GCP_DEPLOY.md

Python GCP deployment instructions

This assumes you have the following installed locally:

• docker
• helm (tested with v3)
• kubectl
• gcloud

---

1. Set up project:

1. Create GCP project
2. Enable Kubernetes Engine
3. Enable Cloud Filestore API

2. Deploy

1. Create a new key on the service account test-deploy in the GCP project. Your JSON key-file will be automatically saved to your computer. Move it to secrets/gcp, rename it to service-account.json and authenticate your gcloud CLI with:

   gcloud auth activate-service-account --key-file secrets/gcp/service-account.json

2. Create and enter conda environment:

   conda env create -f dev_environment.yml
   conda activate codehub

3. Install CLI:

   pip install --editable .

4. You can now use the CLI by typing:
   1. Create cluster

      codehub createcluster -n <cluster-name>

   2. Upgrade cluster

      codehub upgradecluster -n <cluster-name>

   3. Scale cluster

      codehub scalecluster -n <cluster-name>

   4. Delete cluster

      codehub deletecluster -n <cluster-name>

   5. Delete nfs

      codehub deletefs -n <cluster-name>

   6. Get cluster IP

      codehub getip -n <cluster-name>