SHREELEKHYA GAMPA created CARBONDATA-4326:
---------------------------------------------

             Summary: mv created in beeline not hitting in sql/shell and vice 
versa if both beeline and sql/shell are running in parellel
                 Key: CARBONDATA-4326
                 URL: https://issues.apache.org/jira/browse/CARBONDATA-4326
             Project: CarbonData
          Issue Type: Bug
            Reporter: SHREELEKHYA GAMPA


[Steps] :-

When MV is created in spark-shell/spark-sql on table created using Spark 
Dataframe, Explain query hits MV in spark-shell/spark-sql, but doesnt hit MV in 
spark-beeline, Same is the case when MV is created in spark-beeline on table 
created using Spark Dataframe, query hits MV in spark-beeline, but doesnt hit 
MV in spark-shell/spark-sql. This issue is faced when both sessions are running 
in parallel during MV Creation. On restarting the sessions of Spark-shell/ 
Spark-beeline, query hits the MV in both sessions.

Queries Table created using Spark Dataframe:

val geoSchema = StructType(Seq(StructField("timevalue", LongType, nullable = 
true), StructField("longitude", LongType, nullable = false), 
StructField("latitude", LongType, nullable = false))) val geoDf = 
sqlContext.read.option("delimiter", ",").option("header", 
"true").schema(geoSchema).csv("hdfs://hacluster/geodata/geodata.csv")

sql("drop table if exists source_index_df").show() geoDf.write 
.format("carbondata") .option("tableName", "source_index_df") 
.mode(SaveMode.Overwrite) .save()

Queries for MV created in spark-shell: sql("CREATE MATERIALIZED VIEW 
datamap_mv1 as select latitude,longitude from source_index_df group by 
latitude,longitude").show() sql("explain select latitude,longitude from 
source_index_df group by latitude,longitude").show(100,false)

Queries for MV created in spark-beeline/spark-sql: CREATE MATERIALIZED VIEW 
datamap_mv1 as select latitude,longitude from source_index_df group by 
latitude,longitude; explain select latitude,longitude from source_index_df 
group by latitude,longitude;

[Expected Result] :- mv created in beeline should hit the sql/shell and vice 
versa if both beeline and sql/shell are running in parellel

[Actual Issue]:- mv created in beeline not hitting in sql/shell and vice versa 
if both beeline and sql/shell are running in parellel

 




--
This message was sent by Atlassian Jira
(v8.20.1#820001)

Reply via email to