Important

FnServe is archived. Its goal (Lambda-style functions on your own hardware) now lives in HomeCloud, which runs functions on AWS's official runtime images with versions, aliases, layers, async invocation, triggers and the AWS Lambda API. Install HomeCloud: https://homecloud.pages.dev

FnServe

Serverless, without the cloud.

FnServe is a self-hosted, open-source serverless runtime. It brings AWS Lambda–style functions to your own machine or infrastructure, without Kubernetes, vendor lock-in, or hidden costs.

FnServe is part of the HomeCloud ecosystem, but designed to run standalone as well. You can use it independently, or integrate it into your HomeCloud stack when ready.

Why FnServe?

• Lightweight — a single Go binary, no orchestration required
• Language-agnostic — write functions in Go, Python, or anything that speaks JSON
• Local-first — run on your laptop, homelab, or edge server
• Practical — built for small projects and real-world use, not demos
• Scalable — built-in concurrency controls and monitoring
• Developer-friendly — hot reload in development mode

Features

• Run functions once with JSON input/output
• Serve functions as HTTP endpoints
• Development mode with hot reload
• Advanced context injection (request ID, timestamp, deadlines, env vars, parameters)
• Tracing support with trace IDs, span IDs, and parent IDs
• Concurrency controls for high-scale deployments
• Health and statistics endpoints
• Works with Go and Python out of the box
• Graceful shutdown and error handling
• Function timeouts and cancellation

Installation

From source

git clone https://github.com/homecloudhq/fnserve.git
cd fnserve
go build -o fnserve .

Using Go install

go install github.com/homecloudhq/fnserve@latest

Quick Start

Run a function once

# Using a JSON string
fnserve run ./functions/hello.py --event '{"name": "World"}'

# Using a JSON file
fnserve run ./functions/hello.py --event event.json

# Using stdin
echo '{"name": "World"}' | fnserve run ./functions/hello.py

Serve functions as HTTP endpoints

# Basic usage
fnserve serve ./functions --port 8080

# With concurrency controls
fnserve serve ./functions --port 8080 --concurrency 100 --timeout 30s

Development mode with hot reload

fnserve dev ./functions --port 8080 --concurrency 10 --timeout 10s

Functions in ./functions will be available at:

POST /<function-name>

CLI Commands

Root Command

fnserve [command]

Available Commands:

• run: Run a function once with an event
• serve: Serve a directory of functions as HTTP endpoints
• dev: Run in development mode with hot-reload

Run Command

fnserve run [function] [flags]

Flags:

• --event string: Event JSON to pass to function (can be a JSON string or a file path)

Serve Command

fnserve serve [directory] [flags]

Flags:

• --port int: Port to listen on (default 8080)
• --concurrency int: Maximum number of concurrent function executions (default 100)
• --timeout duration: Request timeout duration (default 30s)
• --workers int: Size of the worker pool (default 10)

Dev Command

fnserve dev [directory] [flags]

Flags:

• --port int: Port to listen on (default 8080)
• --concurrency int: Maximum number of concurrent function executions (default 10)
• --timeout duration: Request timeout duration (default 30s)

Function Contract

Input/Output

• Input: JSON (stdin or HTTP body)
• Output: JSON (stdout or HTTP response)
• Must complete within configured timeout

Context

Functions receive a context object with the following structure:

{
  "request_id": "req-uuid",
  "timestamp": "2025-08-18T12:00:00Z",
  "deadline": 30000000000,
  "parameters": {
    "param1": "value1",
    "param2": "value2"
  },
  "env": {
    "X-API-Key": "api-key-value",
    "Authorization": "auth-header"
  },
  "tracing": {
    "trace_id": "trace-uuid",
    "span_id": "span-uuid",
    "parent_id": "parent-span-id"
  }
}

Creating Functions

Python Functions

import sys, json, os

def handler(event, context):
    # Access context information
    request_id = context.get("request_id", "unknown")
    trace_id = context.get("tracing", {}).get("trace_id", "unknown")
    
