[SOLR-7969] Unavailable clustering engines should not fail the core

• Type: Improvement
• Status: Closed
• Priority: Trivial
• Resolution: Fixed
• Fix Version/s: 5.4, 6.0
• Assignee: Dawid Weiss
• Reporter: Dawid Weiss
• Created: 25/Aug/15 09:43
• Updated: 09/May/16 18:47
• Resolved: 25/Aug/15 12:31

Description

Currently the clustering contrib will fail at initialization time if an algorithm (or some of its dependencies) is not available. This shouldn't be so eager, instead:

• the declaration of an engine should allow optionality,
• log the message about unavailable engine/algorithm (and its ID); if it is optional, the message can have a debug level,
• fail at query-time if the engine is requested explicitly.