Tested using an EC2 m6 instance with 64GB disk. Also known to work on a Raspberry Pi 5 with 8GB of RAM.
From the root of edge-endpoint, run:
balena login
balena push <your-fleet>This will build and push two "services" to the edge devices in your chosen fleet. The first is a k3s server named server, which effectively acts as our k3s cluster node. The second is the bastion service, from which a user can access the k3s cluster (e.g. by running kubectl get nodes). The bastion service also contains a copy of this repo at /app/edge-endpoint.
Now, we have our k3s single-node cluster built and running, but we have not started our edge deployment.
Configure the following variables via the <fleet>/Variables or <device>/Device Variables interfaces on the BalenaCloud dashboard:
GROUNDLIGHT_API_TOKEN - so that we can authorize the fetching of edge model binaries
AWS_ACCESS_KEY_ID - so we can pull the edge-endpoint and gl-edge-inference images from ECR
AWS_SECRET_ACCESS_KEY - needed along with AWS_ACCESS_KEY_ID
RUN_EDGE_ENDPOINT - Set this to "RUN_EDGE_ENDPOINT" to start the pods (added for glhub integration)
Optionally you can also configure EDGE_INFERENCE_FLAVOR to use GPU instead. It will default to CPU if not set.
Dockerfile will automatically run the following command as bastion launches so no need to run this anymore but leaving this command as a reference if we need to start the clusters manually.
cd /app/edge-endpoint
INFERENCE_FLAVOR="CPU" DEPLOYMENT_NAMESPACE="default" ./deploy/bin/setup-ee.sh