Merge pull request #9600 from lightningnetwork/yy-more-flakes

lntest+itest: document and fix more flakes

go.mod

@@ -43,7 +43,7 @@ require (
 	github.com/lightningnetwork/lnd/sqldb v1.0.7
 	github.com/lightningnetwork/lnd/ticker v1.1.1
 	github.com/lightningnetwork/lnd/tlv v1.3.0
-	github.com/lightningnetwork/lnd/tor v1.1.5
+	github.com/lightningnetwork/lnd/tor v1.1.6
 	github.com/ltcsuite/ltcd v0.0.0-20190101042124-f37f8bf35796
 	github.com/miekg/dns v1.1.43
 	github.com/pkg/errors v0.9.1

go.sum

@@ -470,8 +470,8 @@ github.com/lightningnetwork/lnd/ticker v1.1.1 h1:J/b6N2hibFtC7JLV77ULQp++QLtCwT6
 github.com/lightningnetwork/lnd/ticker v1.1.1/go.mod h1:waPTRAAcwtu7Ji3+3k+u/xH5GHovTsCoSVpho0KDvdA=
 github.com/lightningnetwork/lnd/tlv v1.3.0 h1:exS/KCPEgpOgviIttfiXAPaUqw2rHQrnUOpP7HPBPiY=
 github.com/lightningnetwork/lnd/tlv v1.3.0/go.mod h1:pJuiBj1ecr1WWLOtcZ+2+hu9Ey25aJWFIsjmAoPPnmc=
-github.com/lightningnetwork/lnd/tor v1.1.5 h1:kZ1Ab4EcPFCny3a179rZGKeDUlgYilEcfKx3tD5+L+k=
+github.com/lightningnetwork/lnd/tor v1.1.6 h1:WHUumk7WgU6BUFsqHuqszI9P6nfhMeIG+rjJBlVE6OE=
-github.com/lightningnetwork/lnd/tor v1.1.5/go.mod h1:qSRB8llhAK+a6kaTPWOLLXSZc6Hg8ZC0mq1sUQ/8JfI=
+github.com/lightningnetwork/lnd/tor v1.1.6/go.mod h1:qSRB8llhAK+a6kaTPWOLLXSZc6Hg8ZC0mq1sUQ/8JfI=
 github.com/ltcsuite/ltcd v0.0.0-20190101042124-f37f8bf35796 h1:sjOGyegMIhvgfq5oaue6Td+hxZuf3tDC8lAPrFldqFw=
 github.com/ltcsuite/ltcd v0.0.0-20190101042124-f37f8bf35796/go.mod h1:3p7ZTf9V1sNPI5H8P3NkTFF4LuwMdPl2DodF60qAKqY=
 github.com/ltcsuite/ltcutil v0.0.0-20181217130922-17f3b04680b6/go.mod h1:8Vg/LTOO0KYa/vlHWJ6XZAevPQThGH5sufO0Hrou/lA=

itest/flakes.go

@@ -104,3 +104,60 @@ func flakeSkipPendingSweepsCheckDarwin(ht *lntest.HarnessTest,
 func isDarwin() bool {
 	return runtime.GOOS == "darwin"
 }
+
+// flakeInconsistentHTLCView documents a flake found that the `ListChannels` RPC
+// can give inaccurate HTLC states, which is found when we call
+// `AssertHTLCNotActive` after a commitment dance is finished. Suppose Carol is
+// settling an invoice with Bob, from Bob's PoV, a typical healthy settlement
+// flow goes like this:
+//
+//	[DBG] PEER brontide.go:2412: Peer([Carol]): Received UpdateFulfillHTLC
+//	[DBG] HSWC switch.go:1315: Closed completed SETTLE circuit for ...
+//	[INF] HSWC switch.go:3044: Forwarded HTLC...
+//	[DBG] PEER brontide.go:2412: Peer([Carol]): Received CommitSig
+//	[DBG] PEER brontide.go:2412: Peer([Carol]): Sending RevokeAndAck
+//	[DBG] PEER brontide.go:2412: Peer([Carol]): Sending CommitSig
+//	[DBG] PEER brontide.go:2412: Peer([Carol]): Received RevokeAndAck
+//	[DBG] HSWC link.go:3617: ChannelLink([ChanPoint: Bob=>Carol]): settle-fail-filter: count=1, filter=[0]
+//	[DBG] HSWC switch.go:3001: Circuit is closing for packet...
+//
+// Bob receives the preimage, closes the circuit, and exchanges commit sig and
+// revoke msgs with Carol. Once Bob receives the `CommitSig` from Carol, the
+// HTLC should be removed from his `LocalCommitment` via
+// `RevokeCurrentCommitment`.
+//
+// However, in the test where `AssertHTLCNotActive` is called, although the
+// above process is finished, the `ListChannels“ still finds the HTLC. Also note
+// that the RPC makes direct call to the channeldb without any locks, which
+// should be fine as the struct `OpenChannel.LocalCommitment` is passed by
+// value, although we need to double check.
+//
+// TODO(yy): In order to fix it, we should make the RPC share the same view of
+// our channel state machine. Instead of making DB queries, it should instead
+// use `lnwallet.LightningChannel` instead to stay consistent.
+//
+//nolint:ll
+func flakeInconsistentHTLCView() {
+	// Perform a sleep so the commiment dance can be finished before we call
+	// the ListChannels.
+	time.Sleep(2 * time.Second)
+}
+
+// flakePaymentStreamReturnEarly documents a flake found in the test which
+// relies on a given payment to be settled before testing other state changes.
+// The issue comes from the payment stream created from the RPC `SendPaymentV2`
+// gives premature settled event for a given payment, which is found in,
+//   - if we force close the channel immediately, we may get an error because
+//     the commitment dance is not finished.
+//   - if we subscribe HTLC events immediately, we may get extra events, which
+//     is also related to the above unfinished commitment dance.
+//
+// TODO(yy): Make sure we only mark the payment being settled once the
+// commitment dance is finished. In addition, we should also fix the exit hop
+// logic in the invoice settlement flow to make sure the invoice is only marked
+// as settled after the commitment dance is finished.
+func flakePaymentStreamReturnEarly() {
+	// Sleep 2 seconds so the pending HTLCs will be removed from the
+	// commitment.
+	time.Sleep(2 * time.Second)
+}

