[HIVE-12607] Hive fails on zero length sequence files

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.2.1
• Component/s: Serializers/Deserializers
• Assignee: Chaoyu Tang
• Reporter: Chaoyu Tang
• Created: 07/Dec/15 20:47
• Updated: 07/Dec/15 20:56

Description

Flume will, at times, generate zero length sequence files which cause the Hive query failure.
To reproduce the issue:

> create external table test (id string)
partitioned by (system string,date string)
STORED AS SEQUENCEFILE
LOCATION '/user/me/test';

hadoop fs -mkdir /user/me/test/logs
hadoop fs -mkdir /user/me/test/logs/date=2014
hadoop fs -touchz /user/me/test/logs/date=2014/a.txt

hive
-> ALTER TABLE test ADD PARTITION (system = 'logs',date='2014') location '/user/me/test/logs/date=2014';
-> select * from test t1,test t2 where t1.id = t2.id;

Issue Links

• is related to: HIVE-7316 Hive fails on zero length files
• is related to: HIVE-11977 Hive should handle an external avro table with zero length files present