feat(metrics): basic auth support (#1579)

* feat(internal): add basic auth HTTP middleware

Signed-off-by: Xe Iaso <me@xeiaso.net>

* feat(config): add HTTP basic auth for metrics

Signed-off-by: Xe Iaso <me@xeiaso.net>

* feat(metrics): wire up basic auth

Signed-off-by: Xe Iaso <me@xeiaso.net>

* doc: document HTTP basic auth for metrics server

Signed-off-by: Xe Iaso <me@xeiaso.net>

* chore: spelling

Signed-off-by: Xe Iaso <me@xeiaso.net>

* docs(admin/policies): give people a python command

Signed-off-by: Xe Iaso <me@xeiaso.net>

---------

Signed-off-by: Xe Iaso <me@xeiaso.net>

.github/actions/spelling/allow.txt

@@ -38,3 +38,4 @@ Samsung
 wenet
 qwertiko
 setuplistener
+mba

.github/actions/spelling/expect.txt

@@ -47,6 +47,7 @@ cachediptoasn
 Caddyfile
 caninetools
 Cardyb
+CAs
 celchecker
 celphase
 cerr
@@ -203,8 +204,10 @@ kagi
 kagibot
 Keyfunc
 keypair
+keypairreloader
 KHTML
 kinda
+kpr
 KUBECONFIG
 lcj
 ldflags
@@ -229,6 +232,7 @@ metarefresh
 metrix
 mimi
 Minfilia
+minica
 mistralai
 mnt
 Mojeek
@@ -313,6 +317,7 @@ searchbot
 searx
 sebest
 secretplans
+selfsigned
 Semrush
 Seo
 setsebool

data/botPolicies.yaml

@@ -174,6 +174,27 @@ status_codes:
 # metrics:
 #   bind: ":9090"
 #   network: "tcp"
+#
+# # To protect your metrics server with basic auth, set credentials below:
+# #
+# # https://anubis.techaro.lol/docs/admin/policies#http-basic-authentication
+#   basicAuth:
+#     username: ""
+#     password: ""
+#
+# # To serve metrics over TLS, set the path to the right TLS certificate and key
+# # here. When the files change on disk, they will automatically be reloaded.
+# #
+# # https://anubis.techaro.lol/docs/admin/policies#tls
+#   tls:
+#     certificate: /path/to/tls.crt
+#     key: /path/to/tls.key
+#
+#     # If you want to secure your metrics endpoint using mutual TLS (mTLS), set
+#     # the path to a certificate authority public certificate here.
+#     #
+#     # https://anubis.techaro.lol/docs/admin/policies#mtls
+#     ca: /path/to/ca.crt
 
 # Anubis can store temporary data in one of a few backends. See the storage
 # backends section of the docs for more information:

docs/docs/CHANGELOG.md

@@ -21,6 +21,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Improve error messages and fix broken REDIRECT_DOMAINS link in docs ([#1193](https://github.com/TecharoHQ/anubis/issues/1193))
 - Add Bulgarian locale ([#1394](https://github.com/TecharoHQ/anubis/pull/1394))
 - Fix CEL internal errors when iterating `headers`/`query` map wrappers by implementing map iterators for `HTTPHeaders` and `URLValues` ([#1465](https://github.com/TecharoHQ/anubis/pull/1465)).
+- Enable [metrics serving via TLS](./admin/policies.mdx#tls), including [mutual TLS (mTLS)](./admin/policies.mdx#mtls).
+- Enable [HTTP basic auth](./admin/policies.mdx#http-basic-authentication) for the metrics server.
 
