Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

FullNode concord-bft changes #3001

Open
wants to merge 3 commits into
base: master
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions bftengine/CMakeLists.txt
Original file line number Diff line number Diff line change
Expand Up @@ -11,6 +11,7 @@ set(corebft_source_files
src/bftengine/DebugStatistics.cpp
src/bftengine/SeqNumInfo.cpp
src/bftengine/ReadOnlyReplica.cpp
src/bftengine/FullNodeReplica.cpp
src/bftengine/ReplicaBase.cpp
src/bftengine/ReplicaForStateTransfer.cpp
src/bftengine/ReplicaImp.cpp
Expand Down
18 changes: 11 additions & 7 deletions bftengine/include/bcstatetransfer/SimpleBCStateTransfer.hpp
Original file line number Diff line number Diff line change
Expand Up @@ -150,8 +150,10 @@ struct Config {
uint16_t cVal = 0;
uint16_t numReplicas = 0; // number of consensus replicas
uint16_t numRoReplicas = 0;
uint16_t numFnReplicas = 0;
bool pedanticChecks = false;
bool isReadOnly = false;
bool isFullNode = false;
bandatarunkumar marked this conversation as resolved.
Show resolved Hide resolved

// sizes
uint32_t maxChunkSize = 0;
Expand Down Expand Up @@ -191,19 +193,21 @@ inline std::ostream &operator<<(std::ostream &os, const Config &c) {
c.cVal,
c.numReplicas,
c.numRoReplicas,
c.numFnReplicas,
c.pedanticChecks,
c.isReadOnly,
c.isFullNode,
c.maxChunkSize,
c.maxNumberOfChunksInBatch,
c.maxBlockSize,
c.maxPendingDataFromSourceReplica,
c.maxNumOfReservedPages,
c.sizeOfReservedPage,
c.gettingMissingBlocksSummaryWindowSize,
c.minPrePrepareMsgsForPrimaryAwareness,
c.fetchRangeSize);
c.gettingMissingBlocksSummaryWindowSize);
os << ",";
os << KVLOG(c.RVT_K,
os << KVLOG(c.minPrePrepareMsgsForPrimaryAwareness,
c.fetchRangeSize,
c.RVT_K,
c.refreshTimerMs,
c.checkpointSummariesRetransmissionTimeoutMs,
c.maxAcceptableMsgDelayMs,
Expand All @@ -216,9 +220,9 @@ inline std::ostream &operator<<(std::ostream &os, const Config &c) {
c.sourcePerformanceSnapshotFrequencySec,
c.runInSeparateThread,
c.enableReservedPages,
c.enableSourceBlocksPreFetch,
c.enableSourceSelectorPrimaryAwareness,
c.enableStoreRvbDataDuringCheckpointing);
c.enableSourceBlocksPreFetch);
os << ",";
os << KVLOG(c.enableSourceSelectorPrimaryAwareness, c.enableStoreRvbDataDuringCheckpointing);
return os;
}
// creates an instance of the state transfer module.
Expand Down
9 changes: 8 additions & 1 deletion bftengine/include/bftengine/ReplicaConfig.hpp
Original file line number Diff line number Diff line change
Expand Up @@ -46,8 +46,10 @@ class ReplicaConfig : public concord::serialize::SerializableFactory<ReplicaConf
}

CONFIG_PARAM(isReadOnly, bool, false, "Am I a read-only replica?");
CONFIG_PARAM(isFullNode, bool, false, "Am I a fullNode replica?");
CONFIG_PARAM(numReplicas, uint16_t, 0, "number of regular replicas");
CONFIG_PARAM(numRoReplicas, uint16_t, 0, "number of read-only replicas");
CONFIG_PARAM(numFnReplicas, uint16_t, 0, "number of full-node replicas");
CONFIG_PARAM(fVal, uint16_t, 0, "F value - max number of faulty/malicious replicas. fVal >= 1");
CONFIG_PARAM(cVal, uint16_t, 0, "C value. cVal >=0");
CONFIG_PARAM(replicaId,
Expand Down Expand Up @@ -327,8 +329,10 @@ class ReplicaConfig : public concord::serialize::SerializableFactory<ReplicaConf

