version: add documentation and tests

.github/copilot-instructions.md

@@ -0,0 +1 @@
+../CLAUDE.md

CLAUDE.md

@@ -0,0 +1,163 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Commands
+
+### Testing
+- Run all tests: `go test ./...`
+- Run tests with verbose output: `go test -v ./...`
+- Run tests for specific package: `go test ./config`
+- Run specific test: `go test -run TestConfigBool ./config`
+
+### Building
+- Build all packages: `go build ./...`
+- Check module dependencies: `go mod tidy`
+- Verify dependencies: `go mod verify`
+
+### Code Quality
+- Format code: `go fmt ./...`
+- Vet code: `go vet ./...`
+- Run static analysis: `staticcheck ./...` (if available)
+
+## Architecture
+
+This is a common library (`go.ntppool.org/common`) providing shared infrastructure for the NTP Pool project. The codebase emphasizes observability, security, and modern Go practices.
+
+### Core Components
+
+**Web Service Foundation:**
+- `ekko/` - Enhanced Echo web framework with pre-configured middleware (OpenTelemetry, Prometheus, logging, security headers)
+- `health/` - Standalone health check HTTP server with `/__health` endpoint
+- `metricsserver/` - Prometheus metrics exposure via `/metrics` endpoint
+
+**Observability Stack:**
+- `logger/` - Structured logging with OpenTelemetry trace integration and multiple output formats
+- `tracing/` - OpenTelemetry distributed tracing with OTLP export support
+- `metricsserver/` - Prometheus metrics with custom registry
+
+**Configuration & Environment:**
+- `config/` - Environment-based configuration with code-generated accessors (`config_accessor.go`)
+- `version/` - Build metadata and version information with Cobra CLI integration
+
+**Security & Communication:**
+- `apitls/` - TLS certificate management with automatic renewal via certman
+- `kafka/` - Kafka client wrapper with TLS support for log streaming
+- `xff/fastlyxff/` - Fastly CDN IP range management for trusted proxy handling
+
+**Utilities:**
+- `ulid/` - Thread-safe ULID generation with monotonic ordering
+- `timeutil/` - JSON-serializable duration types
+- `types/` - Shared data structures (LogScoreAttributes for NTP server scoring)
+
+### Key Patterns
+
+**Functional Options:** Used extensively in `ekko/` for flexible service configuration
+**Interface-Based Design:** `CertificateProvider` in `apitls/` for pluggable certificate management
+**Context Propagation:** Throughout the codebase for cancellation and tracing
+**Graceful Shutdown:** Implemented in web servers and background services
+
+### Dependencies
+
+The codebase heavily uses:
+- Echo web framework with custom middleware stack
+- OpenTelemetry for observability (traces, metrics, logs)
+- Prometheus for metrics collection
+- Kafka for message streaming
+- Cobra for CLI applications
+
+### Code Generation
+
+`config/config_accessor.go` is generated - modify `config.go` and regenerate accessors when adding new configuration options.
+
+## Package Overview
+
+### `apitls/`
+TLS certificate management with automatic renewal support via certman. Provides a CA pool for trusted certificates and interfaces for pluggable certificate providers. Used for secure inter-service communication.
+
+### `config/`
+Environment-based configuration system with code-generated accessor methods. Handles deployment mode, hostname configuration, and TLS settings. Provides URL building utilities for web and management interfaces.
+
+### `ekko/`
+Enhanced Echo web framework wrapper with pre-configured middleware stack including OpenTelemetry tracing, Prometheus metrics, structured logging, gzip compression, and security headers. Supports HTTP/2 with graceful shutdown.
+
+### `health/`
+Standalone HTTP health check server that runs independently from the main application. Exposes `/__health` endpoint with configurable health handlers, timeouts, and graceful shutdown capabilities.
+
+### `kafka/`
+Kafka client wrapper with TLS support for secure log streaming. Provides connection management, broker discovery, and reader/writer factories with compression and batching optimizations.
+
+### `logger/`
+Structured logging system with OpenTelemetry trace integration. Supports multiple output formats (text, OTLP) with configurable log levels, systemd compatibility, and context-aware logging.
+
+### `metricsserver/`
+Dedicated Prometheus metrics HTTP server with custom registry isolation. Exposes `/metrics` endpoint with OpenMetrics support and graceful shutdown handling.
+
+### `timeutil/`
+JSON-serializable duration types that support both string parsing ("30s", "5m") and numeric nanosecond values. Compatible with configuration files and REST APIs.
+
+### `tracing/`
+OpenTelemetry distributed tracing setup with support for OTLP export via gRPC or HTTP. Handles resource detection, propagation, and automatic instrumentation with configurable TLS.
+
+### `types/`
+Shared data structures for the NTP Pool project. Currently contains `LogScoreAttributes` for NTP server scoring with JSON and SQL database compatibility.
+
+### `ulid/`
+Thread-safe ULID (Universally Unique Lexicographically Sortable Identifier) generation using cryptographically secure randomness. Optimized for simplicity and performance in high-concurrency environments.
+
+### `version/`
+Build metadata and version information system with Git integration. Provides CLI commands for Cobra and Kong frameworks, Prometheus build info metrics, and semantic version validation.
+
+### `xff/fastlyxff/`
+Fastly CDN IP range management for trusted proxy handling. Parses Fastly's IP ranges JSON file and generates Echo framework trust options for proper client IP extraction.
+
+## Go Development Best Practices
+
+### Code Style
+- Follow standard Go formatting (`go fmt ./...`)
+- Use `go vet ./...` for static analysis
+- Run `staticcheck ./...` when available
+- Prefer short, descriptive variable names
+- Use interfaces for testability and flexibility
+
+### Error Handling
+- Always handle errors explicitly
+- Use `errors.Join()` for combining multiple errors
+- Wrap errors with context using `fmt.Errorf("context: %w", err)`
+- Return early on errors to reduce nesting
+
+### Testing
+- Write table-driven tests when testing multiple scenarios
+- Use `t.Helper()` in test helper functions
+- Test error conditions, not just happy paths
+- Use `testing.Short()` for integration tests that can be skipped
+
+### Concurrency
+- Use contexts for cancellation and timeouts
+- Prefer channels for communication over shared memory
+- Use `sync.Once` for one-time initialization
+- Always call `defer cancel()` after `context.WithCancel()`
+
+### Performance
+- Use `sync.Pool` for frequently allocated objects
+- Prefer slices over arrays for better performance
+- Use `strings.Builder` for string concatenation in loops
+- Profile before optimizing with `go tool pprof`
+
+### Observability
+- Use structured logging with key-value pairs
+- Add OpenTelemetry spans for external calls
+- Include trace IDs in error messages
+- Use metrics for monitoring application health
+
+### Dependencies
+- Keep dependencies minimal and well-maintained
+- Use `go mod tidy` to clean up unused dependencies
+- Pin major versions to avoid breaking changes
+- Prefer standard library when possible
+
+### Security
+- Never log sensitive information (passwords, tokens)
+- Use `crypto/rand` for cryptographic randomness
+- Validate all inputs at API boundaries
+- Use TLS for all network communication

README.md

@@ -0,0 +1,20 @@
+
+Common library for the NTP Pool project with shared infrastructure components.
+
+## Packages
+
+- **apitls** - TLS setup for NTP Pool internal services with embedded CA
+- **config** - NTP Pool project configuration with environment variables
+- **ekko** - Enhanced Echo web framework with observability middleware
+- **health** - Standalone health check HTTP server
+- **kafka** - Kafka client wrapper with TLS support
+- **logger** - Structured logging with OpenTelemetry integration
+- **metricsserver** - Prometheus metrics HTTP server
+- **timeutil** - JSON-serializable duration types
+- **tracing** - OpenTelemetry distributed tracing setup
+- **types** - Shared data structures for NTP Pool
+- **ulid** - Thread-safe ULID generation
+- **version** - Build metadata and version information
+- **xff/fastlyxff** - Fastly CDN IP range management
+
+[![Go Reference](https://pkg.go.dev/badge/go.ntppool.org/common.svg)](https://pkg.go.dev/go.ntppool.org/common)

health/health_server.go

@@ -1,3 +1,9 @@
+// Package health provides a standalone HTTP server for health checks.
+//
+// This package implements a simple health check server that can be used
+// to expose health status endpoints for monitoring and load balancing.
+// It supports custom health check handlers and provides structured logging
+// with graceful shutdown capabilities.
 package health
 
 import (
@@ -11,11 +17,19 @@ import (
 	"golang.org/x/sync/errgroup"
 )
 
+// Server is a standalone HTTP server dedicated to health checks.
+// It runs separately from the main application server to ensure health
+// checks remain available even if the main server is experiencing issues.
+//
+// The server includes built-in timeouts, graceful shutdown, and structured
+// logging for monitoring and debugging health check behavior.
 type Server struct {
 	log      *slog.Logger
 	healthFn http.HandlerFunc
 }
 
+// NewServer creates a new health check server with the specified health handler.
+// If healthFn is nil, a default handler that returns HTTP 200 "ok" is used.
 func NewServer(healthFn http.HandlerFunc) *Server {
 	if healthFn == nil {
 		healthFn = basicHealth
@@ -27,10 +41,13 @@ func NewServer(healthFn http.HandlerFunc) *Server {
 	return srv
 }
 
+// SetLogger replaces the default logger with a custom one.
 func (srv *Server) SetLogger(log *slog.Logger) {
 	srv.log = log
 }
 
+// Listen starts the health server on the specified port and blocks until ctx is cancelled.
+// The server exposes the health handler at "/__health" with graceful shutdown support.
 func (srv *Server) Listen(ctx context.Context, port int) error {
 	srv.log.Info("starting health listener", "port", port)
 
@@ -72,8 +89,7 @@ func (srv *Server) Listen(ctx context.Context, port int) error {
 	return g.Wait()
 }
 
-// HealthCheckListener runs simple http server on the specified port for
-// health check probes
+// HealthCheckListener runs a simple HTTP server on the specified port for health check probes.
 func HealthCheckListener(ctx context.Context, port int, log *slog.Logger) error {
 	srv := NewServer(nil)
 	srv.SetLogger(log)

metricsserver/metrics.go

@@ -1,3 +1,8 @@
+// Package metricsserver provides a standalone HTTP server for exposing Prometheus metrics.
+//
+// This package implements a dedicated metrics server that exposes application metrics
+// via HTTP. It uses a custom Prometheus registry to avoid conflicts with other metric
+// collectors and provides graceful shutdown capabilities.
 package metricsserver
 
 import (
@@ -13,10 +18,13 @@ import (
 	"go.ntppool.org/common/logger"
 )
 
+// Metrics provides a custom Prometheus registry and HTTP handlers for metrics exposure.
+// It isolates application metrics from the default global registry.
 type Metrics struct {
 	r *prometheus.Registry
 }
 
+// New creates a new Metrics instance with a custom Prometheus registry.
 func New() *Metrics {
 	r := prometheus.NewRegistry()
 
@@ -27,10 +35,13 @@ func New() *Metrics {
 	return m
 }
 
+// Registry returns the custom Prometheus registry.
+// Use this to register your application's metrics collectors.
 func (m *Metrics) Registry() *prometheus.Registry {
 	return m.r
 }
 
+// Handler returns an HTTP handler for the /metrics endpoint with OpenMetrics support.
 func (m *Metrics) Handler() http.Handler {
 	log := logger.NewStdLog("prom http", false, nil)
 
