Endpoints

Setting Up and Communicating with Endpoints on the Nosana Network

When creating a deployment on the Nosana network, you can expose services that will be accessible via endpoints. This means an instance will be accessible via an endpoint with which you can communicate.

This guide will walk you through setting up an Nginx server and interacting with its endpoint. Afterwards, we will set up a Llama instance and start communicating with it.

Proof of concept: Nginx

Nginx is a high-performance web server and reverse proxy server that is widely used for serving static content, load balancing, and handling HTTP and HTTPS traffic.
It'll be a good proof of concept to showcase how to use a Nosana endpoint.

Setting Up an Nginx Server

Step 1: Create a Deployment

First, create a deployment with a job definition that exposes port 80 for Nginx.

TypeScript SDK

const deployment = await client.deployments.create({
  name: 'Nginx Server',
  market: '97G9NnvBDQ2WpKu6fasoMsAKmfj63C9rhysJnkeWodAf', // NVIDIA 4090
  timeout: 60, // 60 minutes
  replicas: 1,
  strategy: 'SIMPLE',
  job_definition: {
    "version": "0.1",
    "type": "container",
    "meta": {
      "trigger": "api"
    },
    "ops": [
      {
        "type": "container/run",
        "id": "nginx",
        "args": {
          "cmd": [],
          "image": "nginx",
          "expose": 80
        }
      }
    ]
  }
});

cURL API

curl -X POST "https://dashboard.k8s.prd.nos.ci/api/deployments/create" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $NOSANA_API_KEY" \
  -d '{
    "name": "Nginx Server",
    "market": "97G9NnvBDQ2WpKu6fasoMsAKmfj63C9rhysJnkeWodAf",
    "timeout": 60,
    "replicas": 1,
    "strategy": "SIMPLE",
    "job_definition": {
      "version": "0.1",
      "type": "container",
      "meta": {
        "trigger": "api"
      },
      "ops": [
        {
          "type": "container/run",
          "id": "nginx",
          "args": {
            "cmd": [],
            "image": "nginx",
            "expose": 80
          }
        }
      ]
    }
  }'

Step 2: Start the Deployment

After creating the deployment, you need to start it for it to actually execute.

TypeScript SDK

// Get the deployment and start it
const deployment = await client.deployments.get(deploymentId);
await deployment.start();

cURL API

curl -X POST \
  -H "Authorization: Bearer $NOSANA_API_KEY" \
  https://dashboard.k8s.prd.nos.ci/api/deployments/<deployment_id>/start

Step 3: Retrieve the Deployment

Once the deployment is running, retrieve it to get the endpoints. The deployment object includes an endpoints array with the service URLs.

TypeScript SDK

// Retrieve the deployment to get endpoints
const deployment = await client.deployments.get(deploymentId);

cURL API

curl -s \
  -H "Authorization: Bearer $NOSANA_API_KEY" \
  https://dashboard.k8s.prd.nos.ci/api/deployments/<deployment_id> | jq .

Step 4: Access the Service Endpoint

The deployment object contains the service URLs in the endpoints field. You can now access your Nginx service at this URL.

TypeScript SDK

// Access the service endpoint from the deployment
const serviceUrl = deployment.endpoints[0]?.url;

cURL API

# First, retrieve the deployment to get the endpoint URL
DEPLOYMENT_RESPONSE=$(curl -s \
  -H "Authorization: Bearer $NOSANA_API_KEY" \
  https://dashboard.k8s.prd.nos.ci/api/deployments/<deployment_id>)

# Extract the service URL from the endpoints array
SERVICE_URL=$(echo $DEPLOYMENT_RESPONSE | jq -r '.endpoints[0].url')

Info

The service endpoint URL will be in the format:
https://<id>.node.k8s.prd.nos.ci

You can find this URL in the endpoints array of the deployment object. Each endpoint includes:

• opId: The operation ID from your job definition
• port: The exposed port number
• url: The full service URL

Navigate to the service URL to find your Nginx service.

Success! Your Nginx instance is running on the Nosana Network.