[DRILL-6171] Querying avro files returns null for not null values when JDBC/ODBC drivers are used

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.10.0
• Component/s: Client - JDBC, Storage - Avro
• Assignee: Unassigned
• Reporter: Engin Sozer
• Created: 20/Feb/18 14:29
• Updated: 20/Feb/18 14:36

Description

Querying an avro file results in incorrect results. (with jdbc/odbc drivers from MapR) For example:

We have a file (test.avro') with columns col1, col2, col3 and col4 for a total of 5000 rows. In the first 4000 rows, col3 is null. In the last 1000 rows, col3 is not null. The issue is that when we write;

select * from dfs.tmp.`test.avro`;
select col1, col2, col3 from dfs.tmp.`test.avro`;

col3 is returned null for all 5000 rows. If we write:

select col1, col2, COALESCE(col3, null) from dfs.tmp.`test.avro`;

This returns the correct results. (col3 = null for first 4000 rows and col3= not null for the next 1000)

This issue did not occur when the same query is run through the Drill Web UI.