void serializeDataMembers(std::ostream& outStream) const {
serialize(outStream, isReadOnly);
serialize(outStream, isFullNode);
serialize(outStream, numReplicas);
serialize(outStream, numRoReplicas);
serialize(outStream, numFnReplicas);
serialize(outStream, fVal);
serialize(outStream, cVal);
serialize(outStream, replicaId);
Expand Down Expand Up @@ -427,8 +431,10 @@ class ReplicaConfig : public concord::serialize::SerializableFactory<ReplicaConf
}
void deserializeDataMembers(std::istream& inStream) {
deserialize(inStream, isReadOnly);
deserialize(inStream, isFullNode);
deserialize(inStream, numReplicas);
deserialize(inStream, numRoReplicas);
deserialize(inStream, numFnReplicas);
deserialize(inStream, fVal);
deserialize(inStream, cVal);
deserialize(inStream, replicaId);
Expand Down Expand Up @@ -605,7 +611,8 @@ inline std::ostream& operator<<(std::ostream& os, const ReplicaConfig& rc) {
rc.maxNumberOfDbCheckpoints,
rc.dbCheckPointWindowSize,
rc.dbCheckpointDirPath,
rc.dbSnapshotIntervalSeconds.count());
rc.dbSnapshotIntervalSeconds.count(),
rc.isFullNode);
os << ",";
const auto replicaMsgSignAlgo =
(concord::crypto::SignatureAlgorithm::EdDSA == rc.replicaMsgSigningAlgo) ? "eddsa" : "undefined";
Expand Down
7 changes: 7 additions & 0 deletions bftengine/include/bftengine/ReplicaFactory.hpp
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,7 @@
#include "Replica.hpp"
#include "ReplicaImp.hpp"
#include "ReadOnlyReplica.hpp"
#include "FullNodeReplica.hpp"
#include "ReplicaLoader.hpp"

namespace preprocessor {
Expand Down Expand Up @@ -48,6 +49,12 @@ class ReplicaFactory {
bft::communication::ICommunication *,
MetadataStorage *);

static IReplicaPtr createFullNodeReplica(const ReplicaConfig &,
std::shared_ptr<IRequestsHandler>,
IStateTransfer *,
bft::communication::ICommunication *,
MetadataStorage *);

static void setAggregator(const std::shared_ptr<concordMetrics::Aggregator> &aggregator);
static logging::Logger logger_;

Expand Down
4 changes: 2 additions & 2 deletions bftengine/src/bcstatetransfer/BCStateTran.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -308,7 +308,7 @@ BCStateTran::BCStateTran(const Config &config, IAppState *const stateApi, DataSt
// Validate input parameters and some of the configuration
ConcordAssertNE(stateApi, nullptr);
ConcordAssertGE(replicas_.size(), 3U * config_.fVal + 1U);
ConcordAssert(replicas_.count(config_.myReplicaId) == 1 || config.isReadOnly);
ConcordAssert(replicas_.count(config_.myReplicaId) == 1 || config.isReadOnly || config.isFullNode);
bandatarunkumar marked this conversation as resolved.
Show resolved Hide resolved
ConcordAssertLT(finalizePutblockTimeoutMilli_, config_.refreshTimerMs);
ConcordAssertEQ(RejectFetchingMsg::reasonMessages.size(), RejectFetchingMsg::Reason::LAST - 1);
if (config_.sourceSessionExpiryDurationMs > 0) {
Expand Down Expand Up @@ -1159,7 +1159,7 @@ void BCStateTran::handleStateTransferMessageImpl(char *msg,
time_in_incoming_events_queue_rec_.end();
histograms_.incoming_events_queue_size->record(incomingEventsQ_->size());
}
bool invalidSender = (senderId >= (config_.numReplicas + config_.numRoReplicas));
bool invalidSender = (senderId >= (config_.numReplicas + config_.numRoReplicas + config_.numFnReplicas));
bool sentFromSelf = senderId == config_.myReplicaId;
bool msgSizeTooSmall = msgLen < sizeof(BCStateTranBaseMsg);
if (msgSizeTooSmall || sentFromSelf || invalidSender) {
Expand Down
3 changes: 2 additions & 1 deletion bftengine/src/bftengine/ClientsManager.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -217,7 +217,8 @@ ClientsManager::ClientsManager(const std::set<NodeIdType>& proxyClients,
ConcordAssert(maxNumOfReqsPerClient_ > 0);
reservedPagesPerRequest_ = reservedPagesPerRequest(sizeOfReservedPage(), maxReplySize_);
reservedPagesPerClient_ = reservedPagesPerClient(sizeOfReservedPage(), maxReplySize_, maxNumOfReqsPerClient_);
for (NodeIdType i = 0; i < ReplicaConfig::instance().numReplicas + ReplicaConfig::instance().numRoReplicas; i++) {
const auto& config = ReplicaConfig::instance();
for (NodeIdType i{}; i < config.numReplicas + config.numRoReplicas + config.numFnReplicas; ++i) {
clientIds_.insert(i);
}
clientIds_.insert(proxyClients_.begin(), proxyClients_.end());
Expand Down
235 changes: 235 additions & 0 deletions bftengine/src/bftengine/FullNodeReplica.cpp
Original file line number Diff line number Diff line change
@@ -0,0 +1,235 @@
// Concord
//
// Copyright (c) 2023 VMware, Inc. All Rights Reserved.
//
// This product is licensed to you under the Apache 2.0 license (the "License"). You may not use this product except in
// compliance with the Apache 2.0 License.
//
// This product may include a number of subcomponents with separate copyright notices and license terms. Your use of
// these subcomponents is subject to the terms and conditions of the sub-component's license, as noted in the LICENSE
// file.

