Merge pull request #261 from ipfs/fix/prune-session-peers

Prune peers that send too many consecutive DONT_HAVEs

internal/session/sessionwantsender.go

@@ -9,8 +9,13 @@ import (
 	peer "github.com/libp2p/go-libp2p-core/peer"
 )
 
-// Maximum number of changes to accept before blocking
-const changesBufferSize = 128
+const (
+	// Maximum number of changes to accept before blocking
+	changesBufferSize = 128
+	// If the session receives this many DONT_HAVEs in a row from a peer,
+	// it prunes the peer from the session
+	peerDontHaveLimit = 16
+)
 
 // BlockPresence indicates whether a peer has a block.
 // Note that the order is important, we decide which peer to send a want to
@@ -76,13 +81,14 @@ type sessionWantSender struct {
 	changes chan change
 	// Information about each want indexed by CID
 	wants map[cid.Cid]*wantInfo
+	// Keeps track of how many consecutive DONT_HAVEs a peer has sent
+	peerConsecutiveDontHaves map[peer.ID]int
 	// Tracks which peers we have send want-block to
 	swbt *sentWantBlocksTracker
 	// Maintains a list of peers and whether they are connected
 	peerAvlMgr *peerAvailabilityManager
 	// Tracks the number of blocks each peer sent us
 	peerRspTrkr *peerResponseTracker
-
 	// Sends wants to peers
 	pm PeerManager
 	// Keeps track of which peer has / doesn't have a block
@@ -101,6 +107,7 @@ func newSessionWantSender(ctx context.Context, sid uint64, pm PeerManager, bpm *
 		sessionID:                sid,
 		changes:                  make(chan change, changesBufferSize),
 		wants:                    make(map[cid.Cid]*wantInfo),
+		peerConsecutiveDontHaves: make(map[peer.ID]int),
 		swbt:                     newSentWantBlocksTracker(),
 		peerAvlMgr:               newPeerAvailabilityManager(),
 		peerRspTrkr:              newPeerResponseTracker(),
@@ -258,6 +265,9 @@ func (spm *sessionWantSender) processAvailability(availability map[peer.ID]bool)
 				if isNowAvailable {
 					newlyAvailable = append(newlyAvailable, p)
 				}
+				// Reset the count of consecutive DONT_HAVEs received from the
+				// peer
+				delete(spm.peerConsecutiveDontHaves, p)
 			}
 		}
 	}
@@ -265,6 +275,12 @@ func (spm *sessionWantSender) processAvailability(availability map[peer.ID]bool)
 	return newlyAvailable
 }
 
+// isAvailable indicates whether the peer is available and whether
+// it's been tracked by the Session (used by the tests)
+func (spm *sessionWantSender) isAvailable(p peer.ID) (bool, bool) {
+	return spm.peerAvlMgr.isAvailable(p)
+}
+
 // trackWant creates a new entry in the map of CID -> want info
 func (spm *sessionWantSender) trackWant(c cid.Cid) {
 	// fmt.Printf("trackWant %s\n", lu.C(c))
@@ -285,6 +301,7 @@ func (spm *sessionWantSender) trackWant(c cid.Cid) {
 
 // processUpdates processes incoming blocks and HAVE / DONT_HAVEs
 func (spm *sessionWantSender) processUpdates(updates []update) {
+	prunePeers := make(map[peer.ID]struct{})
 	dontHaves := cid.NewSet()
 	for _, upd := range updates {
 		// TODO: If there is a timeout for the want from the peer, remove want.sentTo
@@ -308,12 +325,20 @@ func (spm *sessionWantSender) processUpdates(updates []update) {
 					spm.setWantSentTo(c, "")
 				}
 			}
+
+			// Track the number of consecutive DONT_HAVEs each peer receives
+			if spm.peerConsecutiveDontHaves[upd.from] == peerDontHaveLimit {
+				prunePeers[upd.from] = struct{}{}
+			} else {
+				spm.peerConsecutiveDontHaves[upd.from]++
+			}
 		}
 
 		// For each HAVE
 		for _, c := range upd.haves {
 			// Update the block presence for the peer
 			spm.updateWantBlockPresence(c, upd.from)
+			delete(spm.peerConsecutiveDontHaves, upd.from)
 		}
 
 		// For each received block
@@ -325,6 +350,7 @@ func (spm *sessionWantSender) processUpdates(updates []update) {
 				// us the block
 				spm.peerRspTrkr.receivedBlockFrom(upd.from)
 			}
+			delete(spm.peerConsecutiveDontHaves, upd.from)
 		}
 	}
 
