[ZEPPELIN-2582] Clarify docs on interpreters, interpreter groups and interpreter binding modes

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.7.1
• Component/s: documentation
• Assignee: Hoon Park
• Reporter: Lucas Partridge
• Created: 24/May/17 10:29
• Updated: 07/Jul/17 08:24
• Resolved: 07/Jul/17 07:24

Description

I've read the overview docs on interpreters at http://zeppelin.apache.org/docs/0.7.1/manual/interpreters.html and http://zeppelin.apache.org/docs/0.7.1/development/writingzeppelininterpreter.html but I'm still confused! E.g., at http://zeppelin.apache.org/docs/0.7.1/manual/interpreters.html#interpreter-binding-mode it says "In 'isolated' mode, each notebook will create new Interpreter process." Does that mean a new JVM process, or a new process (with its own interpreter instance) in the same JVM as the Interpreter Group?

Please can we have a simple explanation in one place about the relationship between the following entities:

JVM
Spark Driver program
Interpreter group
Interpreter settings
Interpreter
Interpreter binding mode for notebooks

A diagram showing the relationship between these entities for the 3 different interpreter binding notes (shared, scoped, isolated) would be extremely helpful. This would help users understand the implications of the different modes - for example, whether they can share a SparkContext and hence data or variables between different notebooks. Or what would happen to another user's notebook if a given Interpreter Group is restarted. Thanks!

Issue Links

• links to: GitHub Pull Request #2437