refactor: improve runpod client

- Refactor RunPod client to use environment variables for API key.
- Update RunPod example script to reflect changes.
- Remove unused gql_builder.v file.
- Update README.md to reflect changes.
- Improve error handling and logging.
- Use json2 for JSON encoding/decoding.
- Update dependencies.
- Implemented more endpoints for managing pods.

Co-authored-by: mahmmoud.hassanein <mahmmoud.hassanein@gmail.com>
This commit is contained in:
2025-01-23 16:17:12 +02:00
parent 6f9d570a93
commit 01fff39e41
8 changed files with 610 additions and 556 deletions

View File

@@ -2,24 +2,24 @@
// import freeflowuniverse.herolib.core.base
import freeflowuniverse.herolib.clients.runpod
import json
import x.json2
// Example 1: Create client with direct API key
mut rp := runpod.get(
name: 'example1'
api_key: 'rpa_P77PNL3UHJ2XP0EC3XKYCH8M8BZREVLY9U4VGK4E1p4j68'
)!
// Create client with direct API key
// This uses RUNPOD_API_KEY from environment
mut rp := runpod.get()!
// Create a new on demand pod
on_demand_pod_response := rp.create_on_demand_pod(
name: 'RunPod Tensorflow'
image_name: 'runpod/tensorflow'
cloud_type: .all
cloud_type: 'ALL'
gpu_count: 1
volume_in_gb: 5
container_disk_in_gb: 5
min_memory_in_gb: 4
min_vcpu_count: 1
gpu_type_id: 'NVIDIA RTX 2000 Ada'
gpu_type_id: 'NVIDIA RTX A4000'
ports: '8888/http'
volume_mount_path: '/workspace'
env: [
@@ -32,51 +32,62 @@ on_demand_pod_response := rp.create_on_demand_pod(
println('Created pod with ID: ${on_demand_pod_response.id}')
// // create a spot pod
// spot_pod_response := rp.create_spot_pod(
// port: 1826
// bid_per_gpu: 0.2
// cloud_type: .secure
// gpu_count: 1
// volume_in_gb: 5
// container_disk_in_gb: 5
// min_vcpu_count: 1
// min_memory_in_gb: 4
// gpu_type_id: 'NVIDIA RTX 2000 Ada'
// name: 'RunPod Pytorch'
// image_name: 'runpod/pytorch'
// docker_args: ''
// ports: '8888/http'
// volume_mount_path: '/workspace'
// env: [
// runpod.EnvironmentVariableInput{
// key: 'JUPYTER_PASSWORD'
// value: 'rn51hunbpgtltcpac3ol'
// },
// ]
// )!
// println('Created spot pod with ID: ${spot_pod_response.id}')
// create a spot pod
spot_pod_response := rp.create_spot_pod(
port: 1826
bid_per_gpu: 0.2
cloud_type: 'SECURE'
gpu_count: 1
volume_in_gb: 5
container_disk_in_gb: 5
min_vcpu_count: 1
min_memory_in_gb: 4
gpu_type_id: 'NVIDIA RTX A4000'
name: 'RunPod Pytorch'
image_name: 'runpod/pytorch'
docker_args: ''
ports: '8888/http'
volume_mount_path: '/workspace'
env: [
runpod.EnvironmentVariableInput{
key: 'JUPYTER_PASSWORD'
value: 'rn51hunbpgtltcpac3ol'
},
]
)!
println('Created spot pod with ID: ${spot_pod_response.id}')
// // stop on-demand pod
// stop_on_demand_pod := rp.stop_pod(
// pod_id: '${on_demand_pod_response.id}'
// )!
// println('Stopped on-demand pod with ID: ${stop_on_demand_pod.id}')
// stop on-demand pod
stop_on_demand_pod := rp.stop_pod(
pod_id: '${on_demand_pod_response.id}'
)!
println('Stopped on-demand pod with ID: ${stop_on_demand_pod.id}')
// // stop spot pod
// stop_spot_pod := rp.stop_pod(
// pod_id: '${spot_pod_response.id}'
// )!
// println('Stopped spot pod with ID: ${stop_spot_pod.id}')
// stop spot pod
stop_spot_pod := rp.stop_pod(
pod_id: '${spot_pod_response.id}'
)!
println('Stopped spot pod with ID: ${stop_spot_pod.id}')
// // start on-demand pod
// start_on_demand_pod := rp.start_on_demand_pod(pod_id: '${on_demand_pod_response.id}', gpu_count: 1)!
// println('Started on demand pod with ID: ${start_on_demand_pod.id}')
// start on-demand pod
start_on_demand_pod := rp.start_on_demand_pod(pod_id: '${on_demand_pod_response.id}', gpu_count: 1)!
println('Started on demand pod with ID: ${on_demand_pod_response.id}')
// // start spot pod
// start_spot_pod := rp.start_spot_pod(
// pod_id: '${spot_pod_response.id}'
// gpu_count: 1
// bid_per_gpu: 0.2
// )!
// println('Started spot pod with ID: ${start_on_demand_pod.id}')
// start spot pod
start_spot_pod := rp.start_spot_pod(
pod_id: '${spot_pod_response.id}'
gpu_count: 1
bid_per_gpu: 0.2
)!
println('Started spot pod with ID: ${spot_pod_response.id}')
get_pod := rp.get_pod(
pod_id: '${spot_pod_response.id}'
)!
println('Get pod result: ${get_pod}')
rp.terminate_pod(pod_id: '${spot_pod_response.id}')!
println('pod with id ${spot_pod_response.id} is terminated')
rp.terminate_pod(pod_id: '${on_demand_pod_response.id}')!
println('pod with id ${on_demand_pod_response.id} is terminated')