Running Locally

To run Cellarium Cloud services locally, you will need to have the following installed:

• Python 3.10

• Local Postgres Database cluster

• Project Environment variables in a src/settings/.env file. More info.

• Project dependencies installed. pip install -r requirements.txt

• src directory added to your PYTHONPATH environment variable. E.g. export PYTHONPATH=$PYTHONPATH:/path/to/cellarium-cloud/src

Once you have the above installed, you can run the services locally using the following command:

python src/casp/services/<service_name>/main.py

To check the API methods that exist and their documentation, you can visit API docs page

Most of the methods will reuqire you to be authenticated. To do this, you’d need to deploy Admin service:

python src/casp/services/admin/server.py

Once the Admin service is running, you can visit Admin Dashboard to create a user and token.

To run commands that require accessing the vector search API using gRPC, you will need to create a tunnel. To do this you must: - log into the gcloud cli with the command gcloud auth login

You can then run the following command to create a tunnel:

gcloud compute ssh --zone "us-central1-a" "bastion" --project "dsp-cell-annotation-service" -- -NL 10000:localhost:10000

In local development mode, the user agent header sent to the vector search API contains information on what vector API endpoint to route to.

The bastion server is a proxy server that routes requests to the vector search API and is configured using nginx with a reverse proxy that is configured in file /etc/nginx/sites-available/vertex:

map $http_user_agent $forward_ip {
"~target_ip\:([^\s]+)" $1;
default "no-ip-to-forward-to";
}

server {
    listen 10000 http2;

    location / {
        grpc_pass grpc://$forward_ip:10000;
        include proxy_params;
    }
}