begining Kylin will throw error on query like "select * "; but to
> provide a better user experience (also to support some BI tools which need
> load a subset data to warm up), Kylin answers such query from the base
> cuboid (group by all dimensions). The measure column value will be the
> aggregated value; So user could not directly compare the "select *" result
> from a cube with the source data. If you're comparing the aggregated
> queries, I believe they are totally the same.
>
>
>
> 2016-11-26 4:39 GMT+08:00 Alberto Ramón <a.ramonportoles@gmail.com>:
>
>> I have a super-Fact Table with 5 rows
>> [image: Imágenes integradas 2]
>>
>>
>> A- Data in CSV == Hive (OK)
>> B- Select * from Fact, in Kylin some values are different
>>
>> The value 9942758, has been  transformed in 10937033.8 !!!
>>
>
>
>
> --
> Best regards,
>
> Shaofeng Shi 史少锋
>
>

Mime
View raw message