wbtiger
/
foundationdb
mirror of https://github.com/apple/foundationdb.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
foundationdb/fdbclient/MonitorLeader.actor.cpp

874 lines
32 KiB
C++
Raw Blame History

/*
* MonitorLeader.actor.cpp
*
* This source file is part of the FoundationDB open source project
*
* Copyright 2013-2018 Apple Inc. and the FoundationDB project authors
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "fdbclient/MonitorLeader.h"
#include "fdbclient/CoordinationInterface.h"
#include "flow/ActorCollection.h"
#include "flow/UnitTest.h"
#include "fdbrpc/genericactors.actor.h"
#include "flow/Platform.h"
#include "flow/actorcompiler.h" // has to be last include
namespace {
std::string trim(std::string const& connectionString) {
// Strip out whitespace
// Strip out characters between a # and a newline
std::string trimmed;
auto end = connectionString.end();
for (auto c = connectionString.begin(); c != end; ++c) {
if (*c == '#') {
++c;
while (c != end && *c != '\n' && *c != '\r')
++c;
if (c == end)
break;
} else if (*c != ' ' && *c != '\n' && *c != '\r' && *c != '\t')
trimmed += *c;
}
return trimmed;
}
} // namespace
std::pair<std::string, bool> ClusterConnectionFile::lookupClusterFileName(std::string const& filename) {
if (filename.length())
return std::make_pair(filename, false);
std::string f;
bool isDefaultFile = true;
if (platform::getEnvironmentVar(CLUSTER_FILE_ENV_VAR_NAME, f)) {
// If this is set but points to a file that does not
// exist, we will not fallback to any other methods
isDefaultFile = false;
} else if (fileExists("fdb.cluster"))
f = "fdb.cluster";
else
f = platform::getDefaultClusterFilePath();
return std::make_pair(f, isDefaultFile);
}
std::string ClusterConnectionFile::getErrorString(std::pair<std::string, bool> const& resolvedClusterFile,
Error const& e) {
bool isDefault = resolvedClusterFile.second;
if (e.code() == error_code_connection_string_invalid) {
return format("Invalid cluster file `%s': %d %s", resolvedClusterFile.first.c_str(), e.code(), e.what());
} else if (e.code() == error_code_no_cluster_file_found) {
if (isDefault)
return format("Unable to read cluster file `./fdb.cluster' or `%s' and %s unset: %d %s",
platform::getDefaultClusterFilePath().c_str(),
CLUSTER_FILE_ENV_VAR_NAME,
e.code(),
e.what());
else
return format(
"Unable to read cluster file `%s': %d %s", resolvedClusterFile.first.c_str(), e.code(), e.what());
} else {
return format(
"Unexpected error loading cluster file `%s': %d %s", resolvedClusterFile.first.c_str(), e.code(), e.what());
}
}
ClusterConnectionFile::ClusterConnectionFile(std::string const& filename) {
if (!fileExists(filename)) {
throw no_cluster_file_found();
}
cs = ClusterConnectionString(readFileBytes(filename, MAX_CLUSTER_FILE_BYTES));
this->filename = filename;
setConn = false;
}
ClusterConnectionFile::ClusterConnectionFile(std::string const& filename, ClusterConnectionString const& contents) {
this->filename = filename;
cs = contents;
setConn = true;
}
ClusterConnectionString const& ClusterConnectionFile::getConnectionString() const {
return cs;
}
void ClusterConnectionFile::notifyConnected() {
if (setConn) {
this->writeFile();
}
}
bool ClusterConnectionFile::fileContentsUpToDate() const {
ClusterConnectionString temp;
return fileContentsUpToDate(temp);
}
bool ClusterConnectionFile::fileContentsUpToDate(ClusterConnectionString& fileConnectionString) const {
try {
// the cluster file hasn't been created yet so there's nothing to check
if (setConn)
return true;
ClusterConnectionFile temp(filename);
fileConnectionString = temp.getConnectionString();
return fileConnectionString.toString() == cs.toString();
} catch (Error& e) {
TraceEvent(SevWarnAlways, "ClusterFileError").error(e).detail("Filename", filename);
return false; // Swallow the error and report that the file is out of date
}
}
bool ClusterConnectionFile::writeFile() {
setConn = false;
if (filename.size()) {
try {
atomicReplace(filename,
"# DO NOT EDIT!\n# This file is auto-generated, it is not to be edited by hand\n" +
cs.toString().append("\n"));
if (!fileContentsUpToDate()) {
// This should only happen in rare scenarios where multiple processes are updating the same file to
// different values simultaneously In that case, we don't have any guarantees about which file will
// ultimately be written
TraceEvent(SevWarnAlways, "ClusterFileChangedAfterReplace")
.detail("Filename", filename)
.detail("ConnStr", cs.toString());
return false;
}
return true;
} catch (Error& e) {
TraceEvent(SevWarnAlways, "UnableToChangeConnectionFile")
.error(e)
.detail("Filename", filename)
.detail("ConnStr", cs.toString());
}
}
return false;
}
void ClusterConnectionFile::setConnectionString(ClusterConnectionString const& conn) {
ASSERT(filename.size());
cs = conn;
writeFile();
}
std::string ClusterConnectionString::getErrorString(std::string const& source, Error const& e) {
if (e.code() == error_code_connection_string_invalid) {
return format("Invalid connection string `%s: %d %s", source.c_str(), e.code(), e.what());
} else {
return format("Unexpected error parsing connection string `%s: %d %s", source.c_str(), e.code(), e.what());
}
}
ClusterConnectionString::ClusterConnectionString(std::string const& connectionString) {
auto trimmed = trim(connectionString);
// Split on '@' into key@addrs
int pAt = trimmed.find_first_of('@');
if (pAt == trimmed.npos)
throw connection_string_invalid();
std::string key = trimmed.substr(0, pAt);
std::string addrs = trimmed.substr(pAt + 1);
parseKey(key);
coord = NetworkAddress::parseList(addrs);
ASSERT(coord.size() > 0); // parseList() always returns at least one address if it doesn't throw
std::sort(coord.begin(), coord.end());
// Check that there are no duplicate addresses
if (std::unique(coord.begin(), coord.end()) != coord.end())
throw connection_string_invalid();
}
TEST_CASE("/fdbclient/MonitorLeader/parseConnectionString/basic") {
std::string input;
{
input = "asdf:2345@1.1.1.1:345";
ClusterConnectionString cs(input);
ASSERT(input == cs.toString());
}
{
input = "0xxdeadbeef:100100100@1.1.1.1:34534,5.1.5.3:23443";
ClusterConnectionString cs(input);
ASSERT(input == cs.toString());
}
{
input = "0xxdeadbeef:100100100@1.1.1.1:34534,5.1.5.3:23443";
std::string commented("#start of comment\n");
commented += input;
commented += "\n";
commented += "# asdfasdf ##";
ClusterConnectionString cs(commented);
ASSERT(input == cs.toString());
}
{
input = "0xxdeadbeef:100100100@[::1]:1234,[::1]:1235";
std::string commented("#start of comment\n");
commented += input;
commented += "\n";
commented += "# asdfasdf ##";
ClusterConnectionString cs(commented);
ASSERT(input == cs.toString());
}
{
input = "0xxdeadbeef:100100100@[abcd:dcba::1]:1234,[abcd:dcba::abcd:1]:1234";
std::string commented("#start of comment\n");
commented += input;
commented += "\n";
commented += "# asdfasdf ##";
ClusterConnectionString cs(commented);
ASSERT(input == cs.toString());
}
return Void();
}
TEST_CASE("/flow/FlatBuffers/LeaderInfo") {
{
LeaderInfo in;
LeaderInfo out;
in.forward = deterministicRandom()->coinflip();
in.changeID = deterministicRandom()->randomUniqueID();
{
std::string rndString(deterministicRandom()->randomInt(10, 400), 'x');
for (auto& c : rndString) {
c = deterministicRandom()->randomAlphaNumeric();
}
in.serializedInfo = rndString;
}
ObjectWriter writer(IncludeVersion());
writer.serialize(in);
Standalone<StringRef> copy = writer.toStringRef();
ArenaObjectReader reader(copy.arena(), copy, IncludeVersion());
reader.deserialize(out);
ASSERT(in.forward == out.forward);
ASSERT(in.changeID == out.changeID);
ASSERT(in.serializedInfo == out.serializedInfo);
}
LeaderInfo leaderInfo;
leaderInfo.forward = deterministicRandom()->coinflip();
leaderInfo.changeID = deterministicRandom()->randomUniqueID();
{
std::string rndString(deterministicRandom()->randomInt(10, 400), 'x');
for (auto& c : rndString) {
c = deterministicRandom()->randomAlphaNumeric();
}
leaderInfo.serializedInfo = rndString;
}
ErrorOr<EnsureTable<Optional<LeaderInfo>>> objIn(leaderInfo);
ErrorOr<EnsureTable<Optional<LeaderInfo>>> objOut;
Standalone<StringRef> copy;
ObjectWriter writer(IncludeVersion());
writer.serialize(objIn);
copy = writer.toStringRef();
ArenaObjectReader reader(copy.arena(), copy, IncludeVersion());
reader.deserialize(objOut);
ASSERT(!objOut.isError());
ASSERT(objOut.get().asUnderlyingType().present());
LeaderInfo outLeader = objOut.get().asUnderlyingType().get();
ASSERT(outLeader.changeID == leaderInfo.changeID);
ASSERT(outLeader.forward == leaderInfo.forward);
ASSERT(outLeader.serializedInfo == leaderInfo.serializedInfo);
return Void();
}
TEST_CASE("/fdbclient/MonitorLeader/parseConnectionString/fuzz") {
// For a static connection string, add in fuzzed comments and whitespace
// SOMEDAY: create a series of random connection strings, rather than the one we started with
std::string connectionString = "0xxdeadbeef:100100100@1.1.1.1:34534,5.1.5.3:23443";
for (int i = 0; i < 10000; i++) {
std::string output("");
auto c = connectionString.begin();
while (c != connectionString.end()) {
if (deterministicRandom()->random01() < 0.1) // Add whitespace character
output += deterministicRandom()->randomChoice(LiteralStringRef(" \t\n\r"));
if (deterministicRandom()->random01() < 0.5) { // Add one of the input characters
output += *c;
++c;
}
if (deterministicRandom()->random01() < 0.1) { // Add a comment block
output += "#";
int charCount = deterministicRandom()->randomInt(0, 20);
for (int i = 0; i < charCount; i++) {
output += deterministicRandom()->randomChoice(LiteralStringRef("asdfzxcv123345:!@#$#$&()<\"\' \t"));
}
output += deterministicRandom()->randomChoice(LiteralStringRef("\n\r"));
}
}
ClusterConnectionString cs(output);
ASSERT(connectionString == cs.toString());
}
return Void();
}
ClusterConnectionString::ClusterConnectionString(vector<NetworkAddress> servers, Key key) : coord(servers) {
parseKey(key.toString());
}
void ClusterConnectionString::parseKey(std::string const& key) {
// Check the structure of the given key, and fill in this->key and this->keyDesc
// The key must contain one (and only one) : character
int colon = key.find_first_of(':');
if (colon == key.npos)
throw connection_string_invalid();
std::string desc = key.substr(0, colon);
std::string id = key.substr(colon + 1);
// Check that description contains only allowed characters (a-z, A-Z, 0-9, _)
for (auto c = desc.begin(); c != desc.end(); ++c)
if (!(isalnum(*c) || *c == '_'))
throw connection_string_invalid();
// Check that ID contains only allowed characters (a-z, A-Z, 0-9)
for (auto c = id.begin(); c != id.end(); ++c)
if (!isalnum(*c))
throw connection_string_invalid();
this->key = StringRef(key);
this->keyDesc = StringRef(desc);
}
std::string ClusterConnectionString::toString() const {
std::string s = key.toString();
s += '@';
for (int i = 0; i < coord.size(); i++) {
if (i)
s += ',';
s += coord[i].toString();
}
return s;
}
ClientCoordinators::ClientCoordinators(Reference<ClusterConnectionFile> ccf) : ccf(ccf) {
ClusterConnectionString cs = ccf->getConnectionString();
for (auto s = cs.coordinators().begin(); s != cs.coordinators().end(); ++s)
clientLeaderServers.push_back(ClientLeaderRegInterface(*s));
clusterKey = cs.clusterKey();
}
ClientCoordinators::ClientCoordinators(Key clusterKey, std::vector<NetworkAddress> coordinators)
: clusterKey(clusterKey) {
for (const auto& coord : coordinators) {
clientLeaderServers.push_back(ClientLeaderRegInterface(coord));
}
ccf = makeReference<ClusterConnectionFile>(ClusterConnectionString(coordinators, clusterKey));
}
ClientLeaderRegInterface::ClientLeaderRegInterface(NetworkAddress remote)
: getLeader(Endpoint({ remote }, WLTOKEN_CLIENTLEADERREG_GETLEADER)),
openDatabase(Endpoint({ remote }, WLTOKEN_CLIENTLEADERREG_OPENDATABASE)),
checkDescriptorMutable(Endpoint({ remote }, WLTOKEN_CLIENTLEADERREG_DESCRIPTOR_MUTABLE)) {}
ClientLeaderRegInterface::ClientLeaderRegInterface(INetwork* local) {
getLeader.makeWellKnownEndpoint(WLTOKEN_CLIENTLEADERREG_GETLEADER, TaskