Skip to content

README.md

Latest commit

 

History

History
77 lines (51 loc) · 3.27 KB

README.md

File metadata and controls

77 lines (51 loc) · 3.27 KB

LangChain Cassandra Example

This example uses LangChain's Cassandra vector store connector against a ScyllaDB deployment. It creates a demo keyspace if needed, lets LangChain/CassIO create the vector table and indexes automatically, inserts a small document set through LangChain, and prints documents returned by similarity_search_with_score.

The example uses a local FastEmbed embedding model, so it does not require a paid embedding API subscription. The model is downloaded on first use.

The ScyllaDB- and LangChain-specific integration code lives in scylla_langchain.py; langchain_rag.py is the command-line entry point that only parses arguments, loads .env, and reports missing dependencies.

Requirements

This example relies on LangChain/CassIO automatic schema setup, which creates a Storage Attached Index (SAI) on the metadata map column. ScyllaDB supports this CassIO pattern starting with 2026.2 when the enable_cassio_compatibility flag is enabled; ScyllaDB then rewrites the CassIO metadata SAI statement into a regular secondary index. Enable the flag on every node, for example by passing --enable-cassio-compatibility 1 on the command line or setting enable_cassio_compatibility: true in scylla.yaml. The flag is also live-updatable without a restart via UPDATE system.config SET value = 'true' WHERE name = 'enable_cassio_compatibility';.

The ANN search itself is served by the ScyllaDB Vector Store service, which must be configured for the cluster.

Setup

Create and activate a Python virtual environment, then install the example dependencies:

python3 -m venv .venv
. .venv/bin/activate
pip install -r requirements.txt

Copy the environment template and fill in your ScyllaDB connection details:

cp .env.example .env

The script reads .env from this directory. The path is intentionally not configurable.

SCYLLA_CONTACT_POINTS=127.0.0.1
SCYLLA_USERNAME=cassandra
SCYLLA_PASSWORD=cassandra

This minimal example does not configure TLS or CA bundles. Use it only with localhost or a trusted private network; production deployments should configure TLS.

The script always uses:

  • CQL port 9042
  • keyspace langchain_demo
  • table langchain_rag_demo
  • embedding model BAAI/bge-small-en-v1.5
  • embedding dimension 384

Usage

Show the CLI help without connecting to ScyllaDB or loading LangChain dependencies:

python langchain_rag.py --help

Run the example:

python langchain_rag.py --query "How does ScyllaDB support vector search?"

Start from a clean demo table:

python langchain_rag.py --reset --query "How does ScyllaDB support vector search?" --k 3

--reset drops the demo table if it exists. Dropping the table also removes its vector index.

Notes

  • The script uses username/password authentication only and does not configure TLS or CA bundles.
  • LangChain/CassIO creates the table and indexes automatically (SetupMode.SYNC). This requires a ScyllaDB version with enable_cassio_compatibility enabled (see Requirements).
  • The Vector Store service builds the ANN index asynchronously, so the script waits and retries the first search until the index is ready.
  • Do not commit .env with real credentials.
  • The example prints retrieved documents with scores; it does not run text generation.