[SQOOP-2165] Can't use warehouse-dir with parquet

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.4.5
• Component/s: hive-integration
• Assignee: Theodore michael Malaska
• Reporter: Mark Grover
• Created: 03/Mar/15 23:00
• Updated: 08/Apr/15 00:21

Description

Gwen and I were working on some code for Data Warehousing that uses sqoop and we found something interesting.

At one place:
Sqoop1 claims warehouse-dir and target-dir are incompatible:
https://github.com/apache/sqoop/blob/trunk/src/java/org/apache/sqoop/tool/ImportTool.java#L1006

(We should mention this in the docs btw)

But, then if we only put the warehouse-dir (and don't specify the target dir), it complains that the target-dir needs to be specified. See here:
https://github.com/apache/sqoop/blob/trunk/src/java/org/apache/sqoop/tool/ImportTool.java#L1019

And, fyi, here is the query we ran:

sqoop job --create user_upserts_import --meta-connect jdbc:hsqldb:hsql://${SQOOP_METASTORE_HOST}:16000/sqoop \
-- import --connect jdbc:mysql://<MYSQL>:3306/oltp --username root \
-m 8 --incremental append --check-column last_modified --split-by last_modified --as-parquetfile \
--query 'SELECT user.id, user.age, user.gender,
occupation.occupation, zipcode, last_modified FROM user JOIN occupation
ON (user.occupation_id = occupation.id) WHERE $CONDITIONS' \--hive-import --hive-table user_upserts --warehouse-dir /etl/movielens/

If we specify just the target-dir, we get a warning about writing to target-dir and the data goes to default warehouse directory (/usr/hive...), which is pretty unexpected:

15/03/03 14:47:22 WARN util.AppendUtils: Cannot append files to target dir; no such directory: _sqoop/03144643000000600_30456_mgrover-haa2-4.vpc.cloudera.com_f8bf8ac4

obviously the directory in the warning is not the target dir we specified... this looks like something internal to the Kite/Parquet code.