Skip to content

Conversation

@pmpailis
Copy link
Contributor

When using -Dtests.jvm.argline="-Dbuild.snapshot=false" the IVF_FORMAT is not supported for tests, so we're met with the following error:

org.elasticsearch.index.mapper.MapperParsingException: Failed to parse mapping: Unknown vector index options type [bbq_ivf] for field [vector]

In this PR we simply add a new isEnabled method for all VectorIndexType, defaulting to true and overridden just for bbq_ivf.

Closes #129713, #129809, and #129818

@pmpailis pmpailis added >test Issues or PRs that are addressing/adding tests :Search Relevance/Vectors Vector search labels Jun 23, 2025
@elasticsearchmachine
Copy link
Collaborator

Pinging @elastic/es-search-relevance (Team:Search Relevance)

@elasticsearchmachine elasticsearchmachine added Team:Search Relevance Meta label for the Search Relevance team in Elasticsearch v9.1.0 labels Jun 23, 2025
@tteofili
Copy link
Contributor

I wonder if this shouldn't be a cluster feature / capability. Or, IIRC, IVF is behind a feature flag hence we might just add a check for that in the failing tests?

@tteofili
Copy link
Contributor

just noticed that @carlosdelest has just done something like that in #129766

@pmpailis
Copy link
Contributor Author

I wonder if this shouldn't be a cluster feature / capability. Or, IIRC, IVF is behind a feature flag hence we might just add a check for that in the failing tests?

I haven't updated how we define the BBQ_IVF format, we still make use of the feature flag defined in DenseVectorFieldMapper#IVF_FORMAT.

The main issue here is with the -Dtests.jvm.argline="-Dbuild.snapshot=false" param as mentioned in the description. While for standard tests the feature flag for IVF is enabled, when using a release build, this is not the case, so even though we could potentially pick it randomly (given the seed), when resorting to VectorIndexType#fromString we would throw.

The proposed fix in this case is to account for the feature flag when choosing the valid values in the RescoreKnnVectorQueryIT, similarly to how we do in VectorIndexType#fromString.

@pmpailis pmpailis added the test-release Trigger CI checks against release build label Jun 23, 2025
@pmpailis pmpailis merged commit 7d4bbcc into elastic:main Jun 23, 2025
32 checks passed
mridula-s109 pushed a commit to mridula-s109/elasticsearch that referenced this pull request Jun 25, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

:Search Relevance/Vectors Vector search Team:Search Relevance Meta label for the Search Relevance team in Elasticsearch >test Issues or PRs that are addressing/adding tests test-release Trigger CI checks against release build v9.1.0

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[CI] RescoreKnnVectorQueryIT testKnnSearchRescore failing

4 participants