@@ -41,9 +52,8 @@ func (m *Metrics) Handler() http.Handler {
 	})
 }
 
-// ListenAndServe starts a goroutine with a server running on
-// the specified port. The server will shutdown and return when
-// the provided context is done
+// ListenAndServe starts a metrics server on the specified port and blocks until ctx is done.
+// The server exposes the metrics handler and shuts down gracefully when the context is cancelled.
 func (m *Metrics) ListenAndServe(ctx context.Context, port int) error {
 	log := logger.Setup()
 

metricsserver/metrics_test.go

@@ -0,0 +1,242 @@
+package metricsserver
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/prometheus/client_golang/prometheus"
+)
+
+func TestNew(t *testing.T) {
+	metrics := New()
+	
+	if metrics == nil {
+		t.Fatal("New returned nil")
+	}
+	
+	if metrics.r == nil {
+		t.Error("metrics registry is nil")
+	}
+}
+
+func TestRegistry(t *testing.T) {
+	metrics := New()
+	registry := metrics.Registry()
+	
+	if registry == nil {
+		t.Fatal("Registry() returned nil")
+	}
+	
+	if registry != metrics.r {
+		t.Error("Registry() did not return the metrics registry")
+	}
+	
+	// Test that we can register a metric
+	counter := prometheus.NewCounter(prometheus.CounterOpts{
+		Name: "test_counter",
+		Help: "A test counter",
+	})
+	
+	err := registry.Register(counter)
+	if err != nil {
+		t.Errorf("failed to register metric: %v", err)
+	}
+	
+	// Test that the metric is registered
+	metricFamilies, err := registry.Gather()
+	if err != nil {
+		t.Errorf("failed to gather metrics: %v", err)
+	}
+	
+	found := false
+	for _, mf := range metricFamilies {
+		if mf.GetName() == "test_counter" {
+			found = true
+			break
+		}
+	}
+	
+	if !found {
+		t.Error("registered metric not found in registry")
+	}
+}
+
+func TestHandler(t *testing.T) {
+	metrics := New()
+	
+	// Register a test metric
+	counter := prometheus.NewCounterVec(
+		prometheus.CounterOpts{
+			Name: "test_requests_total",
+			Help: "Total number of test requests",
+		},
+		[]string{"method"},
+	)
+	metrics.Registry().MustRegister(counter)
+	counter.WithLabelValues("GET").Inc()
+	
+	// Test the handler
+	handler := metrics.Handler()
+	if handler == nil {
+		t.Fatal("Handler() returned nil")
+	}
+	
+	// Create a test request
+	req := httptest.NewRequest("GET", "/metrics", nil)
+	recorder := httptest.NewRecorder()
+	
+	// Call the handler
+	handler.ServeHTTP(recorder, req)
+	
+	// Check response
+	resp := recorder.Result()
+	defer resp.Body.Close()
+	
+	if resp.StatusCode != http.StatusOK {
+		t.Errorf("expected status 200, got %d", resp.StatusCode)
+	}
+	
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		t.Fatalf("failed to read response body: %v", err)
+	}
+	
+	bodyStr := string(body)
+	
+	// Check for our test metric
+	if !strings.Contains(bodyStr, "test_requests_total") {
+		t.Error("test metric not found in metrics output")
+	}
+	
+	// Check for OpenMetrics format indicators
+	if !strings.Contains(bodyStr, "# TYPE") {
+		t.Error("metrics output missing TYPE comments")
+	}
+}
+
+func TestListenAndServe(t *testing.T) {
+	metrics := New()
+	
+	// Register a test metric
+	counter := prometheus.NewCounterVec(
+		prometheus.CounterOpts{
+			Name: "test_requests_total",
+			Help: "Total number of test requests",
+		},
+		[]string{"method"},
+	)
+	metrics.Registry().MustRegister(counter)
+	counter.WithLabelValues("GET").Inc()
+	
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	
+	// Start server in a goroutine
+	errCh := make(chan error, 1)
+	go func() {
+		// Use a high port number to avoid conflicts
+		errCh <- metrics.ListenAndServe(ctx, 9999)
+	}()
+	
+	// Give the server a moment to start
+	time.Sleep(100 * time.Millisecond)
+	
+	// Test metrics endpoint
+	resp, err := http.Get("http://localhost:9999/metrics")
+	if err != nil {
+		t.Fatalf("failed to GET /metrics: %v", err)
+	}
+	defer resp.Body.Close()
+	
+	if resp.StatusCode != http.StatusOK {
+		t.Errorf("expected status 200, got %d", resp.StatusCode)
+	}
+	
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		t.Fatalf("failed to read response body: %v", err)
+	}
+	
+	bodyStr := string(body)
+	
+	// Check for our test metric
+	if !strings.Contains(bodyStr, "test_requests_total") {
+		t.Error("test metric not found in metrics output")
+	}
+	
+	// Cancel context to stop server
+	cancel()
+	
+	// Wait for server to stop
+	select {
+	case err := <-errCh:
+		if err != nil {
+			t.Errorf("server returned error: %v", err)
+		}
+	case <-time.After(5 * time.Second):
+		t.Error("server did not stop within timeout")
+	}
+}
+
+func TestListenAndServeContextCancellation(t *testing.T) {
+	metrics := New()
+	
+	ctx, cancel := context.WithCancel(context.Background())
+	
+	// Start server
+	errCh := make(chan error, 1)
+	go func() {
+		errCh <- metrics.ListenAndServe(ctx, 9998)
+	}()
+	
+	// Give server time to start
+	time.Sleep(100 * time.Millisecond)
+	
+	// Cancel context
+	cancel()
+	
+	// Server should stop gracefully
+	select {
+	case err := <-errCh:
+		if err != nil {
+			t.Errorf("server returned error on graceful shutdown: %v", err)
+		}
+	case <-time.After(5 * time.Second):
+		t.Error("server did not stop within timeout after context cancellation")
+	}
+}
+
+// Benchmark the metrics handler response time
+func BenchmarkMetricsHandler(b *testing.B) {
+	metrics := New()
+	
+	// Register some test metrics
+	for i := 0; i < 10; i++ {
+		counter := prometheus.NewCounter(prometheus.CounterOpts{
+			Name: fmt.Sprintf("bench_counter_%d", i),
+			Help: "A benchmark counter",
+		})
+		metrics.Registry().MustRegister(counter)
+		counter.Add(float64(i * 100))
+	}
+	
+	handler := metrics.Handler()
+	
+	b.ResetTimer()
+	
+	for i := 0; i < b.N; i++ {
+		req := httptest.NewRequest("GET", "/metrics", nil)
+		recorder := httptest.NewRecorder()
+		handler.ServeHTTP(recorder, req)
+		
+		if recorder.Code != http.StatusOK {
+			b.Fatalf("unexpected status code: %d", recorder.Code)
+		}
+	}
+}

timeutil/duration.go

@@ -1,3 +1,4 @@
+// Package timeutil provides JSON-serializable time utilities.
 package timeutil
 
 import (
@@ -6,14 +7,37 @@ import (
 	"time"
 )
 
