[HIVE-16107] JDBC: HttpClient should retry one more time on NoHttpResponseException

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.0.1, 2.1.1
• Fix Version/s: 2.3.0
• Component/s: HiveServer2, JDBC
• Hadoop Flags: Reviewed
• Assignee: Vaibhav Gumashta
• Reporter: Vaibhav Gumashta
• Created: 03/Mar/17 20:23
• Updated: 21/Jul/17 18:46
• Resolved: 22/Mar/17 18:08

Description

Hive's JDBC client in HTTP transport mode doesn't retry on NoHttpResponseException. We've seen the exception being thrown to the JDBC end user when used with Knox as the proxy, when Knox upgraded its jetty version, which has a smaller value for jetty connector idletimeout, and as a result closes the HTTP connection on server side. The next jdbc query on the client, throws a NoHttpResponseException. However, subsequent queries reconnect, but the JDBC driver should ideally handle this by retrying.