[HIVE-19190] Improve Logging for SemanticException Handling

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 2.3.2, 3.0.0
• Component/s: HiveServer2
• Assignee: Alice Fan
• Reporter: David Mollitor
• Created: 12/Apr/18 13:41
• Updated: 13/Sep/18 17:47
• Resolved: 13/Sep/18 17:47

Description

Please improve the logging for queries that fail with SemanticException. For example, when performing an action on a table that does not exist.  The most common reason why this happens is that someone fat fingers the table/column name.  It is not a system error pe se, but a user validation error. This is not something the cluster administrator should have to worry about. Yet, Hive performs some pretty extreme logging on the matter. I have attached to this JIRA the logging produced by a single submission of the following query:

select * from madeup;

 
For SemanticException exceptions, please print the Exception getMessage() to the server INFO logging so that the query's life-cycle can be traced, but do not blast ERRORs and stack traces to the log file.

Issue Links

• duplicates: HIVE-19302 Logging Too Verbose For TableNotFound