Clarify doc count stats (#56665)

Today we report some statistics in terms of Lucene-level documents, which
differ from Elasticsearch-level documents in a number of ways and include
things like document tombstones which users cannot directly observe. This
commit clarifies the internal nature of these statistics.

Closes #56497
This commit is contained in:
David Turner 2020-05-13 15:07:15 +01:00
parent 9f1ecd52eb
commit 26382dff19

View File

@ -181,15 +181,20 @@ given in the query string.
end::df[]
tag::docs-count[]
Number of non-deleted documents in the segment, such as `25`. This
number is based on Lucene documents and may include documents from
<<nested,nested>> fields.
The number of documents as reported by Lucene. This excludes deleted documents
and counts any <<nested,nested documents>> separately from their parents. It
also excludes documents which were indexed recently and do not yet belong to a
segment.
end::docs-count[]
tag::docs-deleted[]
Number of deleted documents in the segment, such as `0`. This number
is based on Lucene documents. {es} reclaims the disk space of deleted Lucene
documents when a segment is merged.
The number of deleted documents as reported by Lucene, which may be higher or
lower than the number of delete operations you have performed. This number
excludes deletes that were performed recently and do not yet belong to a
segment. Deleted documents are cleaned up by the
<<index-modules-merge,automatic merge process>> if it makes sense to do so.
Also, {es} creates extra deleted documents to internally track the recent
history of operations on a shard.
end::docs-deleted[]
tag::docs-indexed[]