+// Duration is a wrapper around time.Duration that supports JSON marshaling/unmarshaling.
+//
+// When marshaling to JSON, it outputs the duration as a string using time.Duration.String().
+// When unmarshaling from JSON, it accepts both:
+//   - String values that can be parsed by time.ParseDuration (e.g., "30s", "5m", "1h30m")
+//   - Numeric values that represent nanoseconds as a float64
+//
+// This makes it compatible with configuration files and APIs that need to represent
+// durations in a human-readable format.
+//
+// Example usage:
+//
+//	type Config struct {
+//		Timeout timeutil.Duration `json:"timeout"`
+//	}
+//
+//	// JSON: {"timeout": "30s"}
+//	// or:   {"timeout": 30000000000}
 type Duration struct {
 	time.Duration
 }
 
+// MarshalJSON implements json.Marshaler.
+// It marshals the duration as a string using time.Duration.String().
 func (d Duration) MarshalJSON() ([]byte, error) {
 	return json.Marshal(time.Duration(d.Duration).String())
 }
 
+// UnmarshalJSON implements json.Unmarshaler.
+// It accepts both string values (parsed via time.ParseDuration) and
+// numeric values (interpreted as nanoseconds).
 func (d *Duration) UnmarshalJSON(b []byte) error {
 	var v any
 	if err := json.Unmarshal(b, &v); err != nil {

ulid/ulid.go

@@ -1,46 +1,44 @@
+// Package ulid provides thread-safe ULID (Universally Unique Lexicographically Sortable Identifier) generation.
+//
+// ULIDs are 128-bit identifiers that are lexicographically sortable and contain
+// a timestamp component. This package uses cryptographically secure random
+// generation optimized for simplicity and performance in concurrent environments.
 package ulid
 
 import (
 	cryptorand "crypto/rand"
-	"encoding/binary"
-	"io"
-	mathrand "math/rand"
-	"os"
-	"sync"
 	"time"
 
 	oklid "github.com/oklog/ulid/v2"
-	"go.ntppool.org/common/logger"
 )
 
-var monotonicPool = sync.Pool{
-	New: func() any {
-		log := logger.Setup()
-
-		var seed int64
-		err := binary.Read(cryptorand.Reader, binary.BigEndian, &seed)
-		if err != nil {
-			log.Error("crypto/rand error", "err", err)
-			os.Exit(10)
-		}
-
-		rand := mathrand.New(mathrand.NewSource(seed))
-
-		inc := uint64(mathrand.Int63())
-
-		// log.Printf("seed: %d", seed)
-		// log.Printf("inc:  %d", inc)
-
-		// inc = inc & ^uint64(1<<63) // only want 63 bits
-		mono := oklid.Monotonic(rand, inc)
-		return mono
-	},
-}
-
+// MakeULID generates a new ULID with the specified timestamp using cryptographically secure randomness.
+// The function is thread-safe and optimized for high-concurrency environments.
+//
+// This implementation prioritizes simplicity and performance over strict monotonicity within
+// the same millisecond. Each ULID is guaranteed to be unique and lexicographically sortable
+// across different timestamps.
+//
+// Returns a pointer to the generated ULID or an error if generation fails.
+// Generation should only fail under extreme circumstances (entropy exhaustion).
 func MakeULID(t time.Time) (*oklid.ULID, error) {
-	mono := monotonicPool.Get().(io.Reader)
-
-	id, err := oklid.New(oklid.Timestamp(t), mono)
+	id, err := oklid.New(oklid.Timestamp(t), cryptorand.Reader)
+	if err != nil {
+		return nil, err
+	}
+
+	return &id, nil
+}
+
+// Make generates a new ULID with the current timestamp using cryptographically secure randomness.
+// This is a convenience function equivalent to MakeULID(time.Now()).
+//
+// The function is thread-safe and optimized for high-concurrency environments.
+//
+// Returns a pointer to the generated ULID or an error if generation fails.
+// Generation should only fail under extreme circumstances (entropy exhaustion).
+func Make() (*oklid.ULID, error) {
+	id, err := oklid.New(oklid.Now(), cryptorand.Reader)
 	if err != nil {
 		return nil, err
 	}

ulid/ulid_test.go

@@ -1,25 +1,336 @@
 package ulid
 
 import (
+	cryptorand "crypto/rand"
+	"sort"
+	"sync"
 	"testing"
 	"time"
+
+	oklid "github.com/oklog/ulid/v2"
 )
 