 ## v1.25.0: Necron
 

docs/docs/admin/policies.mdx

@@ -138,6 +138,57 @@ metrics:
   socketMode: "0700" # must be a string
 ```
 
+### TLS
+
+If you want to serve the metrics server over TLS, use the `tls` block:
+
+```yaml
+metrics:
+  bind: ":9090"
+  network: "tcp"
+
+  tls:
+    certificate: /path/to/tls.crt
+    key: /path/to/tls.key
+```
+
+The certificate and key will automatically be reloaded when the respective files change.
+
+### mTLS
+
+If you want to validate requests to ensure that they use a client certificate signed by a certificate authority (mutual TLS or mTLS), set the `ca` value in the `tls` block:
+
+```yaml
+metrics:
+  bind: ":9090"
+  network: "tcp"
+
+  tls:
+    certificate: /path/to/tls.crt
+    key: /path/to/tls.key
+    ca: /path/to/ca.crt
+```
+
+As it is not expected for certificate authority certificates to change often, the CA certificate will NOT be automatically reloaded when the respective file changes.
+
+### HTTP basic authentication
+
+Anubis' metrics server also supports setting HTTP basic auth as a lightweight protection against unauthorized users viewing metrics data. As the basic auth credentials are hardcoded in the configuration file, administrators SHOULD use randomly generated credentials, such as type-4 UUIDs or other high entropy strings. These credentials MUST NOT be sensitive or used to protect other high value systems.
+
+Configure it with the `basicAuth` block under `metrics`:
+
+```yaml
+metrics:
+  bind: ":9090"
+  network: "tcp"
+
+  basicAuth:
+    username: azurediamond
+    password: hunter2
+```
+
+If you have Python installed, you can generate a high entropy password with `python -c 'import secrets; print(secrets.token_urlsafe(32))'`.
+
 ## Imprint / Impressum support
 
 Anubis has support for showing imprint / impressum information. This is defined in the `impressum` block of your configuration. See [Imprint / Impressum configuration](./configuration/impressum.mdx) for more information.

internal/basicauth.go

@@ -0,0 +1,52 @@
+package internal
+
+import (
+	"crypto/sha256"
+	"crypto/subtle"
+	"fmt"
+	"log/slog"
+	"net/http"
+)
+
+// BasicAuth wraps next in HTTP Basic authentication using the provided
+// credentials. If either username or password is empty, next is returned
+// unchanged and a debug log line is emitted.
+//
+// Credentials are compared in constant time to avoid leaking information
+// through timing side channels.
+func BasicAuth(realm, username, password string, next http.Handler) http.Handler {
+	if username == "" || password == "" {
+		slog.Debug("skipping middleware, basic auth credentials are empty")
+		return next
+	}
+
+	expectedUser := sha256.Sum256([]byte(username))
+	expectedPass := sha256.Sum256([]byte(password))
+	challenge := fmt.Sprintf("Basic realm=%q, charset=\"UTF-8\"", realm)
+
+	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		user, pass, ok := r.BasicAuth()
+		if !ok {
+			unauthorized(w, challenge)
+			return
+		}
+
+		gotUser := sha256.Sum256([]byte(user))
+		gotPass := sha256.Sum256([]byte(pass))
+
+		userMatch := subtle.ConstantTimeCompare(gotUser[:], expectedUser[:])
+		passMatch := subtle.ConstantTimeCompare(gotPass[:], expectedPass[:])
+
+		if userMatch&passMatch != 1 {
+			unauthorized(w, challenge)
+			return
+		}
+
+		next.ServeHTTP(w, r)
+	})
+}
+
+func unauthorized(w http.ResponseWriter, challenge string) {
+	w.Header().Set("WWW-Authenticate", challenge)
+	http.Error(w, "Unauthorized", http.StatusUnauthorized)
+}

internal/basicauth_test.go

@@ -0,0 +1,138 @@
+package internal
+
+import (
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+)
+
+func okHandler() http.Handler {
+	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.WriteHeader(http.StatusOK)
+		_, _ = w.Write([]byte("ok"))
+	})
+}
+
+func TestBasicAuth(t *testing.T) {
+	t.Parallel()
+
+	const (
+		realm    = "test-realm"
+		username = "admin"
+		password = "hunter2"
+	)
+
+	for _, tt := range []struct {
+		name       string
+		setAuth    bool