itest/lnd_channel_backup_test.go

@@ -1224,13 +1224,7 @@ func testDataLossProtection(ht *lntest.HarnessTest) {
 		// payment hashes.
 		ht.CompletePaymentRequests(carol, payReqs[numInvoices/2:])
 
-		// TODO(yy): remove the sleep once the following bug is fixed.
+		flakePaymentStreamReturnEarly()
-		//
-		// While the payment is reported as settled, the commitment
-		// dance may not be finished, which leaves several HTLCs in the
-		// commitment. Later on, when Carol force closes this channel,
-		// she would have HTLCs there and the test won't pass.
-		time.Sleep(2 * time.Second)
 
 		// Now we shutdown Dave, copying over the its temporary
 		// database state which has the *prior* channel state over his

itest/lnd_coop_close_with_htlcs_test.go

@@ -222,11 +222,14 @@ func coopCloseWithHTLCsWithRestart(ht *lntest.HarnessTest, alice,
 	ht.AssertChannelInactive(bob, chanPoint)
 	ht.AssertChannelInactive(alice, chanPoint)
 
-	// Now restart Alice and Bob.
+	// Shutdown both Alice and Bob.
-	ht.RestartNode(alice)
+	restartAlice := ht.SuspendNode(alice)
-	ht.RestartNode(bob)
+	restartBob := ht.SuspendNode(bob)
 
-	ht.AssertConnected(alice, bob)
+	// Once shutdown, restart and connect them.
+	require.NoError(ht, restartAlice())
+	require.NoError(ht, restartBob())
+	ht.EnsureConnected(alice, bob)
 
 	// Show that both nodes still see the channel as waiting for close after
 	// the restart.

itest/lnd_open_channel_test.go

@@ -837,6 +837,24 @@ func testFundingExpiryBlocksOnPending(ht *lntest.HarnessTest) {
 	param := lntest.OpenChannelParams{Amt: 100000}
 	ht.OpenChannelAssertPending(alice, bob, param)
 
+	// assertExpiry is a helper closure to assert the FundingExpiryBlocks
+	// has been updated to the expected value.
+	assertExpiry := func(hn *node.HarnessNode, expected int32) {
+		err := wait.NoError(func() error {
+			pending := ht.AssertNumPendingOpenChannels(hn, 1)
+			expiry := pending[0].FundingExpiryBlocks
+
+			// Exit early if matched.
+			if expected == expiry {
+				return nil
+			}
+
+			return fmt.Errorf("want %v, got %v", expected, expiry)
+		}, wait.DefaultTimeout)
+		require.NoErrorf(ht, err, "%s: assert FundingExpiryBlocks "+
+			"timeout", hn.Name())
+	}
+
 	// At this point, the channel's funding transaction will have been
 	// broadcast, but not confirmed. Alice and Bob's nodes should reflect
 	// this when queried via RPC. FundingExpiryBlock should decrease
@@ -846,15 +864,16 @@ func testFundingExpiryBlocksOnPending(ht *lntest.HarnessTest) {
 	const numEmptyBlocks = 3
 	for i := int32(0); i < numEmptyBlocks; i++ {
 		expectedVal := lncfg.DefaultMaxWaitNumBlocksFundingConf - i
-		pending := ht.AssertNumPendingOpenChannels(alice, 1)
+
-		require.Equal(ht, expectedVal, pending[0].FundingExpiryBlocks)
+		// Assert Alice and Bob have updated the FundingExpiryBlocks.
-		pending = ht.AssertNumPendingOpenChannels(bob, 1)
+		assertExpiry(alice, expectedVal)
-		require.Equal(ht, expectedVal, pending[0].FundingExpiryBlocks)
+		assertExpiry(bob, expectedVal)
+
 		ht.MineEmptyBlocks(1)
 	}
 
