LOCATION in Hive

Question

While creating hive tables, Can I point the 'LOCATION' to a place in hdfs where data is present. Do I still need to load data or Will the data be available on hive directly?

Answer 1

You can specify any location while creating table and the data will be accessible. If table is partitioned, then use ALTER TABLE ADD PARTITION or MSCK REPAIR TABLE table_name or Amazon version ALTER TABLE table_name RECOVER PARTITIONS , this will add any partitions that exist on HDFS but not in metastore to the metastore. See docs here: https://cwiki.apache.org/confluence/display/Hive/LanguageManual+DDL#LanguageManualDDL-RecoverPartitions(MSCKREPAIRTABLE)

If table is not partitioned, you can simply specify the location with data while creating table or change table location using ALTER TABLE SET LOCATION.