-func TestULID(t *testing.T) {
+func TestMakeULID(t *testing.T) {
 	tm := time.Now()
 	ul1, err := MakeULID(tm)
 	if err != nil {
-		t.Logf("makeULID failed: %s", err)
-		t.Fail()
+		t.Fatalf("MakeULID failed: %s", err)
 	}
 	ul2, err := MakeULID(tm)
 	if err != nil {
-		t.Logf("MakeULID failed: %s", err)
-		t.Fail()
+		t.Fatalf("MakeULID failed: %s", err)
 	}
+
+	if ul1 == nil || ul2 == nil {
+		t.Fatal("MakeULID returned nil ULID")
+	}
+
 	if ul1.String() == ul2.String() {
-		t.Logf("ul1 and ul2 got the same string: %s", ul1.String())
-		t.Fail()
+		t.Errorf("ul1 and ul2 should be different: %s", ul1.String())
 	}
+
+	// Verify they have the same timestamp
+	if ul1.Time() != ul2.Time() {
+		t.Errorf("ULIDs with same input time should have same timestamp: %d != %d", ul1.Time(), ul2.Time())
+	}
+
 	t.Logf("ulid string 1 and 2: %s | %s", ul1.String(), ul2.String())
 }
+
+func TestMake(t *testing.T) {
+	// Test Make() function (uses current time)
+	ul1, err := Make()
+	if err != nil {
+		t.Fatalf("Make failed: %s", err)
+	}
+
+	if ul1 == nil {
+		t.Fatal("Make returned nil ULID")
+	}
+
+	// Sleep a bit and generate another
+	time.Sleep(2 * time.Millisecond)
+
+	ul2, err := Make()
+	if err != nil {
+		t.Fatalf("Make failed: %s", err)
+	}
+
+	// Should be different ULIDs
+	if ul1.String() == ul2.String() {
+		t.Errorf("ULIDs from Make() should be different: %s", ul1.String())
+	}
+
+	// Second should be later (or at least not earlier)
+	if ul1.Time() > ul2.Time() {
+		t.Errorf("second ULID should not have earlier timestamp: %d > %d", ul1.Time(), ul2.Time())
+	}
+
+	t.Logf("Make() ULIDs: %s | %s", ul1.String(), ul2.String())
+}
+
+func TestMakeULIDUniqueness(t *testing.T) {
+	tm := time.Now()
+	seen := make(map[string]bool)
+
+	for i := 0; i < 1000; i++ {
+		ul, err := MakeULID(tm)
+		if err != nil {
+			t.Fatalf("MakeULID failed on iteration %d: %s", i, err)
+		}
+
+		str := ul.String()
+		if seen[str] {
+			t.Errorf("duplicate ULID generated: %s", str)
+		}
+		seen[str] = true
+	}
+}
+
+func TestMakeUniqueness(t *testing.T) {
+	seen := make(map[string]bool)
+
+	for i := 0; i < 1000; i++ {
+		ul, err := Make()
+		if err != nil {
+			t.Fatalf("Make failed on iteration %d: %s", i, err)
+		}
+
+		str := ul.String()
+		if seen[str] {
+			t.Errorf("duplicate ULID generated: %s", str)
+		}
+		seen[str] = true
+	}
+}
+
+func TestMakeULIDTimestampProgression(t *testing.T) {
+	t1 := time.Now()
+	ul1, err := MakeULID(t1)
+	if err != nil {
+		t.Fatalf("MakeULID failed: %s", err)
+	}
+
+	// Wait to ensure different timestamp
+	time.Sleep(2 * time.Millisecond)
+
+	t2 := time.Now()
+	ul2, err := MakeULID(t2)
+	if err != nil {
+		t.Fatalf("MakeULID failed: %s", err)
+	}
+
+	if ul1.Time() >= ul2.Time() {
+		t.Errorf("second ULID should have later timestamp: %d >= %d", ul1.Time(), ul2.Time())
+	}
+
+	if ul1.Compare(*ul2) >= 0 {
+		t.Errorf("second ULID should be greater: %s >= %s", ul1.String(), ul2.String())
+	}
+}
+
+func TestMakeULIDConcurrency(t *testing.T) {
+	const numGoroutines = 10
+	const numULIDsPerGoroutine = 100
+
+	var wg sync.WaitGroup
+	ulidChan := make(chan *oklid.ULID, numGoroutines*numULIDsPerGoroutine)
+	tm := time.Now()
+
+	// Start multiple goroutines generating ULIDs concurrently
+	for i := 0; i < numGoroutines; i++ {
+		wg.Add(1)
+		go func() {
+			defer wg.Done()
+			for j := 0; j < numULIDsPerGoroutine; j++ {
+				ul, err := MakeULID(tm)
+				if err != nil {
+					t.Errorf("MakeULID failed: %s", err)
+					return
+				}
+				ulidChan <- ul
+			}
+		}()
+	}
+
+	wg.Wait()
+	close(ulidChan)
+
+	// Collect all ULIDs and check uniqueness
+	seen := make(map[string]bool)
+	count := 0
+
+	for ul := range ulidChan {
+		str := ul.String()
+		if seen[str] {
+			t.Errorf("duplicate ULID generated in concurrent test: %s", str)
+		}
+		seen[str] = true
+		count++
+	}
+
+	if count != numGoroutines*numULIDsPerGoroutine {
+		t.Errorf("expected %d ULIDs, got %d", numGoroutines*numULIDsPerGoroutine, count)
+	}
+}
+
+func TestMakeConcurrency(t *testing.T) {
+	const numGoroutines = 10
+	const numULIDsPerGoroutine = 100
+
+	var wg sync.WaitGroup
+	ulidChan := make(chan *oklid.ULID, numGoroutines*numULIDsPerGoroutine)
+
+	// Start multiple goroutines generating ULIDs concurrently
+	for i := 0; i < numGoroutines; i++ {
+		wg.Add(1)
+		go func() {
+			defer wg.Done()
+			for j := 0; j < numULIDsPerGoroutine; j++ {
+				ul, err := Make()
+				if err != nil {
+					t.Errorf("Make failed: %s", err)
+					return
+				}
+				ulidChan <- ul
+			}
+		}()
+	}
+
+	wg.Wait()
+	close(ulidChan)
+
+	// Collect all ULIDs and check uniqueness
+	seen := make(map[string]bool)
+	count := 0
+
+	for ul := range ulidChan {
+		str := ul.String()
+		if seen[str] {
+			t.Errorf("duplicate ULID generated in concurrent test: %s", str)
+		}
+		seen[str] = true
+		count++
+	}
+
+	if count != numGoroutines*numULIDsPerGoroutine {
+		t.Errorf("expected %d ULIDs, got %d", numGoroutines*numULIDsPerGoroutine, count)
+	}
+}
+
+func TestMakeULIDErrorHandling(t *testing.T) {
+	// Test with various timestamps
+	timestamps := []time.Time{
+		time.Unix(0, 0),           // Unix epoch
+		time.Now(),                // Current time
+		time.Now().Add(time.Hour), // Future time
+	}
+
+	for i, tm := range timestamps {
+		ul, err := MakeULID(tm)
+		if err != nil {
+			t.Errorf("MakeULID failed with timestamp %d: %s", i, err)
+		}
+		if ul == nil {
+			t.Errorf("MakeULID returned nil ULID with timestamp %d", i)
+		}
+	}
+}
+
+func TestMakeULIDLexicographicOrdering(t *testing.T) {
+	var ulids []*oklid.ULID
+	var timestamps []time.Time
+
+	// Generate ULIDs with increasing timestamps
+	for i := 0; i < 10; i++ {
+		tm := time.Now().Add(time.Duration(i) * time.Millisecond)
+		timestamps = append(timestamps, tm)
+
+		ul, err := MakeULID(tm)
+		if err != nil {
+			t.Fatalf("MakeULID failed: %s", err)
+		}
+		ulids = append(ulids, ul)
+
+		// Small delay to ensure different timestamps
+		time.Sleep(time.Millisecond)
+	}
+
+	// Sort ULID strings lexicographically
+	ulidStrings := make([]string, len(ulids))
+	for i, ul := range ulids {
+		ulidStrings[i] = ul.String()
+	}
+
+	originalOrder := make([]string, len(ulidStrings))
+	copy(originalOrder, ulidStrings)
+
+	sort.Strings(ulidStrings)
+
+	// Verify lexicographic order matches chronological order
+	for i := 0; i < len(originalOrder); i++ {
+		if originalOrder[i] != ulidStrings[i] {
+			t.Errorf("lexicographic order doesn't match chronological order at index %d: %s != %s",
+				i, originalOrder[i], ulidStrings[i])
+		}
+	}
+}
+
+// Benchmark ULID generation performance
+func BenchmarkMakeULID(b *testing.B) {
+	tm := time.Now()
+
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_, err := MakeULID(tm)
+		if err != nil {
+			b.Fatalf("MakeULID failed: %s", err)
+		}
+	}
+}
+
+// Benchmark Make function
+func BenchmarkMake(b *testing.B) {
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_, err := Make()
+		if err != nil {
+			b.Fatalf("Make failed: %s", err)
+		}
+	}
+}
+
+// Benchmark concurrent ULID generation
+func BenchmarkMakeULIDConcurrent(b *testing.B) {
+	tm := time.Now()
+
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			_, err := MakeULID(tm)
+			if err != nil {
+				b.Fatalf("MakeULID failed: %s", err)
+			}
+		}
+	})
+}
+
+// Benchmark concurrent Make function
+func BenchmarkMakeConcurrent(b *testing.B) {
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			_, err := Make()
+			if err != nil {
+				b.Fatalf("Make failed: %s", err)
+			}
+		}
+	})
+}
+
+// Benchmark random number generation
+func BenchmarkCryptoRand(b *testing.B) {
+	buf := make([]byte, 10) // ULID entropy size
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		cryptorand.Read(buf)
+	}
+}

