GET /health

curl --request GET \
  --url http://localhost:8000/health

import requests

url = "http://localhost:8000/health"

response = requests.get(url)

print(response.text)

const options = {method: 'GET'};

fetch('http://localhost:8000/health', options)
  .then(res => res.json())
  .then(res => console.log(res))
  .catch(err => console.error(err));

<?php

$curl = curl_init();

curl_setopt_array($curl, [
  CURLOPT_PORT => "8000",
  CURLOPT_URL => "http://localhost:8000/health",
  CURLOPT_RETURNTRANSFER => true,
  CURLOPT_ENCODING => "",
  CURLOPT_MAXREDIRS => 10,
  CURLOPT_TIMEOUT => 30,
  CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
  CURLOPT_CUSTOMREQUEST => "GET",
]);

$response = curl_exec($curl);
$err = curl_error($curl);

curl_close($curl);

if ($err) {
  echo "cURL Error #:" . $err;
} else {
  echo $response;
}

package main

import (
	"fmt"
	"net/http"
	"io"
)

func main() {

	url := "http://localhost:8000/health"

	req, _ := http.NewRequest("GET", url, nil)

	res, _ := http.DefaultClient.Do(req)

	defer res.Body.Close()
	body, _ := io.ReadAll(res.Body)

	fmt.Println(string(body))

}

HttpResponse<String> response = Unirest.get("http://localhost:8000/health")
  .asString();

require 'uri'
require 'net/http'

url = URI("http://localhost:8000/health")

http = Net::HTTP.new(url.host, url.port)

request = Net::HTTP::Get.new(url)

response = http.request(request)
puts response.read_body

{
  "status": "<string>",
  "model": "<string>",
  "queue_size": 123,
  "device": "<string>",
  "200 OK": {},
  "500 Internal Server Error": {}
}

GET

health

GET /health

curl --request GET \
  --url http://localhost:8000/health

import requests

url = "http://localhost:8000/health"

response = requests.get(url)

print(response.text)

const options = {method: 'GET'};

fetch('http://localhost:8000/health', options)
  .then(res => res.json())
  .then(res => console.log(res))
  .catch(err => console.error(err));

<?php

$curl = curl_init();

curl_setopt_array($curl, [
  CURLOPT_PORT => "8000",
  CURLOPT_URL => "http://localhost:8000/health",
  CURLOPT_RETURNTRANSFER => true,
  CURLOPT_ENCODING => "",
  CURLOPT_MAXREDIRS => 10,
  CURLOPT_TIMEOUT => 30,
  CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
  CURLOPT_CUSTOMREQUEST => "GET",
]);

$response = curl_exec($curl);
$err = curl_error($curl);

curl_close($curl);

if ($err) {
  echo "cURL Error #:" . $err;
} else {
  echo $response;
}

package main

import (
	"fmt"
	"net/http"
	"io"
)

func main() {

	url := "http://localhost:8000/health"

	req, _ := http.NewRequest("GET", url, nil)

	res, _ := http.DefaultClient.Do(req)

	defer res.Body.Close()
	body, _ := io.ReadAll(res.Body)

	fmt.Println(string(body))

}

HttpResponse<String> response = Unirest.get("http://localhost:8000/health")
  .asString();

require 'uri'
require 'net/http'

url = URI("http://localhost:8000/health")

http = Net::HTTP.new(url.host, url.port)

request = Net::HTTP::Get.new(url)

response = http.request(request)
puts response.read_body

{
  "status": "<string>",
  "model": "<string>",
  "queue_size": 123,
  "device": "<string>",
  "200 OK": {},
  "500 Internal Server Error": {}
}

Overview

Check the health and status of the HyperGen server. This endpoint provides information about server status, loaded model, queue size, and device.

This endpoint does not require authentication, making it ideal for monitoring and health checks.

Authentication

string

Not required - This endpoint is publicly accessible

Request

No request body or parameters required.

