factor(*): return consensus status when vp node is un-ready

api/grpc/transaction.go

@@ -14,8 +14,9 @@ import (
 // SendTransaction handles transaction sent by the client.
 // If the transaction is valid, it will return the transaction hash.
 func (cbs *ChainBrokerService) SendTransaction(ctx context.Context, tx *pb.Transaction) (*pb.TransactionHashMsg, error) {
-	if !cbs.api.Broker().OrderReady() {
-		return nil, fmt.Errorf("the system is temporarily unavailable")
+	err := cbs.api.Broker().OrderReady()
+	if err != nil {
+		return nil, fmt.Errorf("the system is temporarily unavailable, err: %s", err.Error())
 	}
 
 	if err := cbs.checkTransaction(tx); err != nil {

config/bitxhub.toml

@@ -36,7 +36,7 @@ server_key_path = "certs/server.key"
   report_caller = false
   [log.module]
     p2p = "info"
-    consensus = "info"
+    consensus = "debug"
     executor = "info"
     router = "info"
     api = "info"
@@ -47,7 +47,7 @@ server_key_path = "certs/server.key"
   verify = true
 
 [order]
-  plugin = "plugins/raft.so"
+  plugin = "plugins/rbft.so"
 
 [executor]
   type = "serial"  # opensource version only supports serial type, commercial version supports serial and parallel types

config/order.toml

@@ -18,21 +18,25 @@ disable_proposal_forwarding = true # This prevents blocks from being accidentall
         fetch_timeout       = "3s"    # How long to wait before fetching missing transactions finished
 
 [rbft]        #RBFT configurations
-set_size         = 100   # How many transactions should the node broadcast at once
+set_size         = 25   # How many transactions should the node broadcast at once
 batch_size       = 500   # How many transactions should the primary pack before sending pre-prepare
 pool_size        = 50000 # How many transactions could the txPool stores in total
-vc_resend_limit  = 10    # How many times of same vc from self before a replica start recovery
 vc_period        = 0     # After how many checkpoint periods( Blocks = 10 * vcperiod ) the primary gets cycled automatically. ( Set 0 to disable )
+check_interval  = "3m"  # interval of the check loop
+tolerance_time   = "5m" # The max tolerance time duration (in seconds) of out-of-date
+batch_mem_limit  = false # Indicates whether limit batch mem size or not
+batch_max_mem    = 10000 # The max memory size of one batch
+
     [rbft.timeout]
-        nego_view        = "6s"  # How long to wait for N-f responses after send negotiate view
-        recovery        = "15s" # How long to wait before recovery finished(This is for release1.2)
-        first_request    = "30s" # How long to wait before first request should come
-        batch           = "0.5s"# Primary send a pre-prepare if there are pending requests, although batchsize isn't reached yet,
-        request         = "6s"  # How long may a request(transaction batch) take between reception and execution, must be greater than the batch timeout
-        validate        = "1s"  # How long may a validate (transaction batch) process will take by local Validation
-        null_request     = "6s"  # Primary send it to inform aliveness, must be greater than request timeout
-        viewchange      = "8s"  # How long may a view change take
-        resend_viewchange= "10s" # How long to wait for a view change quorum before resending (the same) view change
-        clean_viewchange = "60s" # How long to clean out-of-data view change message
-        update          = "4s"  # How long may a update-n take
-        set             = "0.1s" # Node broadcasts transactions if there are cached transactions, although set_size isn't reached yet,
+    	sync_state        = "3s"  # How long to wait quorum sync state response
+        sync_interval     = "1m"  # How long to restart sync state process
+        recovery          = "15s" # How long to wait before recovery finished(This is for release1.2)
+        first_request     = "30s" # How long to wait before first request should come
+        batch             = "0.5s"# Primary send a pre-prepare if there are pending requests, although batchsize isn't reached yet,
+        request           = "6s"  # How long may a request(transaction batch) take between reception and execution, must be greater than the batch timeout
+        null_request      = "9s"  # Primary send it to inform aliveness, must be greater than request timeout
+        viewchange        = "8s"  # How long may a view change take
+        resend_viewchange = "10s" # How long to wait for a view change quorum before resending (the same) view change
+        clean_viewchange  = "60s" # How long to clean out-of-data view change message
+        update            = "4s"  # How long may a update-n take
+        set               = "0.1s" # Node broadcasts transactions if there are cached transactions, although set_size isn't reached yet