@@ -337,6 +363,12 @@ func (spm *sessionWantSender) processUpdates(updates []update) {
 			spm.onPeersExhausted(newlyExhausted)
 		}
 	}
+
+	// If any peers have sent us too many consecutive DONT_HAVEs, remove them
+	// from the session
+	for p := range prunePeers {
+		spm.SignalAvailability(p, false)
+	}
 }
 
 // convenience structs for passing around want-blocks and want-haves for a peer

internal/session/sessionwantsender_test.go

@@ -346,3 +346,198 @@ func TestPeersExhausted(t *testing.T) {
 		t.Fatal("Wrong keys")
 	}
 }
+
+func TestConsecutiveDontHaveLimit(t *testing.T) {
+	cids := testutil.GenerateCids(peerDontHaveLimit + 10)
+	p := testutil.GeneratePeers(1)[0]
+	sid := uint64(1)
+	pm := newMockPeerManager()
+	bpm := bsbpm.New()
+	onSend := func(peer.ID, []cid.Cid, []cid.Cid) {}
+	onPeersExhausted := func([]cid.Cid) {}
+	spm := newSessionWantSender(context.Background(), sid, pm, bpm, onSend, onPeersExhausted)
+
+	go spm.Run()
+
+	// Add all cids as wants
+	spm.Add(cids)
+
+	// Receive a HAVE from peer (adds it to the session)
+	bpm.ReceiveFrom(p, cids[:1], []cid.Cid{})
+	spm.Update(p, []cid.Cid{}, cids[:1], []cid.Cid{}, true)
+
+	// Wait for processing to complete
+	time.Sleep(5 * time.Millisecond)
+
+	// Peer should be available
+	if avail, ok := spm.isAvailable(p); !ok || !avail {
+		t.Fatal("Expected peer to be available")
+	}
+
+	// Receive DONT_HAVEs from peer that do not exceed limit
+	for _, c := range cids[1:peerDontHaveLimit] {
+		bpm.ReceiveFrom(p, []cid.Cid{}, []cid.Cid{c})
+		spm.Update(p, []cid.Cid{}, []cid.Cid{}, []cid.Cid{c}, false)
+	}
+
+	// Wait for processing to complete
+	time.Sleep(5 * time.Millisecond)
+
+	// Peer should be available
+	if avail, ok := spm.isAvailable(p); !ok || !avail {
+		t.Fatal("Expected peer to be available")
+	}
+
+	// Receive DONT_HAVEs from peer that exceed limit
+	for _, c := range cids[peerDontHaveLimit:] {
+		bpm.ReceiveFrom(p, []cid.Cid{}, []cid.Cid{c})
+		spm.Update(p, []cid.Cid{}, []cid.Cid{}, []cid.Cid{c}, false)
+	}
+
+	// Wait for processing to complete
+	time.Sleep(5 * time.Millisecond)
+
+	// Session should remove peer
+	if avail, _ := spm.isAvailable(p); avail {
+		t.Fatal("Expected peer not to be available")
+	}
+}
+
+func TestConsecutiveDontHaveLimitInterrupted(t *testing.T) {
+	cids := testutil.GenerateCids(peerDontHaveLimit + 10)
+	p := testutil.GeneratePeers(1)[0]
+	sid := uint64(1)
+	pm := newMockPeerManager()
+	bpm := bsbpm.New()
+	onSend := func(peer.ID, []cid.Cid, []cid.Cid) {}
+	onPeersExhausted := func([]cid.Cid) {}
+	spm := newSessionWantSender(context.Background(), sid, pm, bpm, onSend, onPeersExhausted)
+
+	go spm.Run()
+
+	// Add all cids as wants
+	spm.Add(cids)
+
+	// Receive a HAVE from peer (adds it to the session)
+	bpm.ReceiveFrom(p, cids[:1], []cid.Cid{})
+	spm.Update(p, []cid.Cid{}, cids[:1], []cid.Cid{}, true)
+
+	// Wait for processing to complete
+	time.Sleep(5 * time.Millisecond)
+
+	// Peer should be available
+	if avail, ok := spm.isAvailable(p); !ok || !avail {
+		t.Fatal("Expected peer to be available")
+	}
+
+	// Receive DONT_HAVE then HAVE then DONT_HAVE from peer,
+	// where consecutive DONT_HAVEs would have exceeded limit
+	// (but they are not consecutive)
+	for _, c := range cids[1:peerDontHaveLimit] {
+		// DONT_HAVEs
