wgengine/magicsock: keep advertising endpoints after we stop discovering them

Previously, when updating endpoints we would immediately stop advertising any endpoint that wasn't discovered during determineEndpoints. This could result in, for example, a case where we performed an incremental netcheck, didn't get any of our three STUN packets back, and then dropped our STUN endpoint from the set of advertised endpoints... which would result in clients falling back to a DERP connection until the next call to determineEndpoints. Instead, let's cache endpoints that we've discovered and continue reporting them to clients until a timeout expires. In the above case where we temporarily don't have a discovered STUN endpoint, we would continue reporting the old value, then re-discover the STUN endpoint again and continue reporting it as normal, so clients never see a withdrawal. Updates tailscale/coral#108 Signed-off-by: Andrew Dunham <andrew@du.nham.ca> Change-Id: I42de72e7418ab328a6c732bdefc74549708cf8b9
2023-04-14 16:37:10 -04:00
parent 4b49ca4a12
commit 80b138f0df
2 changed files with 212 additions and 0 deletions
@@ -18,6 +18,7 @@ import (
 	"net/http/httptest"
 	"net/netip"
 	"os"
+	"reflect"
 	"runtime"
 	"strconv"
 	"strings"
@@ -31,6 +32,7 @@ import (
 	"github.com/tailscale/wireguard-go/tun/tuntest"
 	"go4.org/mem"
 	"golang.org/x/exp/maps"
+	"golang.org/x/exp/slices"
 	"golang.org/x/net/ipv6"
 	"tailscale.com/cmd/testwrapper/flakytest"
 	"tailscale.com/derp"
@@ -390,6 +392,7 @@ collectEndpoints:
 	for {
 		select {
 		case ep := <-epCh:
+			t.Logf("TestNewConn: got endpoint: %v", ep)
 			endpoints = append(endpoints, ep)
 			if strings.HasSuffix(ep, suffix) {
 				break collectEndpoints
@@ -2280,3 +2283,113 @@ func TestIsWireGuardOnlyPeerWithMasquerade(t *testing.T) {
 		t.Fatal("no packet after 1s")
 	}
 }
+
+func TestEndpointTracker(t *testing.T) {
+	local := tailcfg.Endpoint{
+		Addr: netip.MustParseAddrPort("192.168.1.1:12345"),
+		Type: tailcfg.EndpointLocal,
+	}
+
+	stun4_1 := tailcfg.Endpoint{
+		Addr: netip.MustParseAddrPort("1.2.3.4:12345"),
+		Type: tailcfg.EndpointSTUN,
+	}
+	stun4_2 := tailcfg.Endpoint{
+		Addr: netip.MustParseAddrPort("5.6.7.8:12345"),
+		Type: tailcfg.EndpointSTUN,
+	}
+
+	stun6_1 := tailcfg.Endpoint{
+		Addr: netip.MustParseAddrPort("[2a09:8280:1::1111]:12345"),
+		Type: tailcfg.EndpointSTUN,
+	}
+	stun6_2 := tailcfg.Endpoint{
+		Addr: netip.MustParseAddrPort("[2a09:8280:1::2222]:12345"),
+		Type: tailcfg.EndpointSTUN,
+	}
+
+	start := time.Unix(1681503440, 0)
+
+	steps := []struct {
+		name string
+		now  time.Time
+		eps  []tailcfg.Endpoint
+		want []tailcfg.Endpoint
+	}{
+		{
+			name: "initial endpoints",
+			now:  start,
+			eps:  []tailcfg.Endpoint{local, stun4_1, stun6_1},
+			want: []tailcfg.Endpoint{local, stun4_1, stun6_1},
+		},
+		{
+			name: "no change",
+			now:  start.Add(1 * time.Minute),
+			eps:  []tailcfg.Endpoint{local, stun4_1, stun6_1},
+			want: []tailcfg.Endpoint{local, stun4_1, stun6_1},
+		},
+		{
+			name: "missing stun4",
+			now:  start.Add(2 * time.Minute),
+			eps:  []tailcfg.Endpoint{local, stun6_1},
+			want: []tailcfg.Endpoint{local, stun4_1, stun6_1},
+		},
+		{
+			name: "missing stun6",
+			now:  start.Add(3 * time.Minute),
+			eps:  []tailcfg.Endpoint{local, stun4_1},
+			want: []tailcfg.Endpoint{local, stun4_1, stun6_1},
+		},
+		{
+			name: "multiple STUN addresses within timeout",
+			now:  start.Add(4 * time.Minute),
+			eps:  []tailcfg.Endpoint{local, stun4_2, stun6_2},
+			want: []tailcfg.Endpoint{local, stun4_1, stun4_2, stun6_1, stun6_2},
+		},
+		{
+			name: "endpoint extended",
+			now:  start.Add(3*time.Minute + endpointTrackerLifetime - 1),
+			eps:  []tailcfg.Endpoint{local},
+			want: []tailcfg.Endpoint{
+				local, stun4_2, stun6_2,
+				// stun4_1 had its lifetime extended by the
+				// "missing stun6" test above to that start
+				// time plus the lifetime, while stun6 should
+				// have expired a minute sooner. It should thus
+				// be in this returned list.
+				stun4_1,
+			},
+		},
+		{
+			name: "after timeout",
+			now:  start.Add(4*time.Minute + endpointTrackerLifetime + 1),
+			eps:  []tailcfg.Endpoint{local, stun4_2, stun6_2},
+			want: []tailcfg.Endpoint{local, stun4_2, stun6_2},
+		},
+		{
+			name: "after timeout still caches",
+			now:  start.Add(4*time.Minute + endpointTrackerLifetime + time.Minute),
+			eps:  []tailcfg.Endpoint{local},
+			want: []tailcfg.Endpoint{local, stun4_2, stun6_2},
+		},
+	}
+
+	var et endpointTracker
+	for _, tt := range steps {
+		t.Logf("STEP: %s", tt.name)
+
+		got := et.update(tt.now, tt.eps)
+
+		// Sort both arrays for comparison
+		slices.SortFunc(got, func(a, b tailcfg.Endpoint) bool {
+			return a.Addr.String() < b.Addr.String()
+		})
+		slices.SortFunc(tt.want, func(a, b tailcfg.Endpoint) bool {
+			return a.Addr.String() < b.Addr.String()
+		})
+
+		if !reflect.DeepEqual(got, tt.want) {
+			t.Errorf("endpoints mismatch\ngot: %+v\nwant: %+v", got, tt.want)
+		}
+	}
+}