Skip to main content

Your first application with CGC SDK

Think of any application within CGC as compute or database resource. It is container that runs within POD, managed by DEPLOYMENT or STATEFUL-SET.

Below example showcase quick deployment of Nginx container within CGC.

Custom Nginx application​

# Import the SDK resource
import cgc.sdk.resource as resource

# Create a web server - it's like installing an app!
result = resource.resource_create(
    name="my-first-app",        # What you want to call it
    image_name="nginx:latest",   # The application to run (nginx is a web server)
    cpu=1,                       # How many CPU cores (like computer processors)
    memory=2                     # How much RAM in gigabytes
)

# Check if it worked
if result['code'] == 200:
    print("Success! Your web server is starting up.")
else:
    print(f"Something went wrong: {result.get('reason')}")
info

In a custom application you must create a PORT, to expose ports for the network communication within the namespace or web.

Check: PORTS

Using pre-existing apps​

Adding entity with a valid string, allows running pre-configured images within CGC.

Nvidia Pytorch example​

response = resource.resource_create(
    # Required parameters
    name="advanced-app",                   # Unique identifier
    
    # Resource specifications
    entity="nvidia-pytorch",               # Entity type (default: "custom")
    cpu=4,                                 # CPU cores
    memory=16,                             # Memory in GB
    shm_size=2,                            # Shared memory in GB
    gpu=1,                                 # Number of GPUs
    gpu_type="NVIDIA-RTX-A5000",           # GPU model (canonical name)
    environment_data=["ENV=production", "DEBUG=false"],
)

Get entity list​

# Get all available entity types
entities = resource.get_entity_list()
print("Available entities:", entities)

Available GPU types​

GPU types are served by the CGC server, so gpu_type expects the canonical name (NVIDIA-RTX-A5000, NVIDIA-A100-SXM4-80GB, ...). List the ones available in your namespace with the CLI:

cgc status

Resource allocation​

Based on your needs, here are some common specification for the application that you could run:

  • Simple website: 1 CPU, 2GB memory
  • Database: 2-4 CPU, 8-16GB memory
  • Data processing: 4-8 CPU, 16-32GB memory

Listing resources​

Resources are split into three types: compute, db, jobs. This part describes both compute and db.

List all compute resources​

import cgc.sdk.resource as resource

# Get all compute resources
compute_resources = resource.compute_list()

# Parse the response
if compute_resources['code'] == 200:
    pods = compute_resources['details']['pods_list']
    for pod in pods:
        if 'app-name' in pod.get('labels', {}):
            name = pod['labels']['app-name']
            status = pod['status']
            print(f"{name}: {status}")

List all database resources​

# Get all database resources
db_resources = resource.db_list()

if db_resources['code'] == 200:
    for db in db_resources['details']['pods_list']:
        print(f"Database: {db['labels'].get('app-name', 'unknown')}")

Troubleshooting resources​

Resource won't start​

  • Check image name is correct
  • Verify you have sufficient quota
  • Check startup command syntax

Can't connect to resource​

  • Verify ports are configured correctly
  • Check if resource is ready
  • Ensure ingress is enabled for external access

Resource keeps restarting​

  • Check logs for errors
  • Verify environment variables
  • Ensure sufficient memory allocated