    # Access request parameters
    params = context.get("parameters", {})
    
    return {
        "message": f"Hello {event.get('name', 'World')}",
        "request_id": request_id,
        "trace_id": trace_id,
        "params": params
    }

if __name__ == "__main__":
    # Load event from stdin
    try:
        event = json.load(sys.stdin)
    except json.JSONDecodeError:
        event = {}
    
    # Get context from environment variable
    ctx = {}
    if "FN_CONTEXT" in os.environ:
        try:
            ctx = json.loads(os.environ["FN_CONTEXT"])
        except:
            pass
            
    result = handler(event, ctx)
    print(json.dumps(result))

Go Functions

package main

import (
	"encoding/json"
	"fmt"
	"io"
	"os"
	"time"
)

type Event struct {
	Name string `json:"name"`
}

type Context struct {
	RequestID  string                 `json:"request_id"`
	Timestamp  time.Time              `json:"timestamp"`
	Deadline   time.Duration          `json:"deadline"`
	Parameters map[string]string      `json:"parameters"`
	Env        map[string]string      `json:"env"`
	Tracing    map[string]interface{} `json:"tracing"`
}

type Response struct {
	Message   string    `json:"message"`
	RequestID string    `json:"request_id"`
	Timestamp time.Time `json:"timestamp"`
}

func main() {
	// Read input from stdin
	body, err := io.ReadAll(os.Stdin)
	if err != nil {
		fmt.Fprintf(os.Stderr, "Error reading input: %v\n", err)
		os.Exit(1)
	}

	// Parse the event
	var event Event
	if err := json.Unmarshal(body, &event); err != nil {
		fmt.Fprintf(os.Stderr, "Error parsing event: %v\n", err)
		os.Exit(1)
	}

	// Get the context from environment
	var ctx Context
	if contextJSON := os.Getenv("FN_CONTEXT"); contextJSON != "" {
		json.Unmarshal([]byte(contextJSON), &ctx)
	}

	// Create response
	response := Response{
		Message:   fmt.Sprintf("Hello, %s!", event.Name),
		RequestID: ctx.RequestID,
		Timestamp: time.Now(),
	}

	// Output JSON response
	output, _ := json.Marshal(response)
	fmt.Println(string(output))
}

HTTP API

Function Endpoints

Each function in your functions directory is exposed at:

POST /<function-name>

Request

• Method: POST
• Body: JSON payload
• Headers:
  • X-Trace-ID: Optional trace ID for distributed tracing
  • X-Parent-Span: Optional parent span ID for tracing
  • X-API-Key: Optional API key (passed to function)
  • Authorization: Optional auth header (passed to function)

Query Parameters

All query parameters are passed to the function context:

POST /hello?param1=value1&param2=value2

Health Check Endpoint

GET /health

Response:

{"status":"ok"}

Statistics Endpoint

GET /stats

Response:

{
  "active_requests": 2,
  "total_requests": 150,
  "success_requests": 145,
  "failed_requests": 3,
  "avg_execution_ms": 45
}

Concurrency and Scaling

FnServe provides built-in concurrency controls:

• Concurrent Requests Limit: Control the maximum number of concurrent function executions
• Request Timeout: Set execution time limits per function
• Worker Pool: Configure the worker pool size for function execution
• Graceful Shutdown: Properly handle in-flight requests during shutdown

Tracing and Monitoring

• Request IDs: Every function execution gets a unique ID
• Trace and Span IDs: Support for distributed tracing
• Execution Statistics: Track success, failure, and timing metrics

Examples

See the examples directory for more detailed usage examples.

Roadmap

• Support for more runtimes (Node.js, Ruby, etc.)
• Local function deployment
• Function versioning
• Environment management
• Authentication and authorization
• Function logs and metrics storage

License

AGPL-3.0