Response

string

Server health statusValues:

"healthy" - Server is running and ready to process requests
"unhealthy" - Server is experiencing issues (not currently implemented)

string

The model identifier that was loaded at server startupExample: "stabilityai/stable-diffusion-xl-base-1.0"

integer

Current number of pending requests in the queueRange: 0 to max_queue_size (default: 100)

0 - No pending requests, server is idle
>0 - Requests are waiting to be processed

string

Device the model is running onValues:

"cuda" - NVIDIA GPU
"cuda:0", "cuda:1", etc. - Specific GPU device
"cpu" - CPU
"mps" - Apple Silicon GPU

Examples

Basic Health Check

curl http://localhost:8000/health

import requests

response = requests.get("http://localhost:8000/health")
health = response.json()

print(f"Status: {health['status']}")
print(f"Model: {health['model']}")
print(f"Queue: {health['queue_size']}")
print(f"Device: {health['device']}")

const response = await fetch('http://localhost:8000/health');
const health = await response.json();

console.log(`Status: ${health.status}`);
console.log(`Model: ${health.model}`);
console.log(`Queue: ${health.queue_size}`);
console.log(`Device: ${health.device}`);

Response

{
  "status": "healthy",
  "model": "stabilityai/stable-diffusion-xl-base-1.0",
  "queue_size": 0,
  "device": "cuda"
}

Server Under Load

When the server has pending requests:

curl http://localhost:8000/health

{
  "status": "healthy",
  "model": "stabilityai/sdxl-turbo",
  "queue_size": 5,
  "device": "cuda:0"
}

Use Cases

Monitoring Script

Monitor server health and queue status:

import requests
import time

def check_health():
    try:
        response = requests.get("http://localhost:8000/health", timeout=5)
        health = response.json()

        if health["status"] == "healthy":
            print(f" Server healthy - Queue: {health['queue_size']}")
            return True
        else:
            print(f" Server unhealthy")
            return False
    except Exception as e:
        print(f" Server unreachable: {e}")
        return False

# Monitor every 30 seconds
while True:
    check_health()
    time.sleep(30)

Load Balancer Health Check

Use for load balancer health checks (e.g., AWS ALB, nginx):

# nginx configuration
upstream hypergen_servers {
    server 10.0.1.10:8000;
    server 10.0.1.11:8000;
    server 10.0.1.12:8000;
}

server {
    location / {
        proxy_pass http://hypergen_servers;

        # Health check
        health_check uri=/health interval=10s;
    }
}

Kubernetes Liveness Probe

apiVersion: v1
kind: Pod
metadata:
  name: hypergen-server
spec:
  containers:
  - name: hypergen
    image: hypergen:latest
    ports:
    - containerPort: 8000
    livenessProbe:
      httpGet:
        path: /health
        port: 8000
      initialDelaySeconds: 30
      periodSeconds: 10
    readinessProbe:
      httpGet:
        path: /health
        port: 8000
      initialDelaySeconds: 30
      periodSeconds: 5

Wait for Server Ready

Wait for server to be ready before sending requests:

import requests
import time

def wait_for_server(url="http://localhost:8000", timeout=60):
    """Wait for server to be healthy."""
    start = time.time()

    while time.time() - start < timeout:
        try:
            response = requests.get(f"{url}/health", timeout=5)
            if response.json()["status"] == "healthy":
                print("Server is ready!")
                return True
        except:
            pass

        print("Waiting for server...")
        time.sleep(2)

    raise TimeoutError("Server did not become healthy in time")

# Wait for server, then make requests
wait_for_server()

response = requests.post(
    "http://localhost:8000/v1/images/generations",
    json={"prompt": "A cat"}
)

Queue Monitoring

Monitor queue size and alert when backlog grows:

import requests
import time

def monitor_queue(threshold=10):
    """Alert when queue size exceeds threshold."""
    while True:
        try:
            response = requests.get("http://localhost:8000/health")
            health = response.json()
            queue_size = health["queue_size"]

            if queue_size >= threshold:
                print(f"�  WARNING: Queue size is {queue_size} (threshold: {threshold})")
                # Send alert (email, Slack, PagerDuty, etc.)
            else:
                print(f"Queue size: {queue_size}")

        except Exception as e:
            print(f"Error checking health: {e}")

        time.sleep(10)

monitor_queue(threshold=10)