-	// Mine 1 block to confirm the funding transaction, and then close the
+	// Mine 1 block to confirm the funding transaction so clean up the
-	// channel.
+	// mempool.
 	ht.MineBlocksAndAssertNumTxes(1, 1)
 }
 

itest/lnd_route_blinding_test.go

@@ -667,6 +667,10 @@ func testRelayingBlindedError(ht *lntest.HarnessTest) {
 	// so that she can't forward the payment to Dave.
 	testCase.drainCarolLiquidity(false)
 
+	// NOTE: The above draining requires Carol to send a payment, which may
+	// create extra events, causing the `AssertHtlcEvents` to fail below.
+	flakePaymentStreamReturnEarly()
+
 	// Subscribe to Carol's HTLC events so that we can observe the payment
 	// coming in.
 	carolEvents := testCase.carol.RPC.SubscribeHtlcEvents()
@@ -815,6 +819,7 @@ func testErrorHandlingOnChainFailure(ht *lntest.HarnessTest) {
 	ht.MineBlocksAndAssertNumTxes(1, 1)
 
 	// Assert that the HTLC has cleared.
+	flakeInconsistentHTLCView()
 	ht.AssertHTLCNotActive(bob, testCase.channels[0], hash[:])
 	ht.AssertHTLCNotActive(alice, testCase.channels[0], hash[:])
 

itest/lnd_sweep_test.go

@@ -500,6 +500,7 @@ func testSweepCPFPAnchorIncomingTimeout(ht *lntest.HarnessTest) {
 	carol.RPC.SettleInvoice(preimage[:])
 
 	// Bob should have settled his outgoing HTLC with Carol.
+	flakeInconsistentHTLCView()
 	ht.AssertHTLCNotActive(bob, bcChanPoint, payHash[:])
 
 	// We'll now mine enough blocks to trigger Bob to force close channel
@@ -851,6 +852,7 @@ func testSweepHTLCs(ht *lntest.HarnessTest) {
 	carol.RPC.SettleInvoice(preimageSettled[:])
 
 	// Bob should have settled his outgoing HTLC with Carol.
+	flakeInconsistentHTLCView()
 	ht.AssertHTLCNotActive(bob, bcChanPoint, payHashSettled[:])
 
 	// Let Carol go offline so we can focus on testing Bob's sweeping

itest/lnd_wipe_fwdpkgs_test.go

@@ -1,8 +1,6 @@
 package itest
 
 import (
-	"time"
-
 	"github.com/lightningnetwork/lnd/chainreg"
 	"github.com/lightningnetwork/lnd/lnrpc"
 	"github.com/lightningnetwork/lnd/lntest"
@@ -49,13 +47,7 @@ func testWipeForwardingPackages(ht *lntest.HarnessTest) {
 		ht.CompletePaymentRequests(alice, []string{resp.PaymentRequest})
 	}
 
-	// TODO(yy): remove the sleep once the following bug is fixed.
+	flakePaymentStreamReturnEarly()
-	// When the invoice is reported settled, the commitment dance is not
-	// yet finished, which can cause an error when closing the channel,
-	// saying there's active HTLCs. We need to investigate this issue and
-	// reverse the order to, first finish the commitment dance, then report
-	// the invoice as settled.
-	time.Sleep(2 * time.Second)
 
 	// Firstly, Bob force closes the channel.
 	ht.CloseChannelAssertPending(bob, chanPointAB, true)

lntest/harness_assertion.go

@@ -401,7 +401,7 @@ func (h *HarnessTest) assertChannelStatus(hn *node.HarnessNode,
 	}, DefaultTimeout)
 
 	require.NoErrorf(h, err, "%s: timeout checking for channel point: %v",
-		hn.Name(), cp)
+		hn.Name(), h.OutPointFromChannelPoint(cp))
 
 	return channel
 }
@@ -2329,7 +2329,8 @@ func (h *HarnessTest) AssertHtlcEvents(client rpc.HtlcEventsClient,
 		event := h.ReceiveHtlcEvent(client)
 
 		require.Containsf(h, eventTypes, event.EventType,
-			"wrong event type, got %v", userType, event.EventType)
+			"wrong event type, want %v, got %v", userType,
+			event.EventType)
 
 		events = append(events, event)