version/version.go

@@ -1,3 +1,17 @@
+// Package version provides build metadata and version information management.
+//
+// This package manages application version information including semantic version,
+// Git revision, build time, and provides integration with CLI frameworks (Cobra, Kong)
+// and Prometheus metrics for operational visibility.
+//
+// Version information can be injected at build time using ldflags:
+//
+//	go build -ldflags "-X go.ntppool.org/common/version.VERSION=v1.0.0 \
+//	  -X go.ntppool.org/common/version.buildTime=2023-01-01T00:00:00Z \
+//	  -X go.ntppool.org/common/version.gitVersion=abc123"
+//
+// The package also automatically extracts build information from Go's debug.BuildInfo
+// when available, providing fallback values for VCS time and revision.
 package version
 
 import (
@@ -12,21 +26,25 @@ import (
 	"golang.org/x/mod/semver"
 )
 
-// VERSION has the current software version (set in the build process)
+// VERSION contains the current software version (typically set during the build process via ldflags).
+// If not set, defaults to "dev-snapshot". The version should follow semantic versioning.
 var (
-	VERSION     string
-	buildTime   string
-	gitVersion  string
-	gitModified bool
+	VERSION     string // Semantic version (e.g., "1.0.0" or "v1.0.0")
+	buildTime   string // Build timestamp (RFC3339 format)
+	gitVersion  string // Git commit hash
+	gitModified bool   // Whether the working tree was modified during build
 )
 
+// info holds the consolidated version information extracted from build variables and debug.BuildInfo.
 var info Info
 
+// Info represents structured version and build information.
+// This struct is used for JSON serialization and programmatic access to build metadata.
 type Info struct {
-	Version     string `json:",omitempty"`
-	GitRev      string `json:",omitempty"`
-	GitRevShort string `json:",omitempty"`
-	BuildTime   string `json:",omitempty"`
+	Version     string `json:",omitempty"` // Semantic version with "v" prefix
+	GitRev      string `json:",omitempty"` // Full Git commit hash
+	GitRevShort string `json:",omitempty"` // Shortened Git commit hash (7 characters)
+	BuildTime   string `json:",omitempty"` // Build timestamp
 }
 
 func init() {
@@ -79,10 +97,16 @@ func init() {
 	Version()
 }
 
+// VersionCmd creates a Cobra command for displaying version information.
+// The name parameter is used as a prefix in the output (e.g., "myapp v1.0.0").
+// Returns a configured cobra.Command that can be added to any CLI application.
 func VersionCmd(name string) *cobra.Command {
 	versionCmd := &cobra.Command{
 		Use:   "version",
 		Short: "Print version and build information",
+		Long: `Print detailed version information including semantic version,
+Git revision, build time, and Go version. Build information is automatically
+extracted from Go's debug.BuildInfo when available.`,
 		Run: func(cmd *cobra.Command, args []string) {
 			ver := Version()
 			fmt.Printf("%s %s\n", name, ver)
@@ -91,15 +115,23 @@ func VersionCmd(name string) *cobra.Command {
 	return versionCmd
 }
 
+// KongVersionCmd provides a Kong CLI framework compatible version command.
+// The Name field should be set to the application name for proper output formatting.
 type KongVersionCmd struct {
-	Name string `kong:"-"`
+	Name string `kong:"-"` // Application name, excluded from Kong parsing
 }
 
+// Run executes the version command for Kong CLI framework.
+// Prints the application name and version information to stdout.
 func (cmd *KongVersionCmd) Run() error {
 	fmt.Printf("%s %s\n", cmd.Name, Version())
 	return nil
 }
 