+		bpm.ReceiveFrom(p, []cid.Cid{}, []cid.Cid{c})
+		spm.Update(p, []cid.Cid{}, []cid.Cid{}, []cid.Cid{c}, false)
+	}
+	for _, c := range cids[peerDontHaveLimit : peerDontHaveLimit+1] {
+		// HAVEs
+		bpm.ReceiveFrom(p, []cid.Cid{c}, []cid.Cid{})
+		spm.Update(p, []cid.Cid{}, []cid.Cid{c}, []cid.Cid{}, false)
+	}
+	for _, c := range cids[peerDontHaveLimit+1:] {
+		// DONT_HAVEs
+		bpm.ReceiveFrom(p, []cid.Cid{}, []cid.Cid{c})
+		spm.Update(p, []cid.Cid{}, []cid.Cid{}, []cid.Cid{c}, false)
+	}
+
+	// Wait for processing to complete
+	time.Sleep(5 * time.Millisecond)
+
+	// Peer should be available
+	if avail, ok := spm.isAvailable(p); !ok || !avail {
+		t.Fatal("Expected peer to be available")
+	}
+}
+
+func TestConsecutiveDontHaveReinstateAfterRemoval(t *testing.T) {
+	cids := testutil.GenerateCids(peerDontHaveLimit + 10)
+	p := testutil.GeneratePeers(1)[0]
+	sid := uint64(1)
+	pm := newMockPeerManager()
+	bpm := bsbpm.New()
+	onSend := func(peer.ID, []cid.Cid, []cid.Cid) {}
+	onPeersExhausted := func([]cid.Cid) {}
+	spm := newSessionWantSender(context.Background(), sid, pm, bpm, onSend, onPeersExhausted)
+
+	go spm.Run()
+
+	// Add all cids as wants
+	spm.Add(cids)
+
+	// Receive a HAVE from peer (adds it to the session)
+	bpm.ReceiveFrom(p, cids[:1], []cid.Cid{})
+	spm.Update(p, []cid.Cid{}, cids[:1], []cid.Cid{}, true)
+
+	// Wait for processing to complete
+	time.Sleep(5 * time.Millisecond)
+
+	// Peer should be available
+	if avail, ok := spm.isAvailable(p); !ok || !avail {
+		t.Fatal("Expected peer to be available")
+	}
+
+	// Receive DONT_HAVEs from peer that exceed limit
+	for _, c := range cids[1 : peerDontHaveLimit+2] {
+		bpm.ReceiveFrom(p, []cid.Cid{}, []cid.Cid{c})
+		spm.Update(p, []cid.Cid{}, []cid.Cid{}, []cid.Cid{c}, false)
+	}
+
+	// Wait for processing to complete
+	time.Sleep(5 * time.Millisecond)
+
+	// Session should remove peer
+	if avail, _ := spm.isAvailable(p); avail {
+		t.Fatal("Expected peer not to be available")
+	}
+
+	// Receive a HAVE from peer (adds it back into the session)
+	bpm.ReceiveFrom(p, cids[:1], []cid.Cid{})
+	spm.Update(p, []cid.Cid{}, cids[:1], []cid.Cid{}, true)
+
+	// Wait for processing to complete
+	time.Sleep(5 * time.Millisecond)
+
+	// Peer should be available
+	if avail, ok := spm.isAvailable(p); !ok || !avail {
+		t.Fatal("Expected peer to be available")
+	}
+
+	cids2 := testutil.GenerateCids(peerDontHaveLimit + 10)
+
+	// Receive DONT_HAVEs from peer that don't exceed limit
+	for _, c := range cids2[1:peerDontHaveLimit] {
+		bpm.ReceiveFrom(p, []cid.Cid{}, []cid.Cid{c})
+		spm.Update(p, []cid.Cid{}, []cid.Cid{}, []cid.Cid{c}, false)
+	}
+
+	// Wait for processing to complete
+	time.Sleep(5 * time.Millisecond)
+
+	// Peer should be available
+	if avail, ok := spm.isAvailable(p); !ok || !avail {
+		t.Fatal("Expected peer to be available")
+	}
+
+	// Receive DONT_HAVEs from peer that exceed limit
+	for _, c := range cids2[peerDontHaveLimit:] {
+		bpm.ReceiveFrom(p, []cid.Cid{}, []cid.Cid{c})
+		spm.Update(p, []cid.Cid{}, []cid.Cid{}, []cid.Cid{c}, false)
+	}
+
+	// Wait for processing to complete
+	time.Sleep(5 * time.Millisecond)
+
+	// Session should remove peer
+	if avail, _ := spm.isAvailable(p); avail {
+		t.Fatal("Expected peer not to be available")
+	}
+}