Babulal created CARBONDATA-406:
----------------------------------
Summary: Empty Folder is created when data load from dataframe
Key: CARBONDATA-406
URL:
https://issues.apache.org/jira/browse/CARBONDATA-406 Project: CarbonData
Issue Type: Bug
Components: data-load
Affects Versions: 0.1.0-incubating
Reporter: Babulal
Priority: Trivial
Load the data from dataframe to carbon table with tempCSV=false option .
Load is success but emtyFolder is getting created in HDFS
Cluster size= 3 node .
Type:- Stanalone Spark
Steps
val customSchema = StructType(Array( StructField("imei", StringType, true), StructField("deviceInformationId", IntegerType, true), StructField("mac", StringType, true), StructField("productdate", TimestampType , true), StructField("updatetime", TimestampType, true), StructField("gamePointId", DoubleType, true), StructField("contractNumber", DoubleType, true) ));
val df = cc.read.format("com.databricks.spark.csv").option("header", "false").schema(customSchema).load("/opt/data/xyz/100_default_date_11_header.csv");
Start data loading
scala> df.write.format("carbondata").option("tableName","mycarbon2").save();
Check Logs
leges:{}, groupPrivileges:null, rolePrivileges:null))
INFO 10-11 23:52:44,005 - Creating directory if it doesn't exist: hdfs://10.18.102.236:54310/opt/Carbon/Spark/spark/bin/null/bin/carbonshellstore/hivemetadata/mycarbon4
AUDIT 10-11 23:52:44,037 - [BLR1000007781][root][Thread-1]Table created with Database name [default] and Table name [mycarbon4]
INFO 10-11 23:52:44,040 - Successfully able to get the table metadata file lock
In the HDFS this Path is empty
hdfs://10.18.102.236:54310/opt/Carbon/Spark/spark/bin/null/bin/carbonshellstore/hivemetadata/mycarbon4
Actual Store location is :- hdfs://10.18.102.236:54310/opt/Carbon/mystore
Expect :- Empty folder should not be created. . It seems that it is created in SPARK_HOME/bin .
SPARK_HOME is /opt/Carbon/Spark/spark/bin
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)