stacklok / toolhive / 21297033024

60%

Build Type

push

github

Committed by web-flow

Commit Message

Add EmbeddingServer CRD for deploying HuggingFace embedding models (#3323)

* Add MCPEmbedding CRD for embedding model deployment in operator

Introduces a new MCPEmbedding custom resource to deploy HuggingFace
embedding models as MCP servers in Kubernetes. This enables semantic
search and similarity features for MCP tools and resources.

Key Features:
- Custom resource definition for embedding model deployments
- Integration with HuggingFace text-embeddings-inference
- Support for model caching via PersistentVolumeClaims
- Flexible resource configuration and pod customization
- GroupRef support for organizational grouping
- Comprehensive status conditions and phase tracking

Components:
- MCPEmbedding CRD with validation and webhook support
- Controller for managing deployment lifecycle
- Generated CRD manifests and Helm chart templates
- RBAC permissions for managing embeddings
- Example configurations for various use cases

This change is based on the original commit by rebasing onto
jerm/2026-01-13-optimizer-in-vmcp to remove intermediate commits.

* Rename MCPEmbedding crd as EmbeddingServer

* Updated image and model names

* Remove unnecessary GroupRef from EmbeddingServers crd

* Fixed reconciliation loop issue causing no service creation

* Rename examples/operator/embeddings to examples/opeartor/embedding-servers

* Updated embedding server example yamls

* Bump toolhive operator version and fix linting issues

* Added e2e tests and fixed a bug

* Convert EmbeddingServer to use StatefulSets and add HuggingFace token support

This refactors the EmbeddingServer controller with the following changes:

- Convert from Deployment to StatefulSet for better persistent storage support
- Add HFTokenSecretRef field for secure HuggingFace token injection from Kubernetes secrets
- Use StatefulSet volumeClaimTemplates for model cache PVCs instead of separate PVC creation
- Remove Env field from EmbeddingDeploymentOverrides API
- Add comprehensive controll... (continued)

Run Details

455 of 947 new or added lines in 4 files covered. (48.05%)

5 existing lines in 3 files now uncovered.

37818 of 63116 relevant lines covered (59.92%)

76.84 hits per line