129 lines
4.0 KiB
Python
129 lines
4.0 KiB
Python
#!/usr/bin/env python3
|
|
"""
|
|
Ollama GPU Provider Test
|
|
Submits an inference job with prompt "hello" and verifies completion.
|
|
"""
|
|
|
|
import argparse
|
|
import sys
|
|
import time
|
|
from typing import Optional
|
|
|
|
import httpx
|
|
|
|
DEFAULT_COORDINATOR = "http://127.0.0.1:18000"
|
|
DEFAULT_API_KEY = "${CLIENT_API_KEY}"
|
|
DEFAULT_PROMPT = "hello"
|
|
DEFAULT_TIMEOUT = 180
|
|
POLL_INTERVAL = 3
|
|
|
|
|
|
def submit_job(client: httpx.Client, base_url: str, api_key: str, prompt: str) -> Optional[str]:
|
|
payload = {
|
|
"payload": {
|
|
"type": "inference",
|
|
"prompt": prompt,
|
|
"parameters": {"prompt": prompt},
|
|
},
|
|
"ttl_seconds": 900,
|
|
}
|
|
response = client.post(
|
|
f"{base_url}/v1/jobs",
|
|
headers={"X-Api-Key": api_key, "Content-Type": "application/json"},
|
|
json=payload,
|
|
timeout=10,
|
|
)
|
|
if response.status_code != 201:
|
|
print(f"❌ Job submission failed: {response.status_code} {response.text}")
|
|
return None
|
|
return response_seen_id(response)
|
|
|
|
|
|
def response_seen_id(response: httpx.Response) -> Optional[str]:
|
|
try:
|
|
return response.json().get("job_id")
|
|
except Exception:
|
|
return None
|
|
|
|
|
|
def fetch_status(client: httpx.Client, base_url: str, api_key: str, job_id: str) -> Optional[dict]:
|
|
response = client.get(
|
|
f"{base_url}/v1/jobs/{job_id}",
|
|
headers={"X-Api-Key": api_key},
|
|
timeout=10,
|
|
)
|
|
if response.status_code != 200:
|
|
print(f"❌ Status check failed: {response.status_code} {response.text}")
|
|
return None
|
|
return response.json()
|
|
|
|
|
|
def fetch_result(client: httpx.Client, base_url: str, api_key: str, job_id: str) -> Optional[dict]:
|
|
response = client.get(
|
|
f"{base_url}/v1/jobs/{job_id}/result",
|
|
headers={"X-Api-Key": api_key},
|
|
timeout=10,
|
|
)
|
|
if response.status_code != 200:
|
|
print(f"❌ Result fetch failed: {response.status_code} {response.text}")
|
|
return None
|
|
return response.json()
|
|
|
|
|
|
def main() -> int:
|
|
parser = argparse.ArgumentParser(description="Ollama GPU provider end-to-end test")
|
|
parser.add_argument("--url", default=DEFAULT_COORDINATOR, help="Coordinator base URL")
|
|
parser.add_argument("--api-key", default=DEFAULT_API_KEY, help="Client API key")
|
|
parser.add_argument("--prompt", default=DEFAULT_PROMPT, help="Prompt to send")
|
|
parser.add_argument("--timeout", type=int, default=DEFAULT_TIMEOUT, help="Timeout in seconds")
|
|
args = parser.parse_args()
|
|
|
|
with httpx.Client() as client:
|
|
print("🧪 Submitting GPU provider job...")
|
|
job_id = submit_job(client, args.url, args.api_key, args.prompt)
|
|
if not job_id:
|
|
return 1
|
|
print(f"✅ Job submitted: {job_id}")
|
|
|
|
deadline = time.time() + args.timeout
|
|
status = None
|
|
while time.time() < deadline:
|
|
status = fetch_status(client, args.url, args.api_key, job_id)
|
|
if not status:
|
|
return 1
|
|
state = status.get("state")
|
|
print(f"⏳ Job state: {state}")
|
|
if state == "COMPLETED":
|
|
break
|
|
if state in {"FAILED", "CANCELED", "EXPIRED"}:
|
|
print(f"❌ Job ended in state: {state}")
|
|
return 1
|
|
time.sleep(POLL_INTERVAL)
|
|
|
|
if not status or status.get("state") != "COMPLETED":
|
|
print("❌ Job did not complete within timeout")
|
|
return 1
|
|
|
|
result = fetch_result(client, args.url, args.api_key, job_id)
|
|
if result is None:
|
|
return 1
|
|
|
|
payload = result.get("result") or {}
|
|
output = payload.get("output")
|
|
receipt = result.get("receipt")
|
|
if not output:
|
|
print("❌ Missing output in job result")
|
|
return 1
|
|
if not receipt:
|
|
print("❌ Missing receipt in job result (payment/settlement not recorded)")
|
|
return 1
|
|
|
|
print("✅ GPU provider job completed")
|
|
print(f"📝 Output: {output}")
|
|
print(f"🧾 Receipt ID: {receipt.get('receipt_id')}")
|
|
return 0
|
|
|
|
|
|
if __name__ == "__main__":
|
|
sys.exit(main())
|