diff --git a/beacon_chain/networking/eth2_network.nim b/beacon_chain/networking/eth2_network.nim
index 3a339261b..f4ceab67c 100644
--- a/beacon_chain/networking/eth2_network.nim
+++ b/beacon_chain/networking/eth2_network.nim
@@ -372,17 +372,16 @@ proc updateScore*(peer: Peer, score: int) {.inline.} =
 proc join*(peer: Peer): Future[void] =
   var retFuture = newFuture[void]("peer.lifetime.join")
   let peerFut = peer.getFuture()
-  let alreadyFinished = peerFut.finished()
 
   proc continuation(udata: pointer) {.gcsafe.} =
     if not(retFuture.finished()):
       retFuture.complete()
 
   proc cancellation(udata: pointer) {.gcsafe.} =
-    if not(alreadyFinished):
+    if not(isNil(peerFut)):
       peerFut.removeCallback(continuation)
 
-  if alreadyFinished:
+  if peerFut.finished():
     # All the `peer.disconnectedFut` callbacks are already scheduled in current
     # `poll()` call, to avoid race we going to finish only in next `poll()`
     # call.
@@ -391,18 +390,21 @@ proc join*(peer: Peer): Future[void] =
     # `peer.disconnectedFut` is not yet finished, but we want to be scheduled
     # after all callbacks.
     peerFut.addCallback(continuation)
+    retFuture.cancelCallback = cancellation
 
-  return retFuture
+  retFuture
 
-proc notifyAndWait*(peer: Peer): Future[void] =
+proc notifyAndWait*(network: ETh2Node, peer: Peer): Future[void] =
   ## Notify all the waiters that peer life is finished and wait until all
   ## callbacks will be processed.
-  let joinFut = peer.join()
-  let fut = peer.disconnectedFut
+  let
+    joinFut = peer.join()
+    poolFut = network.peerPool.joinPeer(peer)
+    discFut = peer.disconnectedFut
   peer.connectionState = Disconnecting
-  fut.complete()
+  discFut.complete()
   peer.disconnectedFut = nil
-  joinFut
+  allFutures(joinFut, poolFut)
 
 proc calcThroughput(dur: Duration, value: uint64): float =
   let secs = float(chronos.seconds(1).nanoseconds)
@@ -1329,7 +1331,7 @@ proc onConnEvent(node: Eth2Node, peerId: PeerID, event: ConnEvent) {.async.} =
       node.addSeen(peerId, SeenTableTimeReconnect)
 
       if not(isNil(peer.disconnectedFut)):
-        await peer.notifyAndWait()
+        await node.notifyAndWait(peer)
       else:
         # TODO (cheatfate): This could be removed when bug will be fixed inside
         # `nim-libp2p`.
diff --git a/beacon_chain/networking/peer_pool.nim b/beacon_chain/networking/peer_pool.nim
index eef85ca30..c31c78f94 100644
--- a/beacon_chain/networking/peer_pool.nim
+++ b/beacon_chain/networking/peer_pool.nim
@@ -34,6 +34,7 @@ type
     peerType: PeerType
     flags: set[PeerFlags]
     index: int
+    future: Future[void]
 
   PeerIndex = object
     data: int
@@ -311,6 +312,7 @@ proc deletePeer*[A, B](pool: PeerPool[A, B], peer: A, force = false): bool =
           dec(pool.curOutPeersCount)
           dec(pool.acqOutPeersCount)
 
+        let fut = item[].future
         # Indicate that we have an empty space
         pool.fireNotFullEvent(item[])
         # Cleanup storage with default item, and removing key from hashtable.
@@ -318,6 +320,8 @@ proc deletePeer*[A, B](pool: PeerPool[A, B], peer: A, force = false): bool =
         pool.registry.del(key)
         pool.peerDeleted(peer)
         pool.peerCountChanged()
+        # Indicate that peer was deleted
+        fut.complete()
     else:
       if item[].peerType == PeerType.Incoming:
         # If peer is available, then its copy present in heapqueue, so we need
@@ -336,6 +340,7 @@ proc deletePeer*[A, B](pool: PeerPool[A, B], peer: A, force = false): bool =
             break
         dec(pool.curOutPeersCount)
 
+      let fut = item[].future
       # Indicate that we have an empty space
       pool.fireNotFullEvent(item[])
       # Cleanup storage with default item, and removing key from hashtable.
@@ -343,10 +348,44 @@ proc deletePeer*[A, B](pool: PeerPool[A, B], peer: A, force = false): bool =
       pool.registry.del(key)
       pool.peerDeleted(peer)
       pool.peerCountChanged()
+      # Indicate that peer was deleted
+      fut.complete()
     true
   else:
     false
 
+proc joinPeer*[A, B](pool: PeerPool[A, B], peer: A): Future[void] =
+  ## This procedure will only when peer ``peer`` finally leaves PeerPool
+  ## ``pool``.
+  mixin getKey
+  var retFuture = newFuture[void]("peerpool.joinPeer")
+  var future: Future[void]
+
+  proc continuation(udata: pointer) {.gcsafe.} =
+    if not(retFuture.finished()):
+      retFuture.complete()
+
+  proc cancellation(udata: pointer) {.gcsafe.} =
+    if not(isNil(future)):
+      future.removeCallback(continuation)
+
+  let key = getKey(peer)
+  if pool.registry.hasKey(key):
+    let pindex = pool.registry[key].data
+    var item = addr(pool.storage[pindex])
+    future = item[].future
+    # If peer is still in PeerPool, then item[].future should not be finished.
+    doAssert(not(future.finished()))
+    future.addCallback(continuation)
+    retFuture.cancelCallback = cancellation
+  else:
+    # If there no such peer in PeerPool anymore, its possible that
+    # PeerItem.future's callbacks is not yet processed, so we going to complete
+    # retFuture only in next `poll()` call.
+    callSoon(continuation, cast[pointer](retFuture))
+
+  retFuture
+
 proc addPeerImpl[A, B](pool: PeerPool[A, B], peer: A, peerKey: B,
                        peerType: PeerType) =
   mixin getFuture
@@ -354,7 +393,8 @@ proc addPeerImpl[A, B](pool: PeerPool[A, B], peer: A, peerKey: B,
     discard pool.deletePeer(peer)
 
   let item = PeerItem[A](data: peer, peerType: peerType,
-                         index: len(pool.storage))
+                         index: len(pool.storage),
+                         future: newFuture[void]("peerpool.peer"))
   pool.storage.add(item)
   var pitem = addr(pool.storage[^1])
   let pindex = PeerIndex(data: item.index, cmp: pool.cmp)