注:当前Fabric版本 v2.x
先看下命令行的步骤。
以两个组织Org1和Org2加入新创建的通道mychannel为例,给出命令行的步骤如下:
# 1.1 生成创建通道的交易:mychannel.tx
configtxgen -profile TwoOrgsChannel -outputCreateChannelTx ./channel-artifacts/mychannel.tx -channelID mychannel
# 1.2 为组织Org1生成锚节点更新的交易:Org1MSPanchors.tx
configtxgen -profile TwoOrgsChannel -outputAnchorPeersUpdate ./channel-artifacts/Org1MSPanchors.tx -channelID mychannel -asOrg Org1MSP
# 1.3 为组织Org2生成锚节点更新的交易:Org2MSPanchors.tx
configtxgen -profile TwoOrgsChannel -outputAnchorPeersUpdate ./channel-artifacts/Org2MSPanchors.tx -channelID mychannel -asOrg Org2MSP
# 2. 组织Org1:发送创建通道的交易到Orderer节点,并获取该通道的创世块:mychannel.block
peer channel create -o localhost:7050 -c mychannel --ordererTLSHostnameOverride orderer.example.com -f ./channel-artifacts/mychannel.tx --outputBlock ./channel-artifacts/mychannel.block --tls $CORE_PEER_TLS_ENABLED --cafile $ORDERER_CA
# 3.1 组织Org1:peers加入目标通道
peer channel join -b ./channel-artifacts/mychannel.block
# 3.2 组织Org2:peers加入目标通道
peer channel join -b ./channel-artifacts/mychannel.block
# 4.1 组织Org1:更新锚节点
peer channel update -o localhost:7050 --ordererTLSHostnameOverride orderer.example.com -c mychannel -f ./channel-artifacts/${CORE_PEER_LOCALMSPID}anchors.tx --tls $CORE_PEER_TLS_ENABLED --cafile $ORDERER_CA
# 4.2 组织Org2:更新锚节点
peer channel update -o localhost:7050 --ordererTLSHostnameOverride orderer.example.com -c mychannel -f ./channel-artifacts/${CORE_PEER_LOCALMSPID}anchors.tx --tls $CORE_PEER_TLS_ENABLED --cafile $ORDERER_CA
可以看到,总共需要4大步:
peer channel create
,把创建通道的交易发送给Orderer并获得该通道的创世块peer channel join
,加入目标通道peer channel update
,更新锚节点下面简单分析下这三个命令内部做了哪些事:
peer channel create
入口在internal/peer/channel/create.go
func executeCreate(cf *ChannelCmdFactory) error {
// 把创建通道的交易发送给Orderer
err := sendCreateChainTransaction(cf)
if err != nil {
return err
}
// 获得该通道的创世块
block, err := getGenesisBlock(cf)
if err != nil {
return err
}
b, err := proto.Marshal(block)
if err != nil {
return err
}
file := channelID + ".block"
if outputBlock != common.UndefinedParamValue {
file = outputBlock
}
// 保存到本地
err = ioutil.WriteFile(file, b, 0644)
if err != nil {
return err
}
return nil
}
executeCreate
主要做了三件事:
sendCreateChainTransaction
:把创建通道的交易发送给OrderergetGenesisBlock
:获得该通道的创世块ioutil.WriteFile
:把该通道的创世块保存到本地func sendCreateChainTransaction(cf *ChannelCmdFactory) error {
var err error
var chCrtEnv *cb.Envelope
if channelTxFile != "" {
// 根据指定的交易,即mychannel.tx ,把创建通道这个交易打包成信封Envelope
if chCrtEnv, err = createChannelFromConfigTx(channelTxFile); err != nil {
return err
}
} else {
if chCrtEnv, err = createChannelFromDefaults(cf); err != nil {
return err
}
}
// 做一些检查,并对此信封签名
if chCrtEnv, err = sanityCheckAndSignConfigTx(chCrtEnv, cf.Signer); err != nil {
return err
}
var broadcastClient common.BroadcastClient
broadcastClient, err = cf.BroadcastFactory()
if err != nil {
return errors.WithMessage(err, "error getting broadcast client")
}
defer broadcastClient.Close()
// 把此信封发送给Orderer
err = broadcastClient.Send(chCrtEnv)
return err
}
sendCreateChainTransaction
主要做了三件事:
createChannelFromConfigTx
:根据指定的交易,即mychannel.tx ,把创建通道这个交易打包成信封EnvelopesanityCheckAndSignConfigTx
:做一些检查,并对此信封签名broadcastClient.Send(chCrtEnv)
:把此信封发送给Ordererfunc getGenesisBlock(cf *ChannelCmdFactory) (*cb.Block, error) {
timer := time.NewTimer(timeout)
defer timer.Stop()
for {
select {
case <-timer.C:
cf.DeliverClient.Close()
return nil, errors.New("timeout waiting for channel creation")
default:
// 周期性的从Orderer获取新建通道的创世块
if block, err := cf.DeliverClient.GetSpecifiedBlock(0); err != nil {
