- Added support for AWS Secrets Manager in the database configuration to fetch credentials for the follower database and Plane PI database.
- Updated connection methods to utilize secrets when available and AWS credentials are present.
- Introduced new utility functions for fetching secrets from AWS Secrets Manager.
- Enhanced the Celery configuration to support AWS Secrets Manager for message broker credentials.
- Updated example environment file to include new AWS Secrets Manager configuration options.
- Added support for AWS Secrets Manager to retrieve model API keys in LLMConfig.
- Updated .env.example to include new environment variables for AWS Secrets Manager.
- Modified S3 client configuration to use default credential chain when explicit credentials are not provided.
- Improved handling of AWS Bedrock embedding model credentials.
- Updated OpenSearch client to support AWS IAM authentication using SigV4 when AWS credentials are available.
- Modified Celery configuration to allow for dynamic construction of the broker URL using AWS Secrets Manager for credentials.
- Enhanced environment configuration examples to reflect new AWS-related options for OpenSearch and message broker.
- Improved handling of authentication methods in the vector database client, prioritizing AWS credentials when available.
- Updated documentation to clarify the use of AWS Secrets Manager for credential injection in the Celery setup.
- Updated database connection URL methods to use URL encoding for user and password fields, ensuring special characters are handled correctly.
- Modified Alembic configuration to escape '%' in the SQLAlchemy URL to preserve URL-encoded passwords.
- Improved the `test_ml_model` method in the VectorStore class to support both single and batch input formats based on model capabilities.
- Adjusted embedding model validation to respect batch processing capabilities when testing model inference.
- Updated the Alembic configuration to remove URL encoding for the SQLAlchemy connection URL, ensuring it uses the correct format.
- Enhanced error logging in the database migration command to provide clearer messages regarding migration failures, specifically addressing database connectivity and credential issues.
- Introduced asynchronous authentication methods for the OpenSearch client, allowing for improved performance in handling requests.
- Added a new function to build authentication kwargs specifically for the asynchronous client, utilizing AWS SigV4 async signer when AWS credentials are available.
- Updated the VectorStore class to utilize both synchronous and asynchronous authentication methods for OpenSearch connections.
- Added `su-exec` to the Dockerfile for user switching to enhance security.
- Created a new user group and user for the Plane application in the Dockerfile.
- Updated entrypoint scripts to switch to the `plane` user if executed as root, improving security during execution.
- Ensured proper ownership of the application directory for the new user.
- Modified entrypoint scripts for the FastAPI application, Celery beat scheduler, and Celery worker to use `exec` for starting processes. This change ensures that the Python processes replace the shell process, allowing for better signal handling and resource management.
- Added a new user group and user for the Plane application in the Dockerfiles for both admin and web apps to improve security.
- Updated ownership of relevant directories and files to ensure proper permissions for the new user.
- Configured Nginx to use a PID file for better process management.
