@Marcos Da Silva could you please provide create table statement and select query that invokes full table scan? also please run explain and let's see output here. that will definitely help to understand the issue. ... View more

+1 for JIRA reference! Thanks! 🙂 ... View more

@subacini balakrishnan, I'm glad it worked for you. Could you please accept the correct answer, so the question would be marked as answered? thanks! ... View more

@subacini balakrishnan, Have you tried that? ... View more

So, DDLs and DMLs are running successfully, but SELECT fails. Check if you have write permissions on /tmp directory for user you are running query with (looks like it is "hive" user). The difference between running CLI and HS2 APIs - your code is running on different nodes (edge vs master). so make sure you have the same setup for user "hive" on both of them. ... View more

@subacini balakrishnan, There is some mess in logs... In a trace log above, in "HIVE FAILURE OUTPUT" section, you have: FAILED: Execution Error, return code 1 from org.apache.hadoop.hive.ql.exec.DDLTask. AlreadyExistsException(message:Table spark_3_test already exists).... ... dbName:default, tableName:spark_3_test ...Partition not found (server_date = 2016-10-23) Not sure how it is related to the query you are running for "spark_2_test" table and for different partition. I have reproduced all the steps in both Zeppelin and spark-shell. There shouldn't be any issue with running DROP PARTITION from spark shell. Try to give it clean shot - new table, new partitions, no locks of data/directories, no two tables with the same location, etc. Just clean shot. IMO it is related to the specific table configuration/definition. ... View more

@Arkaprova Saha If you are running Java code via JDBC connection to Hive, there is no "LOCAL" directory (except of local mode, which I believe is not the case). Well, actually, there is local dir, but it is on a machine that accepts your query (HS2). Most probably you just don't have writing permissions for the user you are connecting with. I would suggest: 1. run the same query just without "LOCAL". it will create files on HDFS. 2. pull files from HDFS (hdfs get, or hdfs getmerge, or Java API File system). Let me know if that works for you. ... View more