+// RegisterMetric registers a Prometheus gauge metric with build information.
+// If name is provided, it creates a metric named "{name}_build_info", otherwise "build_info".
+// The metric includes labels for version, build time, Git time, and Git revision.
+// This is useful for exposing build information in monitoring systems.
 func RegisterMetric(name string, registry prometheus.Registerer) {
 	if len(name) > 0 {
 		name = strings.ReplaceAll(name, "-", "_")
@@ -110,13 +142,13 @@ func RegisterMetric(name string, registry prometheus.Registerer) {
 	buildInfo := prometheus.NewGaugeVec(
 		prometheus.GaugeOpts{
 			Name: name,
-			Help: "Build information",
+			Help: "Build information including version, build time, and git revision",
 		},
 		[]string{
-			"version",
-			"buildtime",
-			"gittime",
-			"git",
+			"version",   // Combined version/git format (e.g., "v1.0.0/abc123")
+			"buildtime", // Build timestamp from ldflags
+			"gittime",   // Git commit timestamp from VCS info
+			"git",       // Full Git commit hash
 		},
 	)
 	registry.MustRegister(buildInfo)
@@ -131,12 +163,20 @@ func RegisterMetric(name string, registry prometheus.Registerer) {
 	).Set(1)
 }
 
+// v caches the formatted version string to avoid repeated computation.
 var v string
 
+// VersionInfo returns the structured version information.
+// This provides programmatic access to version details for JSON serialization
+// or other structured uses.
 func VersionInfo() Info {
 	return info
 }
 
+// Version returns a human-readable version string suitable for display.
+// The format includes semantic version, Git revision, build time, and Go version.
+// Example: "v1.0.0/abc123f-M (2023-01-01T00:00:00Z, go1.21.0)"
+// The "-M" suffix indicates the working tree was modified during build.
 func Version() string {
 	if len(v) > 0 {
 		return v
@@ -164,10 +204,20 @@ func Version() string {
 	return v
 }
 
+// CheckVersion compares a version against a minimum required version.
+// Returns true if the version meets or exceeds the minimum requirement.
+//
+// Special handling:
+//   - "dev-snapshot" is always considered valid (returns true)
+//   - Git hash suffixes (e.g., "v1.0.0/abc123") are stripped before comparison
+//   - Uses semantic version comparison rules
+//
+// Both version and minimumVersion should follow semantic versioning with "v" prefix.
 func CheckVersion(version, minimumVersion string) bool {
 	if version == "dev-snapshot" {
 		return true
 	}
+	// Strip Git hash suffix if present (e.g., "v1.0.0/abc123" -> "v1.0.0")
 	if idx := strings.Index(version, "/"); idx >= 0 {
 		version = version[0:idx]
 	}

