Python Pipeline on a separate server¶

If your user base on your Seatable Server is growing, one of the first components that you could move to a separate server is the python pipeline. This frees up ressoures for the main server and it increases the security because customer python code is not executed on the main server.

Requirements¶

To install Python Pipeline on a separate server, use at least these ressources:

2 CPU
4 GB RAM
40 GB SSD

Use a separate domain or subdomain that is public available or you need a valid wildcard certificate. Caddy will generate a let's encrypt certificate.

No self signed certificates

Self signed certificates are not recommended. Read the article about custom certificates for more details.

Configure a firewall

Please configure a firewall to only allow access to port 80 and 443. These ports are required in order to receive a certificate from Let's Encrypt.

The value of the variable SEATABLE_SERVER_PUBLIC_IP is used by Caddy to restrict access to the python-scheduler on layer 7. Only the virtual machine running dtable-web should be allowed to access the python-scheduler component.

The Python pipeline should not be deployed on a virtual machine that is attached to a private network since the script containers can access this network interface if you do not configure any additional firewall rules.

Deployment of the Python Pipeline¶

The deployment of a separate python pipeline is simple. Get seatable-release from github like described in the installation of seatable server and only use caddy.yml and python-pipeline-standalone.yml.

Update your .env, that it looks like this and add/update the values according to your needs:

COMPOSE_FILE='caddy.yml,python-pipeline-standalone.yml'
COMPOSE_PATH_SEPARATOR=','

# system settings
TIME_ZONE='Europe/Berlin'

# seatable server url
SEATABLE_SERVER_HOSTNAME=seatable.your-url.com

# database
MARIADB_PASSWORD=your-password

# python-pipeline-standalone
PYTHON_SCHEDULER_HOSTNAME=python.your-url.com
PYTHON_SCHEDULER_AUTH_TOKEN=shared secret with dtable_web_settings.py
# Only this IP is allowed to access the python-scheduler
SEATABLE_SERVER_PUBLIC_IP='IP of seatable.your-url.com'

Execute docker compose up -d to fire up your separate python pipeline.

Configuration of SeaTable Server¶

SeaTable must know how to access the Python Pipeline.

Update the .env on the seatable server and add these two informations:

PYTHON_SCHEDULER_URL=https://python.your-url.com
PYTHON_SCHEDULER_AUTH_TOKEN=shared secret with python scheduler

Restart seatable service and test your python pipeline.