fix: configurable handshake validity interval and improved defaults (#645)

* fix: support configurable rekey timeout interval for peer connectivity tracking (#641)

* change default check-time to 180s

docs/documentation/configuration/overview.md

@@ -28,6 +28,7 @@ core:
   
 backend:
   default: local
+  rekey_timeout_interval: 125s
   local_resolvconf_prefix: tun.
 
 advanced:
@@ -203,6 +204,13 @@ The current MikroTik backend is in **BETA** and may not support all features.
 - **Description:** The default backend to use for managing WireGuard interfaces. 
   Valid options are: `local`, or other backend id's configured in the `mikrotik` section.
 
+### `rekey_timeout_interval`
+- **Default:** `180s`
+- **Environment Variable:** `WG_PORTAL_BACKEND_REKEY_TIMEOUT_INTERVAL`
+- **Description:** The interval after which a WireGuard peer is considered disconnected if no handshake updates are received. 
+  This corresponds to the WireGuard rekey timeout setting of 120 seconds plus a 60-second buffer to account for latency or retry handling.
+  Uses Go duration format (e.g., `10s`, `1m`). If omitted, a default of 180 seconds is used.
+
 ### `local_resolvconf_prefix`
 - **Default:** `tun.`
 - **Environment Variable:** `WG_PORTAL_BACKEND_LOCAL_RESOLVCONF_PREFIX`

internal/app/wireguard/statistics.go

@@ -204,13 +204,13 @@ func (c *StatisticsCollector) collectPeerData(ctx context.Context) {
 
 							// calculate if session was restarted
 							p.UpdatedAt = now
-							p.LastSessionStart = getSessionStartTime(*p, peer.BytesUpload, peer.BytesDownload,
+							p.LastSessionStart = c.getSessionStartTime(*p, peer.BytesUpload, peer.BytesDownload,
 								lastHandshake)
 							p.BytesReceived = peer.BytesUpload      // store bytes that where uploaded from the peer and received by the server
 							p.BytesTransmitted = peer.BytesDownload // store bytes that where received from the peer and sent by the server
 							p.Endpoint = peer.Endpoint
 							p.LastHandshake = lastHandshake
-							p.CalcConnected()
+							p.CalcConnected(c.cfg.Backend.ReKeyTimeoutInterval)
 
 							if wasConnected != p.IsConnected {
 								slog.Debug("peer connection state changed",
@@ -249,7 +249,7 @@ func (c *StatisticsCollector) collectPeerData(ctx context.Context) {
 	}
 }
 
-func getSessionStartTime(
+func (c *StatisticsCollector) getSessionStartTime(
 	oldStats domain.PeerStatus,
 	newReceived, newTransmitted uint64,
 	latestHandshake *time.Time,
@@ -258,7 +258,7 @@ func getSessionStartTime(
 		return nil // currently not connected
 	}
 
-	oldestHandshakeTime := time.Now().Add(-2 * time.Minute) // if a handshake is older than 2 minutes, the peer is no longer connected
+	oldestHandshakeTime := time.Now().Add(-1 * c.cfg.Backend.ReKeyTimeoutInterval) // if a handshake is older than the rekey interval + grace-period, the peer is no longer connected
 	switch {
 	// old session was never initiated
 	case oldStats.BytesReceived == 0 && oldStats.BytesTransmitted == 0 && (newReceived > 0 || newTransmitted > 0):
@@ -369,7 +369,7 @@ func (c *StatisticsCollector) pingWorker(ctx context.Context) {
 					p.LastPing = nil
 				}
 				p.UpdatedAt = time.Now()
-				p.CalcConnected()
+				p.CalcConnected(c.cfg.Backend.ReKeyTimeoutInterval)
 
 				if wasConnected != p.IsConnected {
 					connectionStateChanged = true

internal/app/wireguard/statistics_test.go

@@ -5,10 +5,11 @@ import (
 	"testing"
 	"time"
 
+	"github.com/h44z/wg-portal/internal/config"
 	"github.com/h44z/wg-portal/internal/domain"
 )
 
-func Test_getSessionStartTime(t *testing.T) {
+func TestStatisticsCollector_getSessionStartTime(t *testing.T) {
 	now := time.Now()
 	nowMinus1 := now.Add(-1 * time.Minute)
 	nowMinus3 := now.Add(-3 * time.Minute)
@@ -133,7 +134,14 @@ func Test_getSessionStartTime(t *testing.T) {
 	}
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
-			if got := getSessionStartTime(tt.args.oldStats, tt.args.newReceived, tt.args.newTransmitted,
+			c := &StatisticsCollector{
+				cfg: &config.Config{
+					Backend: config.Backend{
+						ReKeyTimeoutInterval: 180 * time.Second,
+					},
+				},
+			}
+			if got := c.getSessionStartTime(tt.args.oldStats, tt.args.newReceived, tt.args.newTransmitted,
 				tt.args.lastHandshake); !reflect.DeepEqual(got, tt.want) {
 				t.Errorf("getSessionStartTime() = %v, want %v", got, tt.want)
 			}

internal/config/backend.go

@@ -10,6 +10,8 @@ const LocalBackendName = "local"
 type Backend struct {
 	Default string `yaml:"default"` // The default backend to use (defaults to the internal backend)
 
+	ReKeyTimeoutInterval time.Duration `yaml:"rekey_timeout_interval"` // Interval after which a connection is assumed dead
+
 	// Local Backend-specific configuration
 
 	IgnoredLocalInterfaces []string `yaml:"ignored_local_interfaces"` // A list of interface names that should be ignored by this backend (e.g., "wg0")
@@ -115,8 +117,8 @@ func (b *BackendMikrotik) GetApiTimeout() time.Duration {
 type BackendPfsense struct {
 	BackendBase `yaml:",inline"` // Embed the base fields
 
-	ApiUrl       string        `yaml:"api_url"` // The base URL of the pfSense REST API (e.g., "https://pfsense.example.com/api/v2")
-	ApiKey       string        `yaml:"api_key"` // API key for authentication (generated in pfSense under 'System' -> 'REST API' -> 'Keys')
+	ApiUrl       string        `yaml:"api_url"`        // The base URL of the pfSense REST API (e.g., "https://pfsense.example.com/api/v2")
+	ApiKey       string        `yaml:"api_key"`        // API key for authentication (generated in pfSense under 'System' -> 'REST API' -> 'Keys')
 	ApiVerifyTls bool          `yaml:"api_verify_tls"` // Whether to verify the TLS certificate of the pfSense API
 	ApiTimeout   time.Duration `yaml:"api_timeout"`    // Timeout for API requests (default: 30 seconds)
 

internal/config/config.go

@@ -139,6 +139,7 @@ func defaultConfig() *Config {
 
 	cfg.Backend = Backend{
 		Default:                LocalBackendName, // local backend is the default (using wgcrtl)
+		ReKeyTimeoutInterval:   getEnvDuration("WG_PORTAL_BACKEND_REKEY_TIMEOUT_INTERVAL", 180*time.Second),
 		IgnoredLocalInterfaces: getEnvStrSlice("WG_PORTAL_BACKEND_IGNORED_LOCAL_INTERFACES", nil),
 		// Most resolconf implementations use "tun." as a prefix for interface names.
 		// But systemd's implementation uses no prefix, for example.

internal/domain/statistics.go

@@ -21,8 +21,8 @@ type PeerStatus struct {
 	LastSessionStart *time.Time `gorm:"column:last_session_start" json:"LastSessionStart"`
 }
 
-func (s *PeerStatus) CalcConnected() {
-	oldestHandshakeTime := time.Now().Add(-2 * time.Minute) // if a handshake is older than 2 minutes, the peer is no longer connected
+func (s *PeerStatus) CalcConnected(timeout time.Duration) {
+	oldestHandshakeTime := time.Now().Add(-1 * timeout) // if a handshake is older than the rekey-interval + grace-period, the peer is no longer connected
 
 	handshakeValid := false
 	if s.LastHandshake != nil {

internal/domain/statistics_test.go

@@ -9,11 +9,16 @@ func TestPeerStatus_IsConnected(t *testing.T) {
 	now := time.Now()
 	past := now.Add(-3 * time.Minute)
 	recent := now.Add(-1 * time.Minute)
+	defaultTimeout := 125 * time.Second // rekey interval of 120s + 5 seconds grace period
+	past126 := now.Add(-1*defaultTimeout - 1*time.Second)
+	past125 := now.Add(-1 * defaultTimeout)
+	past124 := now.Add(-1*defaultTimeout + 1*time.Second)
 
 	tests := []struct {
-		name   string
-		status PeerStatus
-		want   bool
+		name    string
+		status  PeerStatus
+		timeout time.Duration
+		want    bool
 	}{
 		{
 			name: "Pingable and recent handshake",
@@ -21,7 +26,8 @@ func TestPeerStatus_IsConnected(t *testing.T) {
 				IsPingable:    true,
 				LastHandshake: &recent,
 			},
-			want: true,
+			timeout: defaultTimeout,
+			want:    true,
 		},
 		{
 			name: "Not pingable but recent handshake",
@@ -29,7 +35,8 @@ func TestPeerStatus_IsConnected(t *testing.T) {
 				IsPingable:    false,
 				LastHandshake: &recent,
 			},
-			want: true,
+			timeout: defaultTimeout,
+			want:    true,
 		},
 		{
 			name: "Pingable but old handshake",
@@ -37,15 +44,44 @@ func TestPeerStatus_IsConnected(t *testing.T) {
 				IsPingable:    true,
 				LastHandshake: &past,
 			},
-			want: true,
+			timeout: defaultTimeout,
+			want:    true,
 		},
 		{
-			name: "Not pingable and old handshake",
+			name: "Not pingable and ok handshake (-124s)",
+			status: PeerStatus{
+				IsPingable:    false,
+				LastHandshake: &past124,
+			},
+			timeout: defaultTimeout,
+			want:    true,
+		},
+		{
+			name: "Not pingable and old handshake (-125s)",
+			status: PeerStatus{
+				IsPingable:    false,
+				LastHandshake: &past125,
+			},
+			timeout: defaultTimeout,
+			want:    false,
+		},
+		{
+			name: "Not pingable and old handshake (-126s)",
+			status: PeerStatus{
+				IsPingable:    false,
+				LastHandshake: &past126,
+			},
+			timeout: defaultTimeout,
+			want:    false,
+		},
+		{
+			name: "Not pingable and old handshake (very old)",
 			status: PeerStatus{
 				IsPingable:    false,
 				LastHandshake: &past,
 			},
-			want: false,
+			timeout: defaultTimeout,
+			want:    false,
 		},
 		{
 			name: "Pingable and no handshake",
@@ -53,7 +89,8 @@ func TestPeerStatus_IsConnected(t *testing.T) {
 				IsPingable:    true,
 				LastHandshake: nil,
 			},
-			want: true,
+			timeout: defaultTimeout,
+			want:    true,
 		},
 		{
 			name: "Not pingable and no handshake",
@@ -61,12 +98,13 @@ func TestPeerStatus_IsConnected(t *testing.T) {
 				IsPingable:    false,
 				LastHandshake: nil,
 			},
-			want: false,
+			timeout: defaultTimeout,
+			want:    false,
 		},
 	}
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
-			tt.status.CalcConnected()
+			tt.status.CalcConnected(tt.timeout)
 			if got := tt.status.IsConnected; got != tt.want {
 				t.Errorf("IsConnected = %v, want %v", got, tt.want)
 			}