[CARBONDATA-2832] Block loading error for select query executed after merge index command executed on V1/V2 store table

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 1.4.1
• Fix Version/s: 1.4.1
• Component/s: data-query
• Environment: Spark 2.1
• Assignee: dhatchayani
• Reporter: Chetan Bhat
• Due: 10/Aug/18
• Created: 06/Aug/18 11:05
• Updated: 07/Aug/18 08:37
• Resolved: 07/Aug/18 08:37

Description

Steps :

Create and load data in V1/V2 carbon store:

create table brinjal (imei string,AMSize string,channelsId string,ActiveCountry string, Activecity string,gamePointId double,deviceInformationId double,productionDate Timestamp,deliveryDate timestamp,deliverycharge double) STORED BY 'org.apache.carbondata.format' TBLPROPERTIES('table_blocksize'='1');

LOAD DATA INPATH 'hdfs://hacluster/chetan/vardhandaterestruct.csv' INTO TABLE brinjal OPTIONS('DELIMITER'=',', 'QUOTECHAR'= '"','BAD_RECORDS_ACTION'='FORCE','FILEHEADER'= 'imei,deviceInformationId,AMSize,channelsId,ActiveCountry,Activecity,gamePointId,productionDate,deliveryDate,deliverycharge');

In 1.4.1

refresh table brinjal;

alter table brinjal compact 'segment_index';

select * from brinjal where AMSize='8RAM size';

 

Issue : Block loading error for select query executed after merge index command executed on V1/V2 store table.

0: jdbc:hive2://10.18.98.101:22550/default> select * from brinjal where AMSize='8RAM size';

Error: java.io.IOException: Problem in loading segment blocks. (state=,code=0)

Expected : select query executed after merge index command executed on V1/V2 store table should return correct result set without error**