Chainlit Data Persistency

The Chainlit data layer allows you to persist conversations in chainlit. This project uses a dockerized fork for easier deployment and setup.

In OpenRAG, one can activate Chainlit data layer following these steps:

Step 1: Set up authentication

In fact, chainlit authentication is necessary for data persistency. Set chainlit authentication if not already done (refer to the chainlit auth guide)

Step 2: Add the following variables

To deploy the Chainlit data layer service, add the following variable:

.env

# Persistency services: postgres (localstack (AWS emulator deployed locally)
CHAINLIT_DATALAYER_COMPOSE=extern/chainlit-datalayer/compose.yaml

This provides 2 services:

• a postgres database to store users, feedback, chat history, etc
• “s3 bucket” emulator to store elements (files attached in the chat).

Note

Chainlit datalayer is cloud-compatible, and the same applies for local data persistency. So for local storage, a cloud/s3 service emulator that runs in a container is deployed as well.

• Variables for the postgres data

Tip

Knowing that OpenRAG already has a running postgres service (rdb) (refer to the docker-compose.yaml file), there is no need to deploy another postgres service. In that case, comment out the postgres service definition in the compose.yaml file and add the following variable to your .env

.env

DATABASE_URL=postgresql://root:root_password@rdb:5432/chainlit

• Variables for chainlit to use the S3 Bucket Add the following variables to your .env so that chainlit can use them to connect to the locally deployed S3 bucket

.env

## S3 bucket configuration.
BUCKET_NAME=my-bucket
APP_AWS_ACCESS_KEY=random-key
APP_AWS_SECRET_KEY=random-key
APP_AWS_REGION=eu-central-1
DEV_AWS_ENDPOINT=http://localstack:4566

Tip

If you want to deactivate the service, comment out these variables, especially CHAINLIT_DATALAYER_COMPOSE.