+		user       string
+		pass       string
+		wantStatus int
+		wantBody   string
+		wantChall  bool
+	}{
+		{
+			name:       "valid credentials",
+			setAuth:    true,
+			user:       username,
+			pass:       password,
+			wantStatus: http.StatusOK,
+			wantBody:   "ok",
+		},
+		{
+			name:       "missing credentials",
+			setAuth:    false,
+			wantStatus: http.StatusUnauthorized,
+			wantChall:  true,
+		},
+		{
+			name:       "wrong username",
+			setAuth:    true,
+			user:       "nobody",
+			pass:       password,
+			wantStatus: http.StatusUnauthorized,
+			wantChall:  true,
+		},
+		{
+			name:       "wrong password",
+			setAuth:    true,
+			user:       username,
+			pass:       "wrong",
+			wantStatus: http.StatusUnauthorized,
+			wantChall:  true,
+		},
+		{
+			name:       "empty supplied credentials",
+			setAuth:    true,
+			user:       "",
+			pass:       "",
+			wantStatus: http.StatusUnauthorized,
+			wantChall:  true,
+		},
+	} {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+
+			h := BasicAuth(realm, username, password, okHandler())
+
+			req := httptest.NewRequest(http.MethodGet, "/", nil)
+			if tt.setAuth {
+				req.SetBasicAuth(tt.user, tt.pass)
+			}
+			rec := httptest.NewRecorder()
+			h.ServeHTTP(rec, req)
+
+			if rec.Code != tt.wantStatus {
+				t.Errorf("status = %d, want %d", rec.Code, tt.wantStatus)
+			}
+
+			if tt.wantBody != "" && rec.Body.String() != tt.wantBody {
+				t.Errorf("body = %q, want %q", rec.Body.String(), tt.wantBody)
+			}
+
+			chall := rec.Header().Get("WWW-Authenticate")
+			if tt.wantChall {
+				if chall == "" {
+					t.Error("WWW-Authenticate header missing on 401")
+				}
+				if !strings.Contains(chall, realm) {
+					t.Errorf("WWW-Authenticate = %q, want realm %q", chall, realm)
+				}
+			} else if chall != "" {
+				t.Errorf("unexpected WWW-Authenticate header: %q", chall)
+			}
+		})
+	}
+}
+
+func TestBasicAuthPassthrough(t *testing.T) {
+	t.Parallel()
+
+	for _, tt := range []struct {
+		name     string
+		username string
+		password string
+	}{
+		{name: "empty username", username: "", password: "hunter2"},
+		{name: "empty password", username: "admin", password: ""},
+		{name: "both empty", username: "", password: ""},
+	} {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+
+			h := BasicAuth("realm", tt.username, tt.password, okHandler())
+
+			req := httptest.NewRequest(http.MethodGet, "/", nil)
+			rec := httptest.NewRecorder()
+			h.ServeHTTP(rec, req)
+
+			if rec.Code != http.StatusOK {
+				t.Errorf("status = %d, want %d (passthrough expected)", rec.Code, http.StatusOK)
+			}
+			if rec.Body.String() != "ok" {
+				t.Errorf("body = %q, want %q", rec.Body.String(), "ok")
+			}
+		})
+	}
+}

lib/config/metrics.go

@@ -1,24 +1,38 @@
 package config
 
 import (
+	"crypto/tls"
+	"crypto/x509"
 	"errors"
 	"fmt"
+	"os"
 	"strconv"
 )
 
 var (
 	ErrInvalidMetricsConfig          = errors.New("config: invalid metrics configuration")
+	ErrInvalidMetricsTLSConfig       = errors.New("config: invalid metrics TLS configuration")
+	ErrInvalidMetricsBasicAuthConfig = errors.New("config: invalid metrics basic auth configuration")
 	ErrNoMetricsBind                 = errors.New("config.Metrics: must define bind")
 	ErrNoMetricsNetwork              = errors.New("config.Metrics: must define network")
 	ErrNoMetricsSocketMode           = errors.New("config.Metrics: must define socket mode when using unix sockets")
 	ErrInvalidMetricsSocketMode      = errors.New("config.Metrics: invalid unix socket mode")
 	ErrInvalidMetricsNetwork         = errors.New("config.Metrics: invalid metrics network")
+	ErrNoMetricsTLSCertificate       = errors.New("config.Metrics.TLS: must define certificate file")
+	ErrNoMetricsTLSKey               = errors.New("config.Metrics.TLS: must define key file")
+	ErrInvalidMetricsTLSKeypair      = errors.New("config.Metrics.TLS: keypair is invalid")
+	ErrInvalidMetricsCACertificate   = errors.New("config.Metrics.TLS: invalid CA certificate")
