Skip to content

Mapping Overhaul - More user friendly, cluster aware #34

@kimchy

Description

@kimchy
  1. When adding mapping definitions, they are now merged with the current mapping definitions. Duplicates are silently ignore unless specified in the put mapping API (HTTP param ignoreDuplicates set to false). Note, duplicates refer only to field mapping, object mapping are recursively checked.
  2. Mappings definitions are now clustered. Up until now, when creating mapping, the mappings were not merged (see point 1) but they were broadcasted to the cluster. But, when a document indexed resulted in updated mapping, this fact was lost. Now, this changed mapping are updated on the master and merged, and broadcasted to the whole cluster, which means all the different nodes will know about the new types introduced almost immediately.

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions