timeout handling

discv4/crawler.go

@@ -186,9 +186,14 @@ func (c *Crawler) crawlDiscV4(ctx context.Context, pi PeerInfo) <-chan DiscV4Res
 		// the number of probes to issue against bucket 0
 		probes := 3
 
-		closestMap, closestSet, err := c.probeBucket0(pi, probes)
+		closestMap, closestSet, respondedAt, err := c.probeBucket0(pi, probes, result.RespondedAt != nil)
 
 		if err == nil {
+			// track the respondedAt timestamp if it wasn't already set
+			if result.RespondedAt != nil && !respondedAt.IsZero() {
+				result.RespondedAt = &respondedAt
+			}
+
 			result.Strategy = determineStrategy(closestSet)
 
 			var remainingClosest map[peer.ID]PeerInfo
@@ -241,8 +246,9 @@ func (c *Crawler) crawlDiscV4(ctx context.Context, pi PeerInfo) <-chan DiscV4Res
 	return resultCh
 }
 
-func (c *Crawler) probeBucket0(pi PeerInfo, probes int) (map[peer.ID]PeerInfo, []mapset.Set[peer.ID], error) {
+func (c *Crawler) probeBucket0(pi PeerInfo, probes int, returnedENR bool) (map[peer.ID]PeerInfo, []mapset.Set[peer.ID], time.Time, error) {
 	var (
+		respondedAt time.Time
 		closestMap  = make(map[peer.ID]PeerInfo)
 		closestSets []mapset.Set[peer.ID]
 		errs        []error
@@ -257,13 +263,21 @@ func (c *Crawler) probeBucket0(pi PeerInfo, probes int) (map[peer.ID]PeerInfo, [
 		// first, we generate a random key that falls into bucket 0
 		targetKey, err := GenRandomPublicKey(pi.Node.ID(), 0)
 		if err != nil {
-			return nil, nil, err
+			return nil, nil, time.Time{}, err
 		}
 
 		// second, we do the Find node request
 		closest, err := c.listener.FindNode(pi.Node.ID(), pi.udpAddr, targetKey)
 		if err != nil {
+			// exit early if the node hasn't returned an ENR and the first probe
+			// also timed out
+			if !returnedENR && errors.Is(err, discover.ErrTimeout) {
+				return nil, nil, time.Time{}, fmt.Errorf("failed to probe bucket 0: %w", discover.ErrTimeout)
+			}
+
 			errs = append(errs, err)
+		} else if !respondedAt.IsZero() {
+			respondedAt = time.Now()
 		}
 
 		// third, we parse the responses into our [PeerInfo] struct
@@ -281,10 +295,10 @@ func (c *Crawler) probeBucket0(pi PeerInfo, probes int) (map[peer.ID]PeerInfo, [
 	}
 
 	if len(errs) == probes {
-		return nil, nil, fmt.Errorf("failed to probe bucket 0: %w", errors.Join(errs...))
+		return nil, nil, time.Time{}, fmt.Errorf("failed to probe bucket 0: %w", errors.Join(errs...))
 	}
 
-	return closestMap, closestSets, nil
+	return closestMap, closestSets, respondedAt, nil
 }
 
 type CrawlStrategy string

discv4/driver_crawler.go

@@ -131,7 +131,7 @@ func (cfg *CrawlDriverConfig) CrawlerConfig() *CrawlerConfig {
 		DialTimeout:  cfg.DialTimeout,
 		AddrDialType: cfg.AddrDialType,
 		LogErrors:    cfg.LogErrors,
-		MaxJitter:    time.Duration(cfg.CrawlWorkerCount/100) * time.Second, // 3000 workers -> distributed over 30s
+		MaxJitter:    time.Duration(cfg.CrawlWorkerCount/50) * time.Second, // e.g., 3000 workers evenly distributed over 60s
 		KeepENR:      false,
 	}
 }
@@ -204,6 +204,9 @@ func NewCrawlDriver(dbc db.Client, crawl *models.Crawl, cfg *CrawlDriverConfig)
 		return nil, fmt.Errorf("create unhandled packets counter: %w", err)
 	}
 
+	// set the discovery response timeout
+	discover.RespTimeout = 2 * time.Second
+
 	d := &CrawlDriver{
 		cfg:            cfg,
 		dbc:            dbc,

discv4/gen_test.go

@@ -7,9 +7,8 @@ import (
 
 	"github.com/ethereum/go-ethereum/crypto"
 	"github.com/ethereum/go-ethereum/p2p/enode"
-	"github.com/stretchr/testify/require"
-
 	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
 )
 
 func TestGenRandomPublicKey(t *testing.T) {