Evora

Andor wrapper for Evora and Flask server.

Installation

evora-server requires the proprietary Andor libraries to be installed in /usr/local/lib. The library can be used for debugging without the Andor libraries, but they are necessary to run the actual camera.

To install evora-server, clone the repository and run

pip install .

or to install in editable mode

pip install -e .

Debug mode

To run the server in debug mode, with the dummy module mocking the camera, edit evora/debug.py and set DEBUGGING = True. This will create a folder in the evora-server root that acts as the /data folder.

Running the server

To run the server, from a terminal at the root of the project, execute

python app.py

which is equivalent to

flask --debug run --port 3000

Images

evora-server will save camera files to /data/ecam/DATE where DATE is in the format 20230504 and rotates at midnight UTC.

If /data/ecam does not exist, create it with mkdir -p and make sure it has the right permissions for evora-server to write to it.

Note: Mac OSx doesn't allow the creation of folders in the root / directory, since OSx makes the root directory read-only by default.

Deploying for production

The recommended way to run evora-server in production is by running the app with the Flask development server with a single process and threading. This allows for concurrent routes and asyncio to work (which is required for features such as aborting exposures). At this point this is preferred to using a UWSGI layer such as gunicorn since the camera has a single connection so we cannot run multiple workers.

To run this command in the background as a systemd service, create a file /etc/systemd/system/evora-server.service with the contents

[Unit]
Description=evora-server

[Service]
WorkingDirectory=/home/mrouser/Github/evora-server
ExecStart=/home/mrouser/Github/evora-server/standalone-start.sh

[Install]
WantedBy=multi-user.target

Here we are pointing to the file standalone-start.sh in the repo, which loads the conda environment and starts the server in production mode (port 8000, threading, one worker). This may need to be changed for a location other than MRO. Then start the systemd service with

sudo systemctl daemon-reload
sudo systemctl enable --now enable evora-server
sudo systemctl restart evora-server

Configuring nginx

In addition to running the server, a reverse proxy is needed to run the Evora client and server in the same HTTP server. In Ubuntu, install nginx (alternatively you can use Apache) with

sudo apt update
sudo apt install nginx

and adjust the firewall to open the desired ports. Then start nginx with

sudo systemctl enable --now nginx

We'll now add a new site for Evora. Create a new file /etc/nginx/sites-enabled/evora.conf with

sudo vim /etc/nginx/sites-enabled/evora.conf

and include the configuration

server {
    listen 80;
    listen [::]:80;
    server_name localhost;
    access_log  /usr/local/var/log/nginx/evora.log;

    location /api/ {
        proxy_pass http://127.0.0.1:8000/;
        proxy_set_header Host $host;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_connect_timeout 75s;
        proxy_read_timeout 1800s;
    }

    location /data {
       alias /data;
        autoindex on;
        index index.html index.php;
    }
}

This configuration creates a server running on port 80 (the default HTTP) and adds a reverse proxy to where the evora-server webapp is running. It also creates a route to expose and browse /data.

After this, restart nginx with

sudo systemctl restart nginx

and test that it works by navigating to http://localhost/api/getTemperature.

References

Andor SDK documentation

UWMRO/evora-server