MCP Server Deployment

The MCP server can be deployed either locally, as a Python package, or as a remote HTTP server. This section of the guide explains the details of how to run the MCP server as an HTTP server.

Direct Python Deployment

Install the Exasol MCP Server Python package. Below are examples of pip and uv commands.

pip install exasol-mcp-server

uv tool install exasol-mcp-server@latest

Set the required environment variables and, optionally, the settings json file as described in the MCP Server Setup and other guides referenced there.

Start the HTTP server from the command line, as in the example below.

exasol-mcp-server-http --port <server-port>

Docker Deployment

When running the MCP Server from a Docker container, the environment variables and the settings file shall be passed as the arguments to the docker run command, as in the example below.

Create a file settings.json and save it somewhere in the host machine.

{
    "enable_read_query": true
}

Now create and run the Docker container.

docker run \
    -p 7766:4896 \
    -e EXA_DSN=my_dsn \
    -e EXA_USER=my_user-name \
    -e EXA_PASSWORD=my-password \
    -e FASTMCP_SERVER_AUTH=exa.fastmcp.server.auth.oauth_proxy.OAuthProxy \
    -e EXA_AUTH_UPSTREAM_AUTHORIZATION_ENDPOINT=my_identity_provider/oauth2/authorize \
    -e EXA_AUTH_UPSTREAM_TOKEN_ENDPOINT=my_identity_provider/oauth2/token \
    -e EXA_AUTH_UPSTREAM_CLIENT_ID=my_client_id \
    -e EXA_AUTH_UPSTREAM_CLIENT_SECRET=my_client_secret \
    -e EXA_AUTH_JWKS_URI=my_identity_provider/jwks \
    -e EXA_AUTH_BASE_URL=my_mcp_server \
    -e EXA_MCP_SETTINGS=/app/settings.json \
    -v local_path_to_settings.json:/app/settings.json \
    exadockerci4/exasol-mcp-server:latest
    --port 4896

In this example, the server is configured to use the generic OAuth Proxy provider. The server starts at the Docker container’s port 4896, which is published to the host port 7766. Normally, the base URL of the MCP server will be a secure address of the server, as described in the next section. For example, if the server is deployed in an AWS EC2 instance, the EXA_AUTH_BASE_URL can be set to “https://ec2-xx-xxx-xx-xx.my-region.compute.amazonaws.com”.

Note that the path to the settings.json file must be an absolute path.

Reverse proxy configuration

Exasol MCP Server runs as an HTTP server. Connecting to the server via an unsecured HTTP channel should only be considered as a way of testing it. A production deployment would normally include a reverse proxy, allowing a secure connection to the server from anywhere in the internet. Below is an example of a configuration for Nginx that is commonly used as a reverse proxy.

Check if the Nginx is installed. The following command should print its version.

nginx -v

Create the Nginx configuration file for Exasol MCP Server. This configuration relates to the previous example of running the server in a Docker container. In particular, it assumes that the server is accessible at the local port 7766.

server {
    listen 80;
    # Public name or address of the server
    server_name ec2-xx-xxx-xx-xx.my-region.compute.amazonaws.com;

    # Redirect all HTTP to HTTPS
    return 301 https://$host$request_uri;
}

server {
    listen 443 ssl;
    # Public name or address of the server
    server_name ec2-xx-xxx-xx-xx.my-region.compute.amazonaws.com;

    # SSL certificate
    ssl_certificate /etc/nginx/ssl/my_certificate.pem;
    ssl_certificate_key /etc/nginx/ssl/my_certificate_key.pem;

    # SSL settings
    ssl_protocols TLSv1.2 TLSv1.3;
    ssl_ciphers HIGH:!aNULL:!MD5;

    location / {
        # Local address where the server is accessible from the host
        proxy_pass http://localhost:7766;

        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
    }
}

Point to the created file in the Nginx own configuration file - nginx.conf. The location of this file in a standard installation of Nginx depends on the OS:

* Linux: /etc/nginx/nginx.conf
* macOS: /usr/local/etc/nginx/nginx.conf
* Windows: C:\nginx\conf\nginx.conf

In the http section, add the following line.

http {
    ...
    include <path_to_the_server_config_file>;
    ...
}

Validate the new configuration by running the following command. Administrator privileges might by required.

nginx -t

Let Nginx use the new configuration.

nginx -s reload