Automatic Scaling Decision

Use queue size to make scaling decisions:

import requests

def should_scale_up(queue_threshold=20):
    """Determine if we should add more server instances."""
    try:
        response = requests.get("http://localhost:8000/health")
        health = response.json()

        if health["queue_size"] > queue_threshold:
            print(f"Queue size {health['queue_size']} exceeds threshold {queue_threshold}")
            print("Recommendation: Scale up")
            return True
        else:
            print(f"Queue size {health['queue_size']} is within limits")
            return False

    except Exception as e:
        print(f"Error: {e}")
        return False

# Check if scaling is needed
if should_scale_up():
    # Trigger auto-scaling (AWS Auto Scaling, Kubernetes HPA, etc.)
    pass

Metrics Collection

Prometheus Exporter Example

Export metrics for Prometheus monitoring:

from prometheus_client import start_http_server, Gauge
import requests
import time

# Define metrics
queue_size_gauge = Gauge('hypergen_queue_size', 'Current queue size')
server_status = Gauge('hypergen_server_healthy', 'Server health status (1=healthy, 0=unhealthy)')

def collect_metrics():
    while True:
        try:
            response = requests.get("http://localhost:8000/health", timeout=5)
            health = response.json()

            # Update metrics
            queue_size_gauge.set(health["queue_size"])
            server_status.set(1 if health["status"] == "healthy" else 0)

        except Exception as e:
            print(f"Error collecting metrics: {e}")
            server_status.set(0)

        time.sleep(5)

# Start Prometheus metrics server
start_http_server(9090)
collect_metrics()

Response Status Codes

success

Server is reachable and health check succeeded

error

Server error (rare, as endpoint is very simple)

The /health endpoint should always return 200 OK if the server is running, even if the queue is full or the server is under heavy load.

Best Practices

Monitoring

Poll the /health endpoint every 10-30 seconds
Monitor queue_size to detect backlog
Alert when status is not "healthy"
Track queue_size trends over time

Load Balancing

Use /health for load balancer health checks
Set appropriate timeout (5-10 seconds)
Configure retry logic
Don’t route traffic to instances with high queue_size

Auto-Scaling

Scale up when queue_size consistently exceeds threshold
Scale down when queue_size is consistently 0
Use average queue size over time window (e.g., 5 minutes)
Avoid flapping by using hysteresis

Deployment

Wait for /health to return "healthy" before routing traffic
Use in readiness probes for orchestration platforms
Check /health before running integration tests
Include in pre-deployment smoke tests

Troubleshooting

Server Not Responding

If /health endpoint is not responding:

Check if server is running: ps aux | grep hypergen
Check server logs for errors
Verify port is not blocked by firewall
Ensure server started successfully (check for CUDA errors)

High Queue Size

If queue_size is consistently high:

Generation is too slow (consider using SDXL Turbo)
Too many concurrent requests
Image sizes are too large
Need to scale horizontally (add more servers)

Generate Images

Generate images (the endpoint that fills the queue)

List Models

Get information about the loaded model

POST /v1/images/generations GET /v1/models

Python API

HTTP API

GET /health

Overview

Authentication

Request

Response

Examples

Basic Health Check

Response

Server Under Load

Use Cases

Monitoring Script

Load Balancer Health Check

Kubernetes Liveness Probe

Wait for Server Ready

Queue Monitoring

Automatic Scaling Decision

Metrics Collection

Prometheus Exporter Example

Response Status Codes

Best Practices

Troubleshooting

Server Not Responding

High Queue Size

Generate Images

List Models

​Overview

​Authentication

​Request

​Response

​Examples

​Basic Health Check

​Response

​Server Under Load

​Use Cases

​Monitoring Script

​Load Balancer Health Check

​Kubernetes Liveness Probe

​Wait for Server Ready

​Queue Monitoring

​Automatic Scaling Decision

​Metrics Collection

​Prometheus Exporter Example

​Response Status Codes

​Best Practices

​Troubleshooting

​Server Not Responding

​High Queue Size

​Related Endpoints

Generate Images

List Models

Overview

Authentication

Request

Response

Examples

Basic Health Check

Response

Server Under Load

Use Cases

Monitoring Script

Load Balancer Health Check

Kubernetes Liveness Probe

Wait for Server Ready

Queue Monitoring

Automatic Scaling Decision

Metrics Collection

Prometheus Exporter Example

Response Status Codes

Best Practices

Troubleshooting

Server Not Responding

High Queue Size

Related Endpoints