#include <optional>
#include <functional>
#include <bitset>

#include "bftengine/Replica.hpp"
#include "messages/StateTransferMsg.hpp"
#include "FullNodeReplica.hpp"
bandatarunkumar marked this conversation as resolved.
Show resolved Hide resolved

#include "log/logger.hpp"
#include "MsgHandlersRegistrator.hpp"
#include "messages/CheckpointMsg.hpp"
#include "messages/AskForCheckpointMsg.hpp"
#include "messages/ClientRequestMsg.hpp"
#include "messages/ClientReplyMsg.hpp"
#include "util/kvstream.h"
#include "PersistentStorage.hpp"
#include "MsgsCommunicator.hpp"
#include "SigManager.hpp"
#include "ReconfigurationCmd.hpp"
#include "util/json_output.hpp"
#include "SharedTypes.hpp"
#include "communication/StateControl.hpp"

using concordUtil::Timers;
using namespace std::placeholders;

// Note : The changes in files are inclined with RO replica SateTransfer behavior, all the class functions are inherited
// from ReadOnlyReplica. As we know for timebeing StateTransfer functionality is a temporary solution for FullNode,
// until the ASP/BSP is implemented the functions in this class needs to be changed based on the required accordingly.

namespace bftEngine::impl {

FullNodeReplica::FullNodeReplica(const ReplicaConfig &config,
std::shared_ptr<IRequestsHandler> requests_handler,
IStateTransfer *state_transfer,
std::shared_ptr<MsgsCommunicator> msg_comm,
std::shared_ptr<MsgHandlersRegistrator> msg_handler_reg,
concordUtil::Timers &timers,
MetadataStorage *metadata_storage)
: ReplicaForStateTransfer(config, requests_handler, state_transfer, msg_comm, msg_handler_reg, true, timers),
fn_metrics_{metrics_.RegisterCounter("receivedCheckpointMsgs"),
metrics_.RegisterCounter("sentAskForCheckpointMsgs"),
metrics_.RegisterCounter("receivedInvalidMsgs"),
metrics_.RegisterGauge("lastExecutedSeqNum", lastExecutedSeqNum)},
metadata_storage_{metadata_storage} {
LOG_INFO(GL, "Initialising Full Node Replica");
repsInfo = new ReplicasInfo(config, dynamicCollectorForPartialProofs, dynamicCollectorForExecutionProofs);

registerMsgHandlers();
metrics_.Register();

