[IMPALA-2161] Avro schema's generated by avro-tools can fail silently

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Impala 2.1.1
• Fix Version/s: Impala 2.3.0, Impala 2.2.8
• Labels: usability
• Target Version: Impala 2.2.8
• Assignee: Martin Grund
• Reporter: Peter Ebert
• Created: 28/Jul/15 18:22
• Updated: 10/Nov/15 20:12
• Resolved: 02/Sep/15 04:21

Description

Using avro-tools to create an avro schema resulted in a schema similar to

{"type":"record","name":"name1","namespace":"namespace1","fields":[

{"name":"id","type":["null","int"],"doc":"\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000"}

,

{"name":"temp","type":["null","int"]}

,{"name":"temp1","type":["null",

{"type":"string","avro.java.string":"String"}

]},{"name":"temp2","type":["null",

{"type":"string","avro.java.string":"String"}

]},{"name":"temp3","type":["null",

{"type":"string","avro.java.string":"String"}

]},

{"name":"temp4","type":["null","int"]}

,

{"name":"temp5","type":["null","int"]}

,{"name":"tem6","type":["null",

{"type":"string","avro.java.string":"String"}

]},{"name":"temp7","type":["null",

{"type":"string","avro.java.string":"String"}

]},

{"name":"value","type":["null","double"]}

,

{"name":"insert_dt","type":["null","long"]}

,{"name":"insert_by","type":["null",

{"type":"string","avro.java.string":"String"}

]}]}

The part where \u0000 repeats causes a select * from table to fail instantly and silently with no error reported.

Performing a select * from table limit 1 immediately returns an error.

Ideally there would be no error at all, as it's technically valid, do not see the problem in hive.