To create a worker, we made our own container image very similar to the Flask container. The Dockerfile uses Alpine Linux and explicitly loads Python 3 onto that image, then installs the requirements from a requirements.txt file and adds in two Python files. The first, celery_conf.py, is the Python definition for a couple of tasks taken directly from the Celery documentation. The second, submit_tasks.py, is a short example that is meant to be run interactively to create work and send it over the queue. The container also includes two shell scripts: run.sh and celery_status.sh.
In all of these cases, we used environment variables that we source from the preceding ConfigMap to set up the logging output from the worker, as well ...