SigManager::init(config_.replicaId,
config_.replicaPrivateKey,
config_.publicKeysOfReplicas,
concord::crypto::KeyFormat::HexaDecimalStrippedFormat,
ReplicaConfig::instance().getPublicKeysOfClients(),
concord::crypto::KeyFormat::PemFormat,
{{repsInfo->getIdOfOperator(),
ReplicaConfig::instance().getOperatorPublicKey(),
concord::crypto::KeyFormat::PemFormat}},
*repsInfo);

// Register status handler for Full Node replica
registerStatusHandlers();
bft::communication::StateControl::instance().setGetPeerPubKeyMethod(
[&](uint32_t id) { return SigManager::instance()->getPublicKeyOfVerifier(id); });
}

void FullNodeReplica::start() {
ReplicaForStateTransfer::start();
size_t sendAskForCheckpointMsgPeriodSec = config_.get("concord.bft.ro.sendAskForCheckpointMsgPeriodSec", 30);
askForCheckpointMsgTimer_ = timers_.add(
std::chrono::seconds(sendAskForCheckpointMsgPeriodSec), Timers::Timer::RECURRING, [this](Timers::Handle) {
if (!this->isCollectingState()) {
bandatarunkumar marked this conversation as resolved.
Show resolved Hide resolved
sendAskForCheckpointMsg();
}
});
msgsCommunicator_->startMsgsProcessing(config_.replicaId);
}

void FullNodeReplica::stop() {
timers_.cancel(askForCheckpointMsgTimer_);
ReplicaForStateTransfer::stop();
}

void FullNodeReplica::onTransferringCompleteImp(uint64_t newStateCheckpoint) {
bandatarunkumar marked this conversation as resolved.
Show resolved Hide resolved
last_executed_seq_num_ = newStateCheckpoint * checkpointWindowSize;
fn_metrics_.last_executed_seq_num_.Get().Set(last_executed_seq_num_);
}

void FullNodeReplica::onReportAboutInvalidMessage(MessageBase *msg, const char *reason) {
fn_metrics_.received_invalid_msg_++;
LOG_WARN(GL,
"Node " << config_.replicaId << " received invalid message from Node " << msg->senderId()
<< " type=" << msg->type() << " reason: " << reason);
}
void FullNodeReplica::sendAskForCheckpointMsg() {
bandatarunkumar marked this conversation as resolved.
Show resolved Hide resolved
fn_metrics_.sent_ask_for_checkpoint_msg_++;
LOG_INFO(GL, "sending AskForCheckpointMsg");
AskForCheckpointMsg msg{config_.replicaId};
for (auto id : repsInfo->idsOfPeerReplicas()) send(&msg, id);
}

template <>
void FullNodeReplica::onMessage<StateTransferMsg>(std::unique_ptr<StateTransferMsg> msg) {
ReplicaForStateTransfer::onMessage(move(msg));
}

template <>
bandatarunkumar marked this conversation as resolved.
Show resolved Hide resolved
void FullNodeReplica::onMessage<CheckpointMsg>(std::unique_ptr<CheckpointMsg> msg) {
if (isCollectingState()) {
return;
}
fn_metrics_.received_checkpoint_msg_++;
LOG_INFO(GL,
KVLOG(msg->senderId(),
msg->idOfGeneratedReplica(),
msg->seqNumber(),
msg->epochNumber(),
msg->size(),
msg->isStableState(),
msg->state(),
msg->stateDigest(),
msg->reservedPagesDigest(),
msg->rvbDataDigest()));

// Reconfiguration cmd block is synced to RO replica via reserved pages
EpochNum replicas_last_known_epoch_val = 0;
auto epoch_number_from_res_pages = ReconfigurationCmd::instance().getReconfigurationCommandEpochNumber();
if (epoch_number_from_res_pages.has_value()) replicas_last_known_epoch_val = epoch_number_from_res_pages.value();