version/version_test.go

@@ -0,0 +1,311 @@
+package version
+
+import (
+	"runtime"
+	"strings"
+	"testing"
+
+	"github.com/prometheus/client_golang/prometheus"
+	dto "github.com/prometheus/client_model/go"
+)
+
+func TestCheckVersion(t *testing.T) {
+	tests := []struct {
+		In       string
+		Min      string
+		Expected bool
+	}{
+		// Basic version comparisons
+		{"v3.8.4", "v3.8.5", false},
+		{"v3.9.3", "v3.8.5", true},
+		{"v3.8.5", "v3.8.5", true},
+		// Dev snapshot should always pass
+		{"dev-snapshot", "v3.8.5", true},
+		{"dev-snapshot", "v99.99.99", true},
+		// Versions with Git hashes should be stripped
+		{"v3.8.5/abc123", "v3.8.5", true},
+		{"v3.8.4/abc123", "v3.8.5", false},
+		{"v3.9.0/def456", "v3.8.5", true},
+		// Pre-release versions
+		{"v3.8.5-alpha", "v3.8.5", false},
+		{"v3.8.5", "v3.8.5-alpha", true},
+		{"v3.8.5-beta", "v3.8.5-alpha", true},
+	}
+
+	for _, d := range tests {
+		r := CheckVersion(d.In, d.Min)
+		if r != d.Expected {
+			t.Errorf("CheckVersion(%q, %q) = %t, expected %t", d.In, d.Min, r, d.Expected)
+		}
+	}
+}
+
+func TestVersionInfo(t *testing.T) {
+	info := VersionInfo()
+
+	// Check that we get a valid Info struct
+	if info.Version == "" {
+		t.Error("VersionInfo().Version should not be empty")
+	}
+
+	// Version should start with "v" or be "dev-snapshot"
+	if !strings.HasPrefix(info.Version, "v") && info.Version != "dev-snapshot" {
+		t.Errorf("Version should start with 'v' or be 'dev-snapshot', got: %s", info.Version)
+	}
+
+	// GitRevShort should be <= 7 characters if set
+	if info.GitRevShort != "" && len(info.GitRevShort) > 7 {
+		t.Errorf("GitRevShort should be <= 7 characters, got: %s", info.GitRevShort)
+	}
+
+	// GitRevShort should be prefix of GitRev if both are set
+	if info.GitRev != "" && info.GitRevShort != "" {
+		if !strings.HasPrefix(info.GitRev, info.GitRevShort) {
+			t.Errorf("GitRevShort should be prefix of GitRev: %s not prefix of %s",
+				info.GitRevShort, info.GitRev)
+		}
+	}
+}
+
+func TestVersion(t *testing.T) {
+	version := Version()
+
+	if version == "" {
+		t.Error("Version() should not return empty string")
+	}
+
+	// Should contain Go version
+	if !strings.Contains(version, runtime.Version()) {
+		t.Errorf("Version should contain Go version %s, got: %s", runtime.Version(), version)
+	}
+
+	// Should contain the VERSION variable (or dev-snapshot)
+	info := VersionInfo()
+	if !strings.Contains(version, info.Version) {
+		t.Errorf("Version should contain %s, got: %s", info.Version, version)
+	}
+
+	// Should be in expected format: "version (extras)"
+	if !strings.Contains(version, "(") || !strings.Contains(version, ")") {
+		t.Errorf("Version should be in format 'version (extras)', got: %s", version)
+	}
+}
+
+func TestVersionCmd(t *testing.T) {
+	appName := "testapp"
+	cmd := VersionCmd(appName)
+
+	// Test basic command properties
+	if cmd.Use != "version" {
+		t.Errorf("Expected command use to be 'version', got: %s", cmd.Use)
+	}
+
+	if cmd.Short == "" {
+		t.Error("Command should have a short description")
+	}
+
+	if cmd.Long == "" {
+		t.Error("Command should have a long description")
+	}
+
+	if cmd.Run == nil {
+		t.Error("Command should have a Run function")
+	}
+
+	// Test that the command can be executed without error
+	cmd.SetArgs([]string{})
+	err := cmd.Execute()
+	if err != nil {
+		t.Errorf("VersionCmd execution should not return error, got: %s", err)
+	}
+}
+
+func TestKongVersionCmd(t *testing.T) {
+	cmd := &KongVersionCmd{Name: "testapp"}
+
+	// Test that Run() doesn't return an error
+	err := cmd.Run()
+	if err != nil {
+		t.Errorf("KongVersionCmd.Run() should not return error, got: %s", err)
+	}
+}
+
+func TestRegisterMetric(t *testing.T) {
+	// Create a test registry
+	registry := prometheus.NewRegistry()
+
+	// Test registering metric without name
+	RegisterMetric("", registry)
+
+	// Gather metrics
+	metricFamilies, err := registry.Gather()
+	if err != nil {
+		t.Fatalf("Failed to gather metrics: %s", err)
+	}
+
+	// Find the build_info metric
+	var buildInfoFamily *dto.MetricFamily
+	for _, family := range metricFamilies {
+		if family.GetName() == "build_info" {
+			buildInfoFamily = family
+			break
+		}
+	}
+
+	if buildInfoFamily == nil {
+		t.Fatal("build_info metric not found")
+	}
+
+	if buildInfoFamily.GetHelp() == "" {
+		t.Error("build_info metric should have help text")
+	}
+
+	metrics := buildInfoFamily.GetMetric()
+	if len(metrics) == 0 {
+		t.Fatal("build_info metric should have at least one sample")
+	}
+
+	// Check that the metric has the expected labels
+	metric := metrics[0]
+	labels := metric.GetLabel()
+
+	expectedLabels := []string{"version", "buildtime", "gittime", "git"}
+	labelMap := make(map[string]string)
+
+	for _, label := range labels {
+		labelMap[label.GetName()] = label.GetValue()
+	}
+
+	for _, expectedLabel := range expectedLabels {
+		if _, exists := labelMap[expectedLabel]; !exists {
+			t.Errorf("Expected label %s not found in metric", expectedLabel)
+		}
+	}
+
+	// Check that the metric value is 1
+	if metric.GetGauge().GetValue() != 1 {
+		t.Errorf("Expected build_info metric value to be 1, got %f", metric.GetGauge().GetValue())
+	}
+}
+
+func TestRegisterMetricWithName(t *testing.T) {
+	// Create a test registry
+	registry := prometheus.NewRegistry()
+
+	// Test registering metric with custom name
+	appName := "my-test-app"
+	RegisterMetric(appName, registry)
+
+	// Gather metrics
+	metricFamilies, err := registry.Gather()
+	if err != nil {
+		t.Fatalf("Failed to gather metrics: %s", err)
+	}
+
+	// Find the my_test_app_build_info metric
+	expectedName := "my_test_app_build_info"
+	var buildInfoFamily *dto.MetricFamily
+	for _, family := range metricFamilies {
+		if family.GetName() == expectedName {
+			buildInfoFamily = family
+			break
+		}
+	}
+
+	if buildInfoFamily == nil {
+		t.Fatalf("%s metric not found", expectedName)
+	}
+}
+
+func TestVersionConsistency(t *testing.T) {
+	// Call Version() multiple times and ensure it returns the same result
+	v1 := Version()
+	v2 := Version()
+
+	if v1 != v2 {
+		t.Errorf("Version() should return consistent results: %s != %s", v1, v2)
+	}
+}
+
+func TestVersionInfoConsistency(t *testing.T) {
+	// Ensure VersionInfo() is consistent with Version()
+	info := VersionInfo()
+	version := Version()
+
+	// Version string should contain the semantic version
+	if !strings.Contains(version, info.Version) {
+		t.Errorf("Version() should contain VersionInfo().Version: %s not in %s",
+			info.Version, version)
+	}
+
+	// If GitRevShort is set, version should contain it
+	if info.GitRevShort != "" {
+		if !strings.Contains(version, info.GitRevShort) {
+			t.Errorf("Version() should contain GitRevShort: %s not in %s",
+				info.GitRevShort, version)
+		}
+	}
+}
+
+// Test edge cases
+func TestCheckVersionEdgeCases(t *testing.T) {
+	// Test with empty strings
+	if CheckVersion("", "v1.0.0") {
+		t.Error("Empty version should not be >= v1.0.0")
+	}
+
+	// Test with malformed versions (should be handled gracefully)
+	// Note: semver.Compare might panic or return unexpected results for invalid versions
+	// but our function should handle the common cases
+	tests := []struct {
+		version string
+		minimum string
+		desc    string
+	}{
+		{"v1.0.0/", "v1.0.0", "version with trailing slash"},
+		{"v1.0.0/abc/def", "v1.0.0", "version with multiple slashes"},
+	}
+
+	for _, test := range tests {
+		// This should not panic
+		result := CheckVersion(test.version, test.minimum)
+		t.Logf("%s: CheckVersion(%q, %q) = %t", test.desc, test.version, test.minimum, result)
+	}
+}
+
+// Benchmark version operations
+func BenchmarkVersion(b *testing.B) {
+	// Reset the cached version to test actual computation
+	v = ""
+
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_ = Version()
+	}
+}
+
+func BenchmarkVersionInfo(b *testing.B) {
+	for i := 0; i < b.N; i++ {
+		_ = VersionInfo()
+	}
+}
+
+func BenchmarkCheckVersion(b *testing.B) {
+	version := "v1.2.3/abc123"
+	minimum := "v1.2.0"
+
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_ = CheckVersion(version, minimum)
+	}
+}
+
+func BenchmarkCheckVersionDevSnapshot(b *testing.B) {
+	version := "dev-snapshot"
+	minimum := "v1.2.0"
+
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		_ = CheckVersion(version, minimum)
+	}
+}