one more update

README.md

@@ -185,6 +185,66 @@ python setup.py develop
 11. To test if the if everything is working accordingly, open Python window within the virtual environment and type `import sciris` and `import scirisweb`. If no errors occur, then the import worked.
 
 
+## Multhreaded deployment
+
+The problem with the simple deployment method described above is that requests are single-threaded. If this is an issue, recommended deployment is using `nginx` to serve the static files, and `gunicorn` to run the Flask app. Note that it is common for an application to call several RPCs with each page load. This means that the multithreaded deployment can result in improved site performance even for a single user. 
+
+### Requirements
+
+You must have nginx (`sudo apt install nginx`) and gunicorn (`pip install gunicorn`) installed. 
+
+### Set up nginx
+
+1. Copy `examples/gunicorn/example_nginx_config` to e.g. `/etc/nginx/sites-enabled/my_app` (can change filename if desired)
+2. Edit the copied file to specify
+    - The hostname/URL for the site e.g. `my_app.com`
+    - The full path to the directory containing `index.html` on the system running `nginx`
+    - Change the port in `proxy_pass` line if desired - it must match the port in `launch_gunicorn`
+3. Reload or restart `nginx` e.g. `sudo service nginx reload`
+
+For example, this will start it running at `localhost:8188`:
+
+```script
+server {
+    listen 8188;
+    server_name localhost;
+    location / {
+        root /home/my_username/my_sciris_app;
+    }
+    location /api {
+        proxy_pass http://127.0.0.1:8097/;
+    }
+}
+```
+
+### Run gunicorn
+
+1. Copy `examples/gunicorn/example_launch_gunicorn` to the folder with your app (e.g. `launch_my_app_gunicorn`), and set the number of workers as desired - usual recommendation is twice the number of CPUs but for applications that are CPU bound (e.g., an RPC call runs a model) then it may be better to reduce it to just the number of CPUs.
+2. The example script references the Flask app using `name_of_your_app:flask_app`. The `name_of_your_app` should be importable in Python (either via running Python in the current directory, or installing as a package via `pip`) and `flask_app` is the name of a variable containing the Flask application. So for example, you might have a file `foo.py` containing
+
+```python
+app = sw.ScirisApp(__name__, name="My App")
+the_app = app.flask_app
+```
+in which case the `launch_my_app_gunicorn` script should contain `foo:the_app` instead of `name_of_your_app:flask_app`.
+
+3. Run `launch_my_app_gunicorn`. This will need to be kept running to support the site (so run via `nohup` or `screen` etc.).
+
+For example:
+```script
+cd my_app
+screen -S my_app_session
+./launch_my_app_gunicorn
+<you can now close the terminal>
+
+...
+
+<coming back later, you can restart it with>
+screen -R my_app_session
+```
+
+Note that for local development, you can add the `--reload` flag to the `gunicorn` command to automatically reload the site. This can be helpful if using the `nginx+gunicorn` setup for local development.
+
 
 ## Examples