Deployment Guide

The GSM service is deployed as an SDP component (including its required database backend), which allows full integration testing of the code. Follow the SDP documentation for guidelines.

Running the application locally

The application can be built and run locally, with the backend services deployed using docker or podman.

Starting the database services

Running an etcd instance:

docker run -it  \
  -p 2379:2379 \
  -e ETCD_NAME=ska-sdp-etcd-0 \
  -e ETCD_LISTEN_CLIENT_URLS=http://0.0.0.0:2379 \
  -e ETCD_ADVERTISE_CLIENT_URLS=http://0.0.0.0:2379 \
  -e ETCD_MAX_TXN_OPS=1024 \
  artefact.skao.int/ska-sdp-etcd:3.5.21 \
  etcd

The repository contains two Dockerfiles, one for creating a local PostgreSQL database and another for the GSM Service.

Building the PostgreSQL DB (note the dot at the end of the command):

docker build -t gsm-db -f images/ska-sdp-global-sky-model-db/Dockerfile .

Running the PostgreSQL DB (for a temporary DB, remove the --mount... line):

podman run -d \
  --mount source=gsm_db_vol,target=/var/lib/postgresql/data \
  -e POSTGRES_PASSWORD=pass \
  -e POSTGRES_DB=postgres \
  -p 5432:5432 \
  gsm-db

Starting the GSM Service using docker

Using the GSM Dockerfile, build an image:

docker build -t gsm -f images/ska-sdp-global-sky-model-api/Dockerfile .

Start the service:

podman run -d \
  -e API_VERBOSE=true \
  -e POSTGRES_HOST=localhost \
  -e POSTGRES_PORT=5432 \
  -e POSTGRES_DB=postgres \
  -e POSTGRES_USER=postgres \
  -e POSTGRES_PASSWORD=pass \
  -p 8000:80
  gsm

This will start the GSM service in the background. The API will be available at http://localhost:8000.

Running the GSM service as a script

The API can also be run as a standalone script for development. This requires:

A running PostgreSQL database
A running etcd instance (for flow management features)

You can achieve this using docker, or by installing local versions of the services.

Then run the API directly:

$ POSTGRES_HOST=localhost poetry run uvicorn \
  ska_sdp_global_sky_model.api.app.main:app --reload

The API will be available at http://localhost:8000. To access the API documentation, navigate to http://localhost:8000/docs