The health node might not have received the health info from all
nodes yet before the execution of this test, resulting in an
"unknown" status. We make the status assertion more lenient to
allow for this uncertainty. Additionally, we add some more
assertions for the basic response structure of the other indicators.
Prior to this change the shard might have been moved after restoring.
As a result we were getting an array with 2 recoveries (SNAPSHOT and PEER).
This change ensures the shard is not moved after restoring snapshot.
* Track source for arrays of objects
* Update docs/changelog/108417.yaml
* add missing field name
* add missing field name
* refactor and add test
* merge changes from main
* revert unintended
* add tests
* small refactor
* small refactor
* check for synthetic source
* test setting sourceStored
* order fields
* small refactor
* small refactoring and more tests
* small test fix
This commit removes the copy of the ScalarQuantizedVectorsWriter that we added in 8.14, in favour of a custom FlatVectorsScorer.
Previously we maintained the copy of the writer so that we could add a custom scorer implementation when merging scalar quantized vectors. This is not necessary now as we can provide a FlatVectorsScorer implementation that provides the custom scorer.
This change makes sure we check all queries, even the must_not ones, to decide if we should disable weight matches highlighting or not.
Closes#101667Closes#106693
Add support to the Unified highlighter to combine matches on multiple fields
to highlight a single field: "matched_fields".
Based on Lucene PR: https://github.com/apache/lucene/pull/13268
Lucene PR is based on the concept of masked fields where masked fields
are different from the original highlighted field. This PR in
Elasticsearch uses the already existing highlighter parameter
"matched_fields".
Renames the GET, PUT and DELETE inference APIs removing the model parts:
inference.delete_model -> inference.delete
inference.get_model -> inference.get
inference.put -> inference.put
The GET response now has a endpoints field instead of models
* Track synthetic source for disabled objects
* Update docs/changelog/108051.yaml
* minor refactor
* remove redundant method
* no need to skipChildren
* add mixed test
* add mixed yaml test
* remove extra line
