Skip to content

Add optional Milvus vector store backend - #267

Open
zc277584121 wants to merge 1 commit into
ANative-Lab:mainfrom
zc277584121:feature/milvus-vector-store
Open

Add optional Milvus vector store backend#267
zc277584121 wants to merge 1 commit into
ANative-Lab:mainfrom
zc277584121:feature/milvus-vector-store

Conversation

@zc277584121

Copy link
Copy Markdown

Summary

Adds an optional Milvus vector store backend for EvoAgentX RAG storage.

  • Adds a LlamaIndex-compatible Milvus vector store backed by pymilvus.MilvusClient
  • Keeps FAISS as the default vector store
  • Adds vector_name="milvus" configuration fields for URI, token, database, collection, metric type, consistency level, and overwrite behavior
  • Adds evoagentx[milvus] optional dependency with Milvus Lite support
  • Documents Milvus configuration in the storage and RAG docs

Closes #266.

Validation

  • ruff check evoagentx
  • ruff check evoagentx/storages/vectore_stores tests/src/storages/test_milvus_vector_store.py
  • pytest tests/src/storages/test_milvus_vector_store.py tests/src/storages/test_storagehandler.py -q
  • Manual Milvus Lite + OpenAI text-embedding-3-small RAG smoke test
  • Manual Zilliz Cloud direct vector store smoke test with a temporary collection

Notes

Milvus uses a local ./milvus.db URI by default for Milvus Lite. The same configuration can point to Milvus server or Zilliz Cloud by setting milvus_uri and milvus_token.

Signed-off-by: Cheney Zhang <chen.zhang@zilliz.com>
@zc277584121

Copy link
Copy Markdown
Author

Hi @TedSIWEILIU, I opened this PR following the route discussed in #266. It keeps FAISS as the default, adds Milvus as an optional backend with lazy imports, and follows the existing VectorStoreBase interface. Would appreciate your review when you have time.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Proposal: optional Milvus vector store backend for RAG storage

1 participant