+	ErrCantReadFile                  = errors.New("config: can't read required file")
+	ErrNoMetricsBasicAuthUsername    = errors.New("config.Metrics.BasicAuth: must define username")
+	ErrNoMetricsBasicAuthPassword    = errors.New("config.Metrics.BasicAuth: must define password")
 )
 
 type Metrics struct {
 	Bind       string            `json:"bind" yaml:"bind"`
 	Network    string            `json:"network" yaml:"network"`
 	SocketMode string            `json:"socketMode" yaml:"socketMode"`
+	TLS        *MetricsTLS       `json:"tls" yaml:"tls"`
+	BasicAuth  *MetricsBasicAuth `json:"basicAuth" yaml:"basicAuth"`
 }
 
 func (m *Metrics) Valid() error {
@@ -46,9 +60,107 @@ func (m *Metrics) Valid() error {
 		errs = append(errs, ErrInvalidMetricsNetwork)
 	}
 
+	if m.TLS != nil {
+		if err := m.TLS.Valid(); err != nil {
+			errs = append(errs, err)
+		}
+	}
+
+	if m.BasicAuth != nil {
+		if err := m.BasicAuth.Valid(); err != nil {
+			errs = append(errs, err)
+		}
+	}
+
 	if len(errs) != 0 {
 		return errors.Join(ErrInvalidMetricsConfig, errors.Join(errs...))
 	}
 
 	return nil
 }
+
+type MetricsTLS struct {
+	Certificate string `json:"certificate" yaml:"certificate"`
+	Key         string `json:"key" yaml:"key"`
+	CA          string `json:"ca" yaml:"ca"`
+}
+
+func (mt *MetricsTLS) Valid() error {
+	var errs []error
+
+	if mt.Certificate == "" {
+		errs = append(errs, ErrNoMetricsTLSCertificate)
+	}
+
+	if err := canReadFile(mt.Certificate); err != nil {
+		errs = append(errs, fmt.Errorf("%w %s: %w", ErrCantReadFile, mt.Certificate, err))
+	}
+
+	if mt.Key == "" {
+		errs = append(errs, ErrNoMetricsTLSKey)
+	}
+
+	if err := canReadFile(mt.Key); err != nil {
+		errs = append(errs, fmt.Errorf("%w %s: %w", ErrCantReadFile, mt.Key, err))
+	}
+
+	if _, err := tls.LoadX509KeyPair(mt.Certificate, mt.Key); err != nil {
+		errs = append(errs, fmt.Errorf("%w: %w", ErrInvalidMetricsTLSKeypair, err))
+	}
+
+	if mt.CA != "" {
+		caCert, err := os.ReadFile(mt.CA)
+		if err != nil {
+			errs = append(errs, fmt.Errorf("%w %s: %w", ErrCantReadFile, mt.CA, err))
+		}
+
+		certPool := x509.NewCertPool()
+		if !certPool.AppendCertsFromPEM(caCert) {
+			errs = append(errs, fmt.Errorf("%w %s", ErrInvalidMetricsCACertificate, mt.CA))
+		}
+	}
+
+	if len(errs) != 0 {
+		return errors.Join(ErrInvalidMetricsTLSConfig, errors.Join(errs...))
+	}
+
+	return nil
+}
+
+func canReadFile(fname string) error {
+	fin, err := os.Open(fname)
+	if err != nil {
+		return err
+	}
+	defer fin.Close()
+
+	data := make([]byte, 64)
+	if _, err := fin.Read(data); err != nil {
+		return fmt.Errorf("can't read %s: %w", fname, err)
+	}
+
+	return nil
+}
+
+type MetricsBasicAuth struct {
+	Username string `json:"username" yaml:"username"`
+	Password string `json:"password" yaml:"password"`
+}
+
+func (mba *MetricsBasicAuth) Valid() error {
+	var errs []error
+
+	if mba.Username == "" {
+		errs = append(errs, ErrNoMetricsBasicAuthUsername)
+	}
+
+	if mba.Password == "" {
+		errs = append(errs, ErrNoMetricsBasicAuthPassword)
+	}
+
+	if len(errs) != 0 {
+		return errors.Join(ErrInvalidMetricsBasicAuthConfig, errors.Join(errs...))
+	}
+
+	return nil
+}

lib/config/metrics_test.go

@@ -75,6 +75,161 @@ func TestMetricsValid(t *testing.T) {
 			},
 			err: ErrInvalidMetricsNetwork,
 		},
+		{
+			name: "invalid TLS config",
+			input: &Metrics{
+				Bind:    ":9090",
+				Network: "tcp",
+				TLS:     &MetricsTLS{},
+			},
+			err: ErrInvalidMetricsTLSConfig,
+		},
+		{
+			name: "selfsigned TLS cert",
+			input: &Metrics{
+				Bind:    ":9090",
+				Network: "tcp",
+				TLS: &MetricsTLS{
+					Certificate: "./testdata/tls/selfsigned.crt",
+					Key:         "./testdata/tls/selfsigned.key",
+				},
+			},
+		},
+		{
+			name: "wrong path to selfsigned TLS cert",
+			input: &Metrics{
+				Bind:    ":9090",
+				Network: "tcp",
+				TLS: &MetricsTLS{
+					Certificate: "./testdata/tls2/selfsigned.crt",
+					Key:         "./testdata/tls2/selfsigned.key",
+				},
+			},
+			err: ErrCantReadFile,
+		},
+		{
+			name: "unparseable TLS cert",
+			input: &Metrics{
+				Bind:    ":9090",
+				Network: "tcp",
+				TLS: &MetricsTLS{
+					Certificate: "./testdata/tls/invalid.crt",
+					Key:         "./testdata/tls/invalid.key",
+				},
+			},
+			err: ErrInvalidMetricsTLSKeypair,
+		},
+		{
+			name: "mTLS with CA",
+			input: &Metrics{
+				Bind:    ":9090",
+				Network: "tcp",
+				TLS: &MetricsTLS{
+					Certificate: "./testdata/tls/selfsigned.crt",
+					Key:         "./testdata/tls/selfsigned.key",
+					CA:          "./testdata/tls/minica.pem",
+				},
+			},
+		},
+		{
+			name: "mTLS with nonexistent CA",
+			input: &Metrics{
+				Bind:    ":9090",
+				Network: "tcp",
+				TLS: &MetricsTLS{
+					Certificate: "./testdata/tls/selfsigned.crt",
+					Key:         "./testdata/tls/selfsigned.key",
+					CA:          "./testdata/tls/nonexistent.crt",
+				},
+			},
+			err: ErrCantReadFile,
+		},
+		{
+			name: "mTLS with invalid CA",
+			input: &Metrics{
+				Bind:    ":9090",
+				Network: "tcp",
+				TLS: &MetricsTLS{
+					Certificate: "./testdata/tls/selfsigned.crt",
+					Key:         "./testdata/tls/selfsigned.key",
+					CA:          "./testdata/tls/invalid.crt",
+				},
+			},
+			err: ErrInvalidMetricsCACertificate,
+		},
+		{
+			name: "basic auth credentials set",
+			input: &Metrics{
+				Bind:    ":9090",
+				Network: "tcp",
+				BasicAuth: &MetricsBasicAuth{
+					Username: "admin",
+					Password: "hunter2",
+				},
+			},
+		},
+		{
+			name: "invalid basic auth config",
+			input: &Metrics{
+				Bind:      ":9090",
+				Network:   "tcp",
+				BasicAuth: &MetricsBasicAuth{},
+			},
+			err: ErrInvalidMetricsBasicAuthConfig,
+		},
+	} {
+		t.Run(tt.name, func(t *testing.T) {
+			if err := tt.input.Valid(); !errors.Is(err, tt.err) {
+				t.Logf("wanted error: %v", tt.err)
+				t.Logf("got error:    %v", err)
+				t.Error("validation failed")
+			}
+		})
+	}
+}
+
+func TestMetricsBasicAuthValid(t *testing.T) {
+	for _, tt := range []struct {
+		name  string
+		input *MetricsBasicAuth
+		err   error
+	}{
+		{
+			name: "both set",
+			input: &MetricsBasicAuth{
+				Username: "admin",
+				Password: "hunter2",
+			},
+		},
+		{
+			name:  "empty username and password",
+			input: &MetricsBasicAuth{},
+			err:   ErrInvalidMetricsBasicAuthConfig,
+		},
+		{
+			name: "missing username",
+			input: &MetricsBasicAuth{
+				Password: "hunter2",
+			},
+			err: ErrNoMetricsBasicAuthUsername,
+		},
+		{
+			name: "missing password",
+			input: &MetricsBasicAuth{
+				Username: "admin",
+			},
+			err: ErrNoMetricsBasicAuthPassword,
+		},
+		{
+			name:  "missing both surfaces wrapper error",
+			input: &MetricsBasicAuth{},
+			err:   ErrNoMetricsBasicAuthUsername,
+		},
+		{
+			name:  "missing both surfaces password error",
+			input: &MetricsBasicAuth{},
+			err:   ErrNoMetricsBasicAuthPassword,
+		},
 	} {
 		t.Run(tt.name, func(t *testing.T) {
 			if err := tt.input.Valid(); !errors.Is(err, tt.err) {

lib/config/testdata/tls/1.1.1.1/cert.pem

@@ -0,0 +1,12 @@
+-----BEGIN CERTIFICATE-----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+-----END CERTIFICATE-----

lib/config/testdata/tls/1.1.1.1/key.pem

@@ -0,0 +1,6 @@
+-----BEGIN PRIVATE KEY-----
+MIG2AgEAMBAGByqGSM49AgEGBSuBBAAiBIGeMIGbAgEBBDBN8QsHxxHGJpStu8K7
+D/FmaBBNo6c514KGFSIfqGFuREF5aOL3gN/W11yk2OIibdWhZANiAAS7LgNiyhmx
+AbkgOC05tSminO/0ApFDrIqV0ePg+f97FgWYgB+4llIB8KceWGX9qpGZ+qEXWvpL
+NCS00OlYbXkHnZm+5fOmzHgzEfKeC4Nm9/Jf8RK+0tuVH1AyrnV6zgI=
+-----END PRIVATE KEY-----

lib/config/testdata/tls/minica-key.pem

@@ -0,0 +1,6 @@
+-----BEGIN PRIVATE KEY-----
+MIG2AgEAMBAGByqGSM49AgEGBSuBBAAiBIGeMIGbAgEBBDDr9QQo7ZaTgUL6d73G
+2BG7+YRTFJHAZa0FogRglfc+jYttL1J4/xTig3RmHoqSgrehZANiAASDhijM9Xe0
+G9Vam6AJMeKC6aWDNSLwrxNVmPxemsY/yJ1urBgnxRd9GFH6YW1ki/B8rS+Xl1UX
+NnhBrukLaXvgAQQq782/5IUYGsvK5jw8+dSscYVMCQJwGfmQuaNeczQ=
+-----END PRIVATE KEY-----

lib/config/testdata/tls/minica.pem

@@ -0,0 +1,13 @@
+-----BEGIN CERTIFICATE-----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+-----END CERTIFICATE-----

