From 0f41570c80858e0081392c03977e0f77c0d21299 Mon Sep 17 00:00:00 2001
From: Anton Kaliaev <anton.kalyaev@gmail.com>
Date: Sun, 11 Mar 2018 13:20:51 +0400
Subject: [PATCH] fixes from bucky's review

---
 p2p/pex/pex_reactor.go      | 23 ++++++++++++++---------
 rpc/lib/client/ws_client.go |  4 +---
 2 files changed, 15 insertions(+), 12 deletions(-)

diff --git a/p2p/pex/pex_reactor.go b/p2p/pex/pex_reactor.go
index 73df67dca..aff1ab714 100644
--- a/p2p/pex/pex_reactor.go
+++ b/p2p/pex/pex_reactor.go
@@ -39,6 +39,8 @@ const (
 	defaultSeedDisconnectWaitPeriod = 2 * time.Minute  // disconnect after this
 	defaultCrawlPeerInterval        = 2 * time.Minute  // dont redial for this. TODO: back-off
 	defaultCrawlPeersPeriod         = 30 * time.Second // check some peers every this
+
+	maxAttemptsToDial = 16 // ~ 35h in total (last attempt - 18h sleep)
 )
 
 // PEXReactor handles PEX (peer exchange) and ensures that an
@@ -61,7 +63,7 @@ type PEXReactor struct {
 	requestsSent         *cmn.CMap // ID->struct{}: unanswered send requests
 	lastReceivedRequests *cmn.CMap // ID->time.Time: last time peer requested from us
 
-	attemptsToDial sync.Map // dial addr -> number of attempts to dial (for exponential backoff)
+	attemptsToDial sync.Map // dial address (string) -> number of attempts (int) to dial (for exponential backoff)
 }
 
 // PEXReactorConfig holds reactor specific configuration data.
@@ -365,22 +367,25 @@ func (r *PEXReactor) ensurePeers() {
 }
 
 func (r *PEXReactor) dialPeer(addr *p2p.NetAddress) {
-	// 1s == (1e9 ns) == (1 Billion ns)
-	billionNs := float64(time.Second.Nanoseconds())
+	attempts := r.AttemptsToDial(addr)
+
+	if attempts > maxAttemptsToDial {
+		r.Logger.Error("Reached max attempts to dial", "addr", addr, "attempts", attempts)
+		r.book.MarkBad(addr)
+		return
+	}
 
 	// exponential backoff if it's not our first attempt to dial given address
-	var attempts int
-	if lAttempts, attempted := r.attemptsToDial.Load(addr.DialString()); attempted {
-		attempts = lAttempts.(int)
-		jitterSeconds := time.Duration(rand.Float64() * billionNs)
+	if attempts != 0 {
+		jitterSeconds := time.Duration(rand.Float64() * float64(time.Second)) // 1s == (1e9 ns)
 		backoffDuration := jitterSeconds + ((1 << uint(attempts)) * time.Second)
-		r.Logger.Debug(fmt.Sprintf("Dialing %v", addr), "attempts", attempts, "backoff_duration", backoffDuration)
+		r.Logger.Debug("Sleeping before dialing", "addr", addr, "dur", backoffDuration)
 		time.Sleep(backoffDuration)
 	}
 
 	err := r.Switch.DialPeerWithAddress(addr, false)
 	if err != nil {
-		r.Logger.Error("Dialing failed", "err", err)
+		r.Logger.Error("Dialing failed", "addr", addr, "err", err, "attempts", attempts)
 		// TODO: detect more "bad peer" scenarios
 		if _, ok := err.(p2p.ErrSwitchAuthenticationFailure); ok {
 			r.book.MarkBad(addr)
diff --git a/rpc/lib/client/ws_client.go b/rpc/lib/client/ws_client.go
index ca75ad561..fe15cda21 100644
--- a/rpc/lib/client/ws_client.go
+++ b/rpc/lib/client/ws_client.go
@@ -254,10 +254,8 @@ func (c *WSClient) reconnect() error {
 		c.mtx.Unlock()
 	}()
 
-	// 1s == (1e9 ns) == (1 Billion ns)
-	billionNs := float64(time.Second.Nanoseconds())
 	for {
-		jitterSeconds := time.Duration(rand.Float64() * billionNs)
+		jitterSeconds := time.Duration(rand.Float64() * float64(time.Second)) // 1s == (1e9 ns)
 		backoffDuration := jitterSeconds + ((1 << uint(attempt)) * time.Second)
 
 		c.Logger.Info("reconnecting", "attempt", attempt+1, "backoff_duration", backoffDuration)