[BIGTOP-2268] Untangle the convoluted Hive-support from Spark deployment

• Type: Bug
• Status: Open
• Priority: Blocker
• Resolution: Unresolved
• Affects Version/s: 1.1.0
• Fix Version/s: backlog
• Component/s: debian, deployment, rpm, spark
• Assignee: Unassigned
• Reporter: Konstantin I Boudnik
• Created: 22/Jan/16 20:21
• Updated: 10/May/17 15:29

Description

As explained in BIGTOP-2154, forcing Hive support as a default option of Spark packaging breaks a number of things (like spark-shell) unless very unrelated libraries from spark-datanucleus are installed. The JIRA provides a very hackish way of working this issue around just for the sake of getting 1.1.0 out.

In the next release it needs to be fixed in a more structural and architecturally sound way.

Issue Links

• depends upon: BIGTOP-2154 spark-shell doesn't start anymore without Hive libs in the classpath
• is related to: BIGTOP-2764 deployment failure when roles include spark::common and spark::yarn*