lib/config/testdata/tls/selfsigned.crt

@@ -0,0 +1,11 @@
+-----BEGIN CERTIFICATE-----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+-----END CERTIFICATE-----

lib/config/testdata/tls/selfsigned.key

@@ -0,0 +1,3 @@
+-----BEGIN PRIVATE KEY-----
+MC4CAQAwBQYDK2VwBCIEIL0HxjjfVlg6zQPB9/zTLq0IBzfp8gEoifEYzQZYIj+T
+-----END PRIVATE KEY-----

lib/metrics/keypairreloader.go

@@ -0,0 +1,81 @@
+package metrics
+
+import (
+	"crypto/tls"
+	"fmt"
+	"log/slog"
+	"os"
+	"sync"
+	"time"
+)
+
+type KeypairReloader struct {
+	certMu   sync.RWMutex
+	cert     *tls.Certificate
+	certPath string
+	keyPath  string
+	modTime  time.Time
+	lg       *slog.Logger
+}
+
+func NewKeypairReloader(certPath, keyPath string, lg *slog.Logger) (*KeypairReloader, error) {
+	result := &KeypairReloader{
+		certPath: certPath,
+		keyPath:  keyPath,
+		lg:       lg,
+	}
+	cert, err := tls.LoadX509KeyPair(certPath, keyPath)
+	if err != nil {
+		return nil, err
+	}
+	result.cert = &cert
+
+	st, err := os.Stat(certPath)
+	if err != nil {
+		return nil, err
+	}
+	result.modTime = st.ModTime()
+
+	return result, nil
+}
+
+func (kpr *KeypairReloader) maybeReload() error {
+	kpr.lg.Debug("loading new keypair", "cert", kpr.certPath, "key", kpr.keyPath)
+	newCert, err := tls.LoadX509KeyPair(kpr.certPath, kpr.keyPath)
+	if err != nil {
+		return err
+	}
+
+	st, err := os.Stat(kpr.certPath)
+	if err != nil {
+		return err
+	}
+
+	kpr.certMu.Lock()
+	defer kpr.certMu.Unlock()
+	kpr.cert = &newCert
+	kpr.modTime = st.ModTime()
+
+	return nil
+}
+
+func (kpr *KeypairReloader) GetCertificate(*tls.ClientHelloInfo) (*tls.Certificate, error) {
+	st, err := os.Stat(kpr.certPath)
+	if err != nil {
+		return nil, fmt.Errorf("stat(%q): %w", kpr.certPath, err)
+	}
+
+	kpr.certMu.RLock()
+	needsReload := st.ModTime().After(kpr.modTime)
+	kpr.certMu.RUnlock()
+
+	if needsReload {
+		if err := kpr.maybeReload(); err != nil {
+			return nil, fmt.Errorf("reload cert: %w", err)
+		}
+	}
+
+	kpr.certMu.RLock()
+	defer kpr.certMu.RUnlock()
+	return kpr.cert, nil
+}

lib/metrics/keypairreloader_test.go

@@ -0,0 +1,265 @@
+package metrics
+
+import (
+	"bytes"
+	"crypto/rand"
+	"crypto/rsa"
+	"crypto/x509"
+	"crypto/x509/pkix"
+	"encoding/pem"
+	"errors"
+	"log/slog"
+	"math/big"
+	"os"
+	"path/filepath"
+	"testing"
+	"time"
+)
+
+func discardLogger() *slog.Logger {
+	return slog.New(slog.DiscardHandler)
+}
+
+// writeKeypair generates a fresh self-signed cert + RSA key and writes them
+// as PEM files in dir. Returns the paths and the cert's DER bytes so callers
+// can identify which pair was loaded.
+func writeKeypair(t *testing.T, dir, prefix string) (certPath, keyPath string, certDER []byte) {
+	t.Helper()
+
+	priv, err := rsa.GenerateKey(rand.Reader, 2048)
+	if err != nil {
+		t.Fatalf("rsa.GenerateKey: %v", err)
+	}
+
+	tmpl := &x509.Certificate{
+		SerialNumber:          big.NewInt(time.Now().UnixNano()),
+		Subject:               pkix.Name{CommonName: "keypairreloader-test"},
+		NotBefore:             time.Now().Add(-time.Hour),
+		NotAfter:              time.Now().Add(time.Hour),
+		KeyUsage:              x509.KeyUsageDigitalSignature | x509.KeyUsageKeyEncipherment,
+		ExtKeyUsage:           []x509.ExtKeyUsage{x509.ExtKeyUsageServerAuth},
+		BasicConstraintsValid: true,
+		DNSNames:              []string{"keypairreloader-test"},
+	}
+
+	der, err := x509.CreateCertificate(rand.Reader, tmpl, tmpl, &priv.PublicKey, priv)
+	if err != nil {
+		t.Fatalf("x509.CreateCertificate: %v", err)
+	}
+
+	certPath = filepath.Join(dir, prefix+"cert.pem")
+	keyPath = filepath.Join(dir, prefix+"key.pem")
+
+	certPEM := pem.EncodeToMemory(&pem.Block{Type: "CERTIFICATE", Bytes: der})
+	keyPEM := pem.EncodeToMemory(&pem.Block{Type: "RSA PRIVATE KEY", Bytes: x509.MarshalPKCS1PrivateKey(priv)})
+
+	if err := os.WriteFile(certPath, certPEM, 0o600); err != nil {
+		t.Fatalf("write cert: %v", err)
+	}
+	if err := os.WriteFile(keyPath, keyPEM, 0o600); err != nil {
+		t.Fatalf("write key: %v", err)
+	}