// not relevant
if (!msg->isStableState() || msg->seqNumber() <= lastExecutedSeqNum ||
msg->epochNumber() < replicas_last_known_epoch_val) {
return;
}
// no self certificate
static std::map<SeqNum, CheckpointInfo<false>> checkpoints_info;
const auto msg_seq_num = msg->seqNumber();
const auto id_of_generated_eplica = msg->idOfGeneratedReplica();
checkpoints_info[msg_seq_num].addCheckpointMsg(msg.release(), id_of_generated_eplica);
// if enough - invoke state transfer
if (checkpoints_info[msg_seq_num].isCheckpointCertificateComplete()) {
persistCheckpointDescriptor(msg_seq_num, checkpoints_info[msg_seq_num]);
checkpoints_info.clear();
LOG_INFO(GL, "call to startCollectingState()");
stateTransfer->startCollectingState();
}
}

void FullNodeReplica::persistCheckpointDescriptor(const SeqNum &seqnum, const CheckpointInfo<false> &chckpinfo) {
bandatarunkumar marked this conversation as resolved.
Show resolved Hide resolved
std::vector<CheckpointMsg *> msgs;
msgs.reserve(chckpinfo.getAllCheckpointMsgs().size());
for (const auto &m : chckpinfo.getAllCheckpointMsgs()) {
msgs.push_back(m.second);
LOG_INFO(GL,
KVLOG(m.second->seqNumber(),
m.second->epochNumber(),
m.second->state(),
m.second->stateDigest(),
m.second->reservedPagesDigest(),
m.second->rvbDataDigest(),
m.second->idOfGeneratedReplica()));
}
DescriptorOfLastStableCheckpoint desc(ReplicaConfig::instance().getnumReplicas(), msgs);
const size_t buf_len = DescriptorOfLastStableCheckpoint::maxSize(ReplicaConfig::instance().getnumReplicas());
concord::serialize::UniquePtrToChar desc_buf(new char[buf_len]);
char *desc_buf_ptr = desc_buf.get();
size_t actual_size = 0;
desc.serialize(desc_buf_ptr, buf_len, actual_size);
ConcordAssertNE(actual_size, 0);

// TODO [TK] S3KeyGenerator
// checkpoints/<BlockId>/<RepId>
std::ostringstream oss;
oss << "checkpoints/" << msgs[0]->state() << "/" << config_.replicaId;
metadata_storage_->atomicWriteArbitraryObject(oss.str(), desc_buf.get(), actual_size);
}

template <>
void FullNodeReplica::onMessage<ClientRequestMsg>(std::unique_ptr<ClientRequestMsg> msg) {
bandatarunkumar marked this conversation as resolved.
Show resolved Hide resolved
const NodeIdType sender_id = msg->senderId();
const NodeIdType client_id = msg->clientProxyId();
const ReqId req_seq_num = msg->requestSeqNum();
const uint64_t flags = msg->flags();

SCOPED_MDC_CID(msg->getCid());
LOG_DEBUG(CNSUS, KVLOG(client_id, req_seq_num, sender_id) << " flags: " << std::bitset<sizeof(uint64_t) * 8>(flags));

const auto &span_context = msg->spanContext<std::remove_pointer<ClientRequestMsg>::type>();
auto span = concordUtils::startChildSpanFromContext(span_context, "bft_client_request");
span.setTag("rid", config_.getreplicaId());
span.setTag("cid", msg->getCid());
span.setTag("seq_num", req_seq_num);

// TODO: handle reconfiguration request here, refer ReadOnlyReplica class
}

void FullNodeReplica::registerStatusHandlers() {
auto h = concord::diagnostics::StatusHandler(
"replica-sequence-numbers", "Last executed sequence number of the full node replica", [this]() {
concordUtils::BuildJson bj;

bj.startJson();
bj.startNested("sequenceNumbers");
bj.addKv("lastExecutedSeqNum", last_executed_seq_num_);
bj.endNested();
bj.endJson();

return bj.getJson();
});
concord::diagnostics::RegistrarSingleton::getInstance().status.registerHandler(h);
}

void FullNodeReplica::registerMsgHandlers() {
msgHandlers_->registerMsgHandler(MsgCode::Checkpoint,
std::bind(&FullNodeReplica::messageHandler<CheckpointMsg>, this, _1));
msgHandlers_->registerMsgHandler(MsgCode::ClientRequest,
std::bind(&FullNodeReplica::messageHandler<ClientRequestMsg>, this, _1));
msgHandlers_->registerMsgHandler(MsgCode::StateTransfer,
std::bind(&FullNodeReplica::messageHandler<StateTransferMsg>, this, _1));
}

} // namespace bftEngine::impl
Loading