internal/app/bitxhub.go

@@ -279,7 +279,8 @@ func (bxh *BitXHub) Stop() error {
 func (bxh *BitXHub) printLogo() {
 	for {
 		time.Sleep(100 * time.Millisecond)
-		if bxh.Order.Ready() {
+		err :=bxh.Order.Ready()
+		if err == nil {
 			bxh.logger.WithFields(logrus.Fields{
 				"plugin_path": bxh.repo.Config.Order.Plugin,
 			}).Info("Order is ready")

internal/coreapi/api/api.go

@@ -38,7 +38,7 @@ type BrokerAPI interface {
 	GetInterchainTxWrappers(pid string, begin, end uint64, ch chan<- *pb.InterchainTxWrappers) error
 
 	// OrderReady
-	OrderReady() bool
+	OrderReady() error
 
 	FetchSignsFromOtherPeers(content string, typ pb.GetMultiSignsRequest_Type) map[string][]byte
 	GetSign(content string, typ pb.GetMultiSignsRequest_Type) (string, []byte, error)

internal/coreapi/broker.go

@@ -109,7 +109,7 @@ func (b *BrokerAPI) RemovePier(pid string, isUnion bool) {
 	b.bxh.Router.RemovePier(pid, isUnion)
 }
 
-func (b *BrokerAPI) OrderReady() bool {
+func (b *BrokerAPI) OrderReady() error {
 	return b.bxh.Order.Ready()
 }
 

pkg/order/etcdraft/node.go

@@ -3,6 +3,7 @@ package etcdraft
 import (
 	"context"
 	"encoding/binary"
+	"errors"
 	"fmt"
 	"path/filepath"
 	"sync"
@@ -161,7 +162,8 @@ func (n *Node) Stop() {
 
 // Add the transaction into txpool and broadcast it to other nodes
 func (n *Node) Prepare(tx *pb.Transaction) error {
-	if !n.Ready() {
+	err := n.Ready()
+	if err != nil {
 		return nil
 	}
 	return n.mempool.RecvTransaction(tx)
@@ -248,8 +250,12 @@ func (n *Node) IsLeader() bool {
 	return n.leader == n.id
 }
 
-func (n *Node) Ready() bool {
-	return n.leader != 0
+func (n *Node) Ready() error {
+	hashLeader := n.leader != 0
+	if !hashLeader {
+		return errors.New("in leader election status")
+	}
+	return nil
 }
 
 // main work loop

pkg/order/order.go

@@ -25,7 +25,7 @@ type Order interface {
 	Step(ctx context.Context, msg []byte) error
 
 	// Ready means whether order has finished electing leader
-	Ready() bool
+	Ready() error
 
 	// ReportState means block was persisted and report it to the consensus engine
 	ReportState(height uint64, hash types.Hash)

pkg/order/solo/node.go

@@ -72,8 +72,8 @@ func (n *Node) Step(ctx context.Context, msg []byte) error {
 	return nil
 }
 
-func (n *Node) Ready() bool {
-	return true
+func (n *Node) Ready() error {
+	return nil
 }
 
 func (n *Node) ReportState(height uint64, hash types.Hash) {

tester/tester_test.go

@@ -21,10 +21,11 @@ func TestTester(t *testing.T) {
 	node4 := setupNode(t, "./test_data/config/node4")
 
 	for {
-		if node1.Broker().OrderReady() &&
-			node2.Broker().OrderReady() &&
-			node3.Broker().OrderReady() &&
-			node4.Broker().OrderReady() {
+		err1 := node1.Broker().OrderReady()
+		err2 := node2.Broker().OrderReady()
+		err3 := node3.Broker().OrderReady()
+		err4 := node4.Broker().OrderReady()
+		if err1 != nil && err2 != nil && err3 != nil && err4 != nil {
 			break
 		}