+
+	return certPath, keyPath, der
+}
+
+func TestNewKeypairReloader(t *testing.T) {
+	dir := t.TempDir()
+	goodCert, goodKey, _ := writeKeypair(t, dir, "good-")
+
+	garbagePath := filepath.Join(dir, "garbage.pem")
+	if err := os.WriteFile(garbagePath, []byte("not a pem file"), 0o600); err != nil {
+		t.Fatalf("write garbage: %v", err)
+	}
+
+	tests := []struct {
+		name     string
+		certPath string
+		keyPath  string
+		wantErr  error
+		wantNil  bool
+	}{
+		{
+			name:     "valid cert and key",
+			certPath: goodCert,
+			keyPath:  goodKey,
+		},
+		{
+			name:     "missing cert file",
+			certPath: filepath.Join(dir, "does-not-exist.pem"),
+			keyPath:  goodKey,
+			wantErr:  os.ErrNotExist,
+			wantNil:  true,
+		},
+		{
+			name:     "missing key file",
+			certPath: goodCert,
+			keyPath:  filepath.Join(dir, "does-not-exist-key.pem"),
+			wantErr:  os.ErrNotExist,
+			wantNil:  true,
+		},
+		{
+			name:     "cert file is garbage",
+			certPath: garbagePath,
+			keyPath:  goodKey,
+			wantNil:  true,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			kpr, err := NewKeypairReloader(tt.certPath, tt.keyPath, discardLogger())
+
+			if tt.wantErr != nil && !errors.Is(err, tt.wantErr) {
+				t.Errorf("err = %v, want errors.Is(..., %v)", err, tt.wantErr)
+			}
+			if tt.wantErr == nil && !tt.wantNil && err != nil {
+				t.Errorf("unexpected err: %v", err)
+			}
+			if tt.wantNil && kpr != nil {
+				t.Errorf("kpr = %+v, want nil", kpr)
+			}
+			if !tt.wantNil && kpr == nil {
+				t.Errorf("kpr is nil, want non-nil")
+			}
+		})
+	}
+}
+
+func TestKeypairReloader_GetCertificate(t *testing.T) {
+	tests := []struct {
+		name string
+		run  func(t *testing.T)
+	}{
+		{
+			name: "returns loaded cert",
+			run: func(t *testing.T) {
+				dir := t.TempDir()
+				certPath, keyPath, wantDER := writeKeypair(t, dir, "a-")
+
+				kpr, err := NewKeypairReloader(certPath, keyPath, discardLogger())
+				if err != nil {
+					t.Fatalf("NewKeypairReloader: %v", err)
+				}
+
+				got, err := kpr.GetCertificate(nil)
+				if err != nil {
+					t.Fatalf("GetCertificate: %v", err)
+				}
+				if len(got.Certificate) == 0 || !bytes.Equal(got.Certificate[0], wantDER) {
+					t.Errorf("GetCertificate returned wrong cert bytes")
+				}
+			},
+		},
+		{
+			name: "reloads when mtime advances",
+			run: func(t *testing.T) {
+				dir := t.TempDir()
+				certPath, keyPath, _ := writeKeypair(t, dir, "a-")
+
+				kpr, err := NewKeypairReloader(certPath, keyPath, discardLogger())
+				if err != nil {
+					t.Fatalf("NewKeypairReloader: %v", err)
+				}
+
+				// Overwrite with a new pair at the same paths and bump mtime.
+				newCertPath, newKeyPath, newDER := writeKeypair(t, dir, "b-")
+				mustRename(t, newCertPath, certPath)
+				mustRename(t, newKeyPath, keyPath)
+				future := time.Now().Add(time.Hour)
+				if err := os.Chtimes(certPath, future, future); err != nil {
+					t.Fatalf("Chtimes: %v", err)
+				}
+
+				got, err := kpr.GetCertificate(nil)
+				if err != nil {
+					t.Fatalf("GetCertificate: %v", err)
+				}
+				if len(got.Certificate) == 0 || !bytes.Equal(got.Certificate[0], newDER) {
+					t.Errorf("GetCertificate did not return reloaded cert")
+				}
+			},
+		},
+		{
+			name: "does not reload when mtime unchanged",
+			run: func(t *testing.T) {
+				dir := t.TempDir()
+				certPath, keyPath, originalDER := writeKeypair(t, dir, "a-")
+
+				kpr, err := NewKeypairReloader(certPath, keyPath, discardLogger())
+				if err != nil {
+					t.Fatalf("NewKeypairReloader: %v", err)
+				}
+
+				// Overwrite the cert/key files with a *different* keypair, then
+				// rewind mtime so the reloader must not pick up the change.
+				newCertPath, newKeyPath, newDER := writeKeypair(t, dir, "b-")
+				mustRename(t, newCertPath, certPath)
+				mustRename(t, newKeyPath, keyPath)
+				past := time.Unix(0, 0)
+				if err := os.Chtimes(certPath, past, past); err != nil {
+					t.Fatalf("Chtimes: %v", err)
+				}
+
+				got, err := kpr.GetCertificate(nil)
