Check if worker is initialized

[nkaretnikov]

Fixes #605.

Description

conda-store can run via docker and in standalone mode:

• in standalone mode, the --standalone CLI parameter must be used to start the worker sub-process
• when running via docker, this CLI parameter is not used because a dedicated worker process is started by docker-compose.

However, there might be cases where users forget to pass the --standalone parameter when running in standalone mode. This is a problem because the server cannot execute any tasks without the worker, so it must be running.

This PR adds a check that verifies whether the worker is running and prints an error message if not.

The worker is configured via a number of parameters, like celery_broker_url, which can either point to a database connection or use Redis:

    @default("celery_broker_url")
    def _default_celery_broker_url(self):
        if self.redis_url is not None:
            return self.redis_url
        return f"sqla+{self.database_url}"

Similarly, checks like conda_store.celery_app.control.inspect().ping() won't work outside of Redis and RabbitMQ.

So we use a database entry to check whether the worker is initialized. The database is used because it will work regardless of parameters used to configure the worker.

Example output:

[CondaStoreServer] WARNING | Waiting for worker... Use --standalone if running outside of docker
[CondaStoreServer] WARNING | Waiting for worker... Use --standalone if running outside of docker
[CondaStoreServer] WARNING | Waiting for worker... Use --standalone if running outside of docker
[CondaStoreServer] Worker initialized

The warning message is printed in red, the info message is printed in green.

The simplest way to test this:

• run the server in one terminal, wait for warnings to be printed
• run the worker in another terminal, wait for the info message to be printed.

The commands to run from conda-store/conda-store-server:

terminal1: python -m conda_store_server.server
terminal2: python -m conda_store_server.worker

To test in standalone mode (also starts the worker):

python -m conda_store_server.server --standalone

To test via docker:

docker-compose build --build-arg RELEASE_VERSION=2023.10.1 && docker-compose up

Pull request checklist

• Did you test this change locally?
• Did you update the documentation (if required)?
• Did you add/update relevant tests for this change (if required)?

Additional information

[netlify]

✅ Deploy Preview for kaleidoscopic-dango-0cf31d canceled.

Name	Link
🔨 Latest commit	1863ec0
🔍 Latest deploy log	https://app.netlify.com/sites/kaleidoscopic-dango-0cf31d/deploys/65abf0a47d1b2500087bb149

[nkaretnikov]

Tested this on Linux and Windows. Example output from Windows (server and worker are running in separate terminals):

[nkaretnikov]

@dcmcand Please take a look!

Summary of changes:

• The constants and method are accessible globally, but they are marked as private to avoid polluting the public namespace.
• The time delay value is now a parameter.
• Removed the DB constraint, too.
• Fixed a typo.

[dcmcand]

🚀 Thanks @nkaretnikov