Merge pull request #5281 from guggero/router-chain-sync

itest flakes: fix multiple issues around router subsystem being out of sync

lntest/harness.go

@@ -177,9 +177,16 @@ func (n *NetworkHarness) SetUp(testCase string, lndArgs []string) error {
 	default:
 	}
 
+	// First, make a connection between the two nodes. This will wait until
+	// both nodes are fully started since the Connect RPC is guarded behind
+	// the server.Started() flag that waits for all subsystems to be ready.
+	ctxb := context.Background()
+	if err := n.ConnectNodes(ctxb, n.Alice, n.Bob); err != nil {
+		return err
+	}
+
 	// Load up the wallets of the seeder nodes with 10 outputs of 1 BTC
 	// each.
-	ctxb := context.Background()
 	addrReq := &lnrpc.NewAddressRequest{
 		Type: lnrpc.AddressType_WITNESS_PUBKEY_HASH,
 	}
@@ -216,8 +223,13 @@ func (n *NetworkHarness) SetUp(testCase string, lndArgs []string) error {
 		return err
 	}
 
-	// Finally, make a connection between both of the nodes.
+	// Now we want to wait for the nodes to catch up.
-	if err := n.ConnectNodes(ctxb, n.Alice, n.Bob); err != nil {
+	ctxt, cancel := context.WithTimeout(ctxb, DefaultTimeout)
+	defer cancel()
+	if err := n.Alice.WaitForBlockchainSync(ctxt); err != nil {
+		return err
+	}
+	if err := n.Bob.WaitForBlockchainSync(ctxt); err != nil {
 		return err
 	}
 

lntest/itest/lnd_test.go

@@ -11787,7 +11787,15 @@ func testSwitchOfflineDelivery(net *lntest.NetworkHarness, t *harnessTest) {
 			err)
 	}
 
-	time.Sleep(time.Millisecond * 50)
+	// Make sure all nodes are fully synced before we continue.
+	ctxt, cancel := context.WithTimeout(ctxb, defaultTimeout)
+	defer cancel()
+	for _, node := range nodes {
+		err := node.WaitForBlockchainSync(ctxt)
+		if err != nil {
+			t.Fatalf("unable to wait for sync: %v", err)
+		}
+	}
 
 	// Using Carol as the source, pay to the 5 invoices from Bob created
 	// above.
@@ -11856,6 +11864,16 @@ func testSwitchOfflineDelivery(net *lntest.NetworkHarness, t *harnessTest) {
 		t.Fatalf("htlc mismatch: %v", predErr)
 	}
 
+	// Make sure all nodes are fully synced again.
+	ctxt, cancel = context.WithTimeout(ctxb, defaultTimeout)
+	defer cancel()
+	for _, node := range nodes {
+		err := node.WaitForBlockchainSync(ctxt)
+		if err != nil {
+			t.Fatalf("unable to wait for sync: %v", err)
+		}
+	}
+
 	// Now that the settles have reached Dave, reconnect him with Alice,
 	// allowing the settles to return to the sender.
 	ctxt, _ = context.WithTimeout(ctxb, defaultTimeout)

lntest/itest/log_error_whitelist.txt

@@ -155,6 +155,7 @@
 <time> [ERR] NTFN: unable to get hash from block with height 790
 <time> [ERR] NTFN: unable to get missed blocks: starting height <height> is greater than ending height <height>
 <time> [ERR] NTFN: Unable to rewind chain from height <height> to height <height>: unable to find blockhash for disconnected height=<height>: -1: Block number out of range
+<time> [ERR] NTFN: Unable to rewind chain from height <height> to height <height>: unable to find blockhash for disconnected height=<height>: -8: Block height out of range
 <time> [ERR] NTNF: unable to get hash from block with height <height>
 <time> [ERR] PEER: Allowed test error from <ip> (inbound): ReadMessage: unhandled command [sendaddrv2]
 <time> [ERR] PEER: resend failed: unable to fetch channel sync messages for peer <hex>@<ip>: unable to find closed channel summary

lntest/itest/log_substitutions.txt

@@ -10,7 +10,7 @@ s/HTLC ID = [[:digit:]]+/HTLC ID = <id>/g
 s/height=[[:digit:]]+/height=<height>/g
 s/collecting result for shard [[:digit:]]+/collecting result for shard <number>/g
 s/sending attempt [[:digit:]]+/sending attempt <number>/g
-s/Unable to rewind chain from height [[:digit:]]+ to height [[:digit:]]+/Unable to rewind chain from height <height> to height <height>/g
+s/Unable to rewind chain from height [[:digit:]]+ to height -?[[:digit:]]+/Unable to rewind chain from height <height> to height <height>/g
 s/NTFN: unable to get missed blocks: starting height [[:digit:]]+ is greater than ending height [[:digit:]]+/NTFN: unable to get missed blocks: starting height <height> is greater than ending height <height>/g
 s/BTCN: Broadcast attempt failed: rejected by <ip>: replacement transaction <hex> has an insufficient fee rate: needs more than [[:digit:]]+, has [[:digit:]]+/BTCN: Broadcast attempt failed: rejected by <ip>: replacement transaction <hex> has an insufficient fee rate: needs more than <amt>, has <amt>/g
 s/pid=[[:digit:]]+/pid=<pid>/g

routing/router.go

@@ -563,6 +563,14 @@ func (r *ChannelRouter) Start() error {
 			}
 		}
 
+		// The graph pruning might have taken a while and there could be
+		// new blocks available.
+		bestHash, bestHeight, err = r.cfg.Chain.GetBestBlock()
+		if err != nil {
+			return err
+		}
+		r.bestHeight = uint32(bestHeight)
+
 		// Before we begin normal operation of the router, we first need
 		// to synchronize the channel graph to the latest state of the
 		// UTXO set.
@@ -2479,6 +2487,13 @@ func (r *ChannelRouter) CurrentBlockHeight() (uint32, error) {
 	return uint32(height), err
 }
 
+// SyncedHeight returns the block height to which the router subsystem currently
+// is synced to. This can differ from the above chain height if the goroutine
+// responsible for processing the blocks isn't yet up to speed.
+func (r *ChannelRouter) SyncedHeight() uint32 {
+	return atomic.LoadUint32(&r.bestHeight)
+}
+
 // GetChannelByID return the channel by the channel id.
 //
 // NOTE: This method is part of the ChannelGraphSource interface.

rpcserver.go

@@ -2507,6 +2507,17 @@ func (r *rpcServer) GetInfo(ctx context.Context,
 			"with current best block in the main chain: %v", err)
 	}
 
+	// The router has a lot of work to do for each block. So it might be
+	// possible that it isn't yet up to date with the most recent block,
+	// even if the wallet is. This can happen in environments with high CPU
+	// load (such as parallel itests). Since the `synced_to_chain` flag in
+	// the response of this call is used by many wallets (and also our
+	// itests) to make sure everything's up to date, we add the router's
+	// state to it. So the flag will only toggle to true once the router was
+	// also able to catch up.
+	routerHeight := r.server.chanRouter.SyncedHeight()
+	isSynced = isSynced && uint32(bestHeight) == routerHeight
+
 	network := lncfg.NormalizeNetwork(r.cfg.ActiveNetParams.Name)
 	activeChains := make([]*lnrpc.Chain, r.cfg.registeredChains.NumActiveChains())
 	for i, chain := range r.cfg.registeredChains.ActiveChains() {