+				if err != nil {
+					t.Fatalf("GetCertificate: %v", err)
+				}
+				if len(got.Certificate) == 0 {
+					t.Fatal("empty cert chain")
+				}
+				if bytes.Equal(got.Certificate[0], newDER) {
+					t.Errorf("GetCertificate reloaded despite unchanged mtime")
+				}
+				if !bytes.Equal(got.Certificate[0], originalDER) {
+					t.Errorf("GetCertificate did not return original cert")
+				}
+			},
+		},
+		{
+			name: "does not panic when reload fails after mtime bump",
+			run: func(t *testing.T) {
+				dir := t.TempDir()
+				certPath, keyPath, _ := writeKeypair(t, dir, "a-")
+
+				kpr, err := NewKeypairReloader(certPath, keyPath, discardLogger())
+				if err != nil {
+					t.Fatalf("NewKeypairReloader: %v", err)
+				}
+
+				// Corrupt the cert file and bump mtime. maybeReload will fail.
+				if err := os.WriteFile(certPath, []byte("not a pem file"), 0o600); err != nil {
+					t.Fatalf("corrupt cert: %v", err)
+				}
+				future := time.Now().Add(time.Hour)
+				if err := os.Chtimes(certPath, future, future); err != nil {
+					t.Fatalf("Chtimes: %v", err)
+				}
+
+				defer func() {
+					if r := recover(); r != nil {
+						t.Fatalf("GetCertificate panicked on reload failure: %v", r)
+					}
+				}()
+
+				got, err := kpr.GetCertificate(nil)
+				if err == nil {
+					t.Errorf("GetCertificate returned nil err for corrupt cert; got %+v", got)
+				}
+			},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			tt.run(t)
+		})
+	}
+}
+
+func mustRename(t *testing.T, from, to string) {
+	t.Helper()
+	if err := os.Rename(from, to); err != nil {
+		t.Fatalf("rename %q -> %q: %v", from, to, err)
+	}
+}

lib/metrics/metrics.go

@@ -2,11 +2,14 @@ package metrics
 
 import (
 	"context"
+	"crypto/tls"
+	"crypto/x509"
 	"errors"
 	"fmt"
 	"log/slog"
 	"net/http"
 	"net/http/pprof"
+	"os"
 	"time"
 
 	"github.com/TecharoHQ/anubis/internal"
@@ -20,10 +23,16 @@ type Server struct {
 	Log    *slog.Logger
 }
 
-func (s *Server) Run(ctx context.Context, done func()) error {
+func (s *Server) Run(ctx context.Context, done func()) {
 	defer done()
 	lg := s.Log.With("subsystem", "metrics")
 
+	if err := s.run(ctx, lg); err != nil {
+		lg.Error("can't serve metrics server", "err", err)
+	}
+}
+
+func (s *Server) run(ctx context.Context, lg *slog.Logger) error {
 	mux := http.NewServeMux()
 	mux.HandleFunc("GET /debug/pprof/", pprof.Index)
 	mux.HandleFunc("GET /debug/pprof/cmdline", pprof.Cmdline)
@@ -62,6 +71,39 @@ func (s *Server) Run(ctx context.Context, done func()) error {
 
 	defer ln.Close()
 
+	if s.Config.TLS != nil {
+		kpr, err := NewKeypairReloader(s.Config.TLS.Certificate, s.Config.TLS.Key, lg)
+		if err != nil {
+			return fmt.Errorf("can't setup keypair reloader: %w", err)
+		}
+
+		srv.TLSConfig = &tls.Config{
+			GetCertificate: kpr.GetCertificate,
+		}
+
+		if s.Config.TLS.CA != "" {
+			caCert, err := os.ReadFile(s.Config.TLS.CA)
+			if err != nil {
+				return fmt.Errorf("%w %s: %w", config.ErrCantReadFile, s.Config.TLS.CA, err)
+			}
+
+			certPool := x509.NewCertPool()
+			if !certPool.AppendCertsFromPEM(caCert) {
+				return fmt.Errorf("%w %s", config.ErrInvalidMetricsCACertificate, s.Config.TLS.CA)
+			}
+
+			srv.TLSConfig.ClientCAs = certPool
+			srv.TLSConfig.ClientAuth = tls.RequireAndVerifyClientCert
+		}
+	}
+
+	if s.Config.BasicAuth != nil {
+		var h http.Handler = mux
+		h = internal.BasicAuth("anubis-metrics", s.Config.BasicAuth.Username, s.Config.BasicAuth.Password, mux)
+
+		srv.Handler = h
+	}
+
 	lg.Debug("listening for metrics", "url", metricsURL)
 
 	go func() {
@@ -73,9 +115,16 @@ func (s *Server) Run(ctx context.Context, done func()) error {
 		}
 	}()
 
+	switch s.Config.TLS != nil {
+	case true:
+		if err := srv.ServeTLS(ln, "", ""); !errors.Is(err, http.ErrServerClosed) {
+			return fmt.Errorf("can't serve TLS metrics server: %w", err)
+		}
+	case false:
 		if err := srv.Serve(ln); !errors.Is(err, http.ErrServerClosed) {
 			return fmt.Errorf("can't serve metrics server: %w", err)
 		}
+	}
 
 	return nil
 }