cf.DeliverClient.Close()
cf, err = InitCmdFactory(EndorserNotRequired, PeerDeliverNotRequired, OrdererRequired)
if err != nil {
return nil, errors.WithMessage(err, "failed connecting")
}
time.Sleep(200 * time.Millisecond)
} else {
cf.DeliverClient.Close()
return block, nil
}
}
}
}
getGenesisBlock
就做了一件事:周期性的从Orderer获取新建通道的创世块,即索引为0的块
至此,peer channel create
本地侧的逻辑分析完了,下面看下Orderer侧的逻辑,即从本地通过peer命令把创建通道的交易发给Orderer到收到Orderer给出的新建通道的创世块的这段过程中Orderer做了哪些工作。
Orderer接收信封并处理的逻辑在orderer/common/broadcast/broadcast.go
// Handle reads requests from a Broadcast stream, processes them, and returns the responses to the stream
func (bh *Handler) Handle(srv ab.AtomicBroadcast_BroadcastServer) error {
addr := util.ExtractRemoteAddress(srv.Context())
logger.Debugf("Starting new broadcast loop for %s", addr)
for {
// 接收信封消息
msg, err := srv.Recv()
if err == io.EOF {
logger.Debugf("Received EOF from %s, hangup", addr)
return nil
}
if err != nil {
logger.Warningf("Error reading from %s: %s", addr, err)
return err
}
// 处理消息
resp := bh.ProcessMessage(msg, addr)
// 返回处理结果
err = srv.Send(resp)
if resp.Status != cb.Status_SUCCESS {
return err
}
if err != nil {
logger.Warningf("Error sending to %s: %s", addr, err)
return err
}
}
}
Handle
就做了三件事:
Recv
:接收信封消息ProcessMessage
:处理消息Send
:返回处理结果重点看下ProcessMessage
中对消息的处理:
// ProcessMessage validates and enqueues a single message
func (bh *Handler) ProcessMessage(msg *cb.Envelope, addr string) (resp *ab.BroadcastResponse) {
...
// 获取接收到的消息的Header、判断是否为配置信息、获取消息处理器
chdr, isConfig, processor, err := bh.SupportRegistrar.BroadcastChannelSupport(msg)
if chdr != nil {
tracker.ChannelID = chdr.ChannelId
tracker.TxType = cb.HeaderType(chdr.Type).String()
}
if err != nil {
logger.Warningf("[channel: %s] Could not get message processor for serving %s: %s", tracker.ChannelID, addr, err)
return &ab.BroadcastResponse{Status: cb.Status_BAD_REQUEST, Info: err.Error()}
}
if !isConfig {
...
} else { // isConfig
logger.Debugf("[channel: %s] Broadcast is processing config update message from %s", chdr.ChannelId, addr)
// 对于配置消息,使用处理器处理配置变更消息
config, configSeq, err := processor.ProcessConfigUpdateMsg(msg)
if err != nil {
logger.Warningf("[channel: %s] Rejecting broadcast of config message from %s because of error: %s", chdr.ChannelId, addr, err)
return &ab.BroadcastResponse{Status: ClassifyError(err), Info: err.Error()}
}
tracker.EndValidate()
tracker.BeginEnqueue()
if err = processor.WaitReady(); err != nil {
logger.Warningf("[channel: %s] Rejecting broadcast of message from %s with SERVICE_UNAVAILABLE: rejected by Consenter: %s", chdr.ChannelId, addr, err)
return &ab.BroadcastResponse{Status: cb.Status_SERVICE_UNAVAILABLE, Info: err.Error()}
}
// 把配置变更的交易进行全网共识
err = processor.Configure(config, configSeq)
if err != nil {
logger.Warningf("[channel: %s] Rejecting broadcast of config message from %s with SERVICE_UNAVAILABLE: rejected by Configure: %s", chdr.ChannelId, addr, err)
return &ab.BroadcastResponse{Status: cb.Status_SERVICE_UNAVAILABLE, Info: err.Error()}
}
}
logger.Debugf("[channel: %s] Broadcast has successfully enqueued message of type %s from %s", chdr.ChannelId, cb.HeaderType_name[chdr.Type], addr)
// 返回处理结果
return &ab.BroadcastResponse{Status: cb.Status_SUCCESS}
}
ProcessMessage
主要做了四件事:
BroadcastChannelSupport
:获取接收到的消息的Header、判断是否为配置信息、获取消息处理器processor.ProcessConfigUpdateMsg
:对于配置消息,使用处理器处理配置变更消息processor.Configure
:把配置变更的交易进行全网共识BroadcastResponse
:返回处理结果peer channel create
主要流程如下:
TODO:
TODO: