fix(p2p): update peer address on reconnect and optimize reacher (#5348)

Author: gacevicljubisa

pkg/p2p/libp2p/internal/reacher/heap.go

@@ -0,0 +1,33 @@
+// Copyright 2026 The Swarm Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package reacher
+
+// peerHeap is a min-heap of peers ordered by retryAfter time.
+type peerHeap []*peer
+
+func (h peerHeap) Len() int           { return len(h) }
+func (h peerHeap) Less(i, j int) bool { return h[i].retryAfter.Before(h[j].retryAfter) }
+func (h peerHeap) Swap(i, j int) {
+	h[i], h[j] = h[j], h[i]
+	h[i].index = i
+	h[j].index = j
+}
+
+func (h *peerHeap) Push(x any) {
+	n := len(*h)
+	p := x.(*peer)
+	p.index = n
+	*h = append(*h, p)
+}
+
+func (h *peerHeap) Pop() any {
+	old := *h
+	n := len(old)
+	p := old[n-1]
+	old[n-1] = nil // avoid memory leak
+	p.index = -1   // for safety
+	*h = old[0 : n-1]
+	return p
+}

pkg/p2p/libp2p/internal/reacher/reacher.go

@@ -7,6 +7,7 @@
 package reacher
 
 import (
+	"container/heap"
 	"context"
 	"sync"
 	"time"
@@ -27,11 +28,13 @@ type peer struct {
 	overlay    swarm.Address
 	addr       ma.Multiaddr
 	retryAfter time.Time
+	index      int // index in the heap
 }
 
 type reacher struct {
-	mu    sync.Mutex
-	peers map[string]*peer
+	mu        sync.Mutex
+	peerHeap  peerHeap         // min-heap ordered by retryAfter
+	peerIndex map[string]*peer // lookup by overlay for O(1) access
 
 	newPeer chan struct{}
 	quit    chan struct{}
@@ -53,12 +56,13 @@ type Options struct {
 
 func New(streamer p2p.Pinger, notifier p2p.ReachableNotifier, o *Options, log log.Logger) *reacher {
 	r := &reacher{
-		newPeer:  make(chan struct{}, 1),
-		quit:     make(chan struct{}),
-		pinger:   streamer,
-		peers:    make(map[string]*peer),
-		notifier: notifier,
-		logger:   log.WithName("reacher").Register(),
+		newPeer:   make(chan struct{}, 1),
+		quit:      make(chan struct{}),
+		pinger:    streamer,
+		peerHeap:  make(peerHeap, 0),
+		peerIndex: make(map[string]*peer),
+		notifier:  notifier,
+		logger:    log.WithName("reacher").Register(),
 	}
 
 	if o == nil {
@@ -79,7 +83,7 @@ func New(streamer p2p.Pinger, notifier p2p.ReachableNotifier, o *Options, log lo
 func (r *reacher) manage() {
 	defer r.wg.Done()
 
-	c := make(chan *peer)
+	c := make(chan peer)
 	defer close(c)
 
 	ctx, cancel := context.WithCancel(context.Background())
@@ -92,7 +96,7 @@ func (r *reacher) manage() {
 
 	for {
 
-		p, tryAfter := r.tryAcquirePeer()
+		p, ok, tryAfter := r.tryAcquirePeer()
 
 		// if no peer is returned,
 		// wait until either more work or the closest retry-after time.
@@ -110,7 +114,7 @@ func (r *reacher) manage() {
 		}
 
 		// wait for work
-		if p == nil {
+		if !ok {
 			select {
 			case <-r.quit:
 				return
@@ -128,7 +132,7 @@ func (r *reacher) manage() {
 	}
 }
 
-func (r *reacher) ping(c chan *peer, ctx context.Context) {
+func (r *reacher) ping(c chan peer, ctx context.Context) {
 	defer r.wg.Done()
 	for p := range c {
 		func() {
@@ -146,38 +150,34 @@ func (r *reacher) ping(c chan *peer, ctx context.Context) {
 	}
 }
 
-func (r *reacher) tryAcquirePeer() (*peer, time.Duration) {
+func (r *reacher) tryAcquirePeer() (peer, bool, time.Duration) {
 	r.mu.Lock()
 	defer r.mu.Unlock()
 
-	var (
-		now         = time.Now()
-		nextClosest time.Time
-	)
+	if len(r.peerHeap) == 0 {
+		return peer{}, false, 0
+	}
 
-	for _, p := range r.peers {
+	now := time.Now()
 
-		// retry after has expired, retry
-		if now.After(p.retryAfter) {
-			p.retryAfter = time.Now().Add(r.options.RetryAfterDuration)
-			return p, 0
-		}
+	// Peek at the peer with the earliest retryAfter
+	p := r.peerHeap[0]
 
-		// here, we find the peer with the earliest retry after
-		if nextClosest.IsZero() || p.retryAfter.Before(nextClosest) {
-			nextClosest = p.retryAfter
-		}
+	// If retryAfter has not expired, return time to wait
+	if now.Before(p.retryAfter) {
+		return peer{}, false, time.Until(p.retryAfter)
 	}
 
-	if nextClosest.IsZero() {
-		return nil, 0
-	}
+	// Update retryAfter and fix heap position
+	p.retryAfter = time.Now().Add(r.options.RetryAfterDuration)
+	heap.Fix(&r.peerHeap, p.index)
 
-	// return the time to wait until the closest retry after
-	return nil, time.Until(nextClosest)
+	// Return a copy so callers can read fields without holding the lock.
+	return *p, true, 0
 }
 
 // Connected adds a new peer to the queue for testing reachability.
+// If the peer already exists, its address is updated.
 func (r *reacher) Connected(overlay swarm.Address, addr ma.Multiaddr) {
 	if addr == nil {
 		return
@@ -186,8 +186,15 @@ func (r *reacher) Connected(overlay swarm.Address, addr ma.Multiaddr) {
 	r.mu.Lock()
 	defer r.mu.Unlock()
 
-	if _, ok := r.peers[overlay.ByteString()]; !ok {
-		r.peers[overlay.ByteString()] = &peer{overlay: overlay, addr: addr}
+	key := overlay.ByteString()
+	if existing, ok := r.peerIndex[key]; ok {
+		existing.addr = addr              // Update address for reconnecting peer
+		existing.retryAfter = time.Time{} // Reset to trigger immediate re-ping
+		heap.Fix(&r.peerHeap, existing.index)
+	} else {
+		p := &peer{overlay: overlay, addr: addr}
+		r.peerIndex[key] = p
+		heap.Push(&r.peerHeap, p)
 	}
 
 	select {
@@ -201,7 +208,11 @@ func (r *reacher) Disconnected(overlay swarm.Address) {
 	r.mu.Lock()
 	defer r.mu.Unlock()
 
-	delete(r.peers, overlay.ByteString())
+	key := overlay.ByteString()
+	if p, ok := r.peerIndex[key]; ok {
+		heap.Remove(&r.peerHeap, p.index)
+		delete(r.peerIndex, key)
+	}
 }
 
 // Close stops the worker. Must be called once.

pkg/p2p/libp2p/internal/reacher/reacher_test.go

@@ -7,6 +7,7 @@ package reacher_test
 import (
 	"context"
 	"errors"
+	"sync"
 	"testing"
 	"time"
 
@@ -127,6 +128,167 @@ func TestDisconnected(t *testing.T) {
 	})
 }
 
+func TestAddressUpdateOnReconnect(t *testing.T) {
+	t.Parallel()
+
+	synctest.Test(t, func(t *testing.T) {
+		// Use 1 worker and a known retry duration to make timing deterministic.
+		options := reacher.Options{
+			PingTimeout:        time.Second * 5,
+			Workers:            1,
+			RetryAfterDuration: time.Minute,
+		}
+
+		overlay := swarm.RandAddress(t)
+		oldAddr, _ := ma.NewMultiaddr("/ip4/127.0.0.1/tcp/7071/p2p/16Uiu2HAmTBuJT9LvNmBiQiNoTsxE5mtNy6YG3paw79m94CRa9sRb")
+		newAddr, _ := ma.NewMultiaddr("/ip4/192.168.1.1/tcp/7072/p2p/16Uiu2HAmTBuJT9LvNmBiQiNoTsxE5mtNy6YG3paw79m94CRa9sRb")
+
+		var pingsMu sync.Mutex
+		var pings []ma.Multiaddr
+		pinged := make(chan struct{}, 8)
+
+		pingFunc := func(_ context.Context, a ma.Multiaddr) (time.Duration, error) {
+			pingsMu.Lock()
+			pings = append(pings, a)
+			pingsMu.Unlock()
+			pinged <- struct{}{}
+			return 0, nil
+		}
+
+		reachableFunc := func(addr swarm.Address, status p2p.ReachabilityStatus) {}
+
+		mock := newMock(pingFunc, reachableFunc)
+
+		r := reacher.New(mock, mock, &options, log.Noop)
+		testutil.CleanupCloser(t, r)
+
+		// First connection with old address – triggers initial ping.
+		r.Connected(overlay, oldAddr)
+
+		select {
+		case <-time.After(time.Second * 10):
+			t.Fatal("timed out waiting for initial ping")
+		case <-pinged:
+		}
+
+		// Verify old address was pinged first.
+		pingsMu.Lock()
+		if len(pings) != 1 {
+			t.Fatalf("expected 1 ping after initial connect, got %d", len(pings))
+		}
+		if !pings[0].Equal(oldAddr) {
+			t.Fatalf("first ping should use old address, got %s", pings[0])
+		}
+		pingsMu.Unlock()
+
+		// Reconnect with a new address — should trigger immediate re-ping.
+		r.Connected(overlay, newAddr)
+
+		select {
+		case <-time.After(time.Second * 10):
+			t.Fatal("timed out waiting for reconnect ping")
+		case <-pinged:
+		}
+
+		// Verify the reconnect pinged the new address.
+		pingsMu.Lock()
+		if len(pings) != 2 {
+			t.Fatalf("expected 2 pings after reconnect, got %d", len(pings))
+		}
+		if !pings[1].Equal(newAddr) {
+			t.Fatalf("reconnect ping should use new address, got %s", pings[1])
+		}
+		pingsMu.Unlock()
+
+		// Advance time past the retry duration — should trigger a scheduled re-ping.
+		time.Sleep(options.RetryAfterDuration + time.Second)
+
+		select {
+		case <-time.After(time.Second * 10):
+			t.Fatal("timed out waiting for scheduled re-ping")
+		case <-pinged:
+		}
+
+		// Verify the scheduled re-ping used the new address.
+		pingsMu.Lock()
+		if len(pings) != 3 {
+			t.Fatalf("expected 3 pings after retry duration, got %d", len(pings))
+		}
+		if !pings[2].Equal(newAddr) {
+			t.Fatalf("scheduled re-ping should use new address, got %s", pings[2])
+		}
+		pingsMu.Unlock()
+	})
+}
+
+func TestHeapOrdering(t *testing.T) {
+	t.Parallel()
+
+	synctest.Test(t, func(t *testing.T) {
+		// Use single worker to ensure sequential processing
+		options := reacher.Options{
+			PingTimeout:        time.Second * 5,
+			Workers:            1,
+			RetryAfterDuration: time.Second * 10,
+		}
+
+		var pingOrder []swarm.Address
+		var pingOrderMu sync.Mutex
+		allPinged := make(chan struct{})
+
+		overlay1 := swarm.RandAddress(t)
+		overlay2 := swarm.RandAddress(t)
+		overlay3 := swarm.RandAddress(t)
+		addr, _ := ma.NewMultiaddr("/ip4/127.0.0.1/tcp/7071/p2p/16Uiu2HAmTBuJT9LvNmBiQiNoTsxE5mtNy6YG3paw79m94CRa9sRb")
+
+		pingFunc := func(_ context.Context, _ ma.Multiaddr) (time.Duration, error) {
+			return 0, nil
+		}
+
+		reachableFunc := func(overlay swarm.Address, status p2p.ReachabilityStatus) {
+			pingOrderMu.Lock()
+			pingOrder = append(pingOrder, overlay)
+			if len(pingOrder) == 3 {
+				close(allPinged)
+			}
+			pingOrderMu.Unlock()
+		}
+
+		mock := newMock(pingFunc, reachableFunc)
+
+		r := reacher.New(mock, mock, &options, log.Noop)
+		testutil.CleanupCloser(t, r)
+
+		// Add peers - they should all be pinged since retryAfter starts at zero
+		r.Connected(overlay1, addr)
+		r.Connected(overlay2, addr)
+		r.Connected(overlay3, addr)
+
+		select {
+		case <-time.After(time.Second * 5):
+			t.Fatalf("test timed out, only %d peers pinged", len(pingOrder))
+		case <-allPinged:
+		}
+
+		// Verify all three peers were pinged
+		pingOrderMu.Lock()
+		defer pingOrderMu.Unlock()
+
+		if len(pingOrder) != 3 {
+			t.Fatalf("expected 3 peers pinged, got %d", len(pingOrder))
+		}
+
+		// Verify all overlays are present (order may vary due to heap with same retryAfter)
+		seen := make(map[string]bool)
+		for _, o := range pingOrder {
+			seen[o.String()] = true
+		}
+		if !seen[overlay1.String()] || !seen[overlay2.String()] || !seen[overlay3.String()] {
+			t.Fatalf("not all peers were pinged")
+		}
+	})
+}
+
 type mock struct {
 	pingFunc      func(context.Context, ma.Multiaddr) (time.Duration, error)
 	reachableFunc func(swarm.Address, p2p.ReachabilityStatus)