[DOC] index:false default and index:true rejection on pluggable data format indices on some specific field types

[aggarwalmayank]

What do you want to do?

• [ X] Request a change to existing documentation
• [X ] Add new documentation
• Report a technical problem with the documentation
• Other

Tell us about your request. Provide a summary of the request.
OpenSearch core PR opensearch-project/OpenSearch#22712 introduces new mapping behavior for
indices that use a pluggable data format: a specific set of field types now default index
to false, and an explicit index: true on those fields is rejected. This behavior is
undocumented. (The pluggable data format settings themselves already exist — see "Context" —
and are not introduced by this PR.)

What needs to be documented (the change in this PR)

1. index defaults to false for these field types on a pluggable-data-format index

Affected field types:
byte, short, integer, long, float, double, half_float, unsigned_long,
scaled_float, date, date_nanos, ip, boolean.

• These fields remain queryable (range, term, terms) — search is served by the
  pluggable format's columnar storage / doc values rather than by the Lucene index.
• The index: false default is implicit — it is not written into the stored mapping,
  consistent with OpenSearch's existing behavior of not persisting unspecified defaults.
• keyword and text are NOT affected — their search is still served by the Lucene
  inverted index, so they keep index: true.

Why these specific types: on a pluggable-data-format index, the format backs only columnar
storage (and bloom filters) for the affected types and does not provide their search
capability (POINT_RANGE for numeric/date/date_nanos/ip/scaled_float; FULL_TEXT_SEARCH
for boolean). Since the Lucene secondary does not write that search structure, index
defaults to false and search falls back to the columnar/doc-values path. keyword/text
retain their FULL_TEXT_SEARCH capability, so they are unchanged.

Implementation reference (exact mappers changed in the core PR — for traceability):

Field type(s)	Mapper class	Module
byte, short, integer, long, float, double, half_float, unsigned_long	NumberFieldMapper	server
scaled_float	ScaledFloatFieldMapper	modules/mapper-extras
date, date_nanos	DateFieldMapper	server
ip	IpFieldMapper	server
boolean	BooleanFieldMapper	server

KeywordFieldMapper and TextFieldMapper were intentionally not changed.

2. index: true is rejected on these fields

Setting index: true explicitly on an affected field type (at index creation, via a mapping
update, or through an index template) on a pluggable-data-format index returns a 400
MapperParsingException, indicating the data format cannot provide the requested search
capability (for example, "cannot collectively cover: [POINT_RANGE]", or FULL_TEXT_SEARCH
for boolean).

Pages likely affected

• Field type pages (numeric types, date/date_nanos, ip, boolean, scaled_float) —
  add a note about the pluggable-data-format index default and the index: true rejection,
  and explicitly call out that keyword/text are unaffected.
• The pluggable data format documentation section (wherever the existing settings are/should
  be described) — add the mapping behavior.

Version: List the OpenSearch version to which this issue applies, e.g. 2.14, 2.12--2.14, or all.
3.9

What other resources are available? Provide links to related issues, POCs, steps for testing, etc.

• Core PR: Default index to false for doc-values-backed fields on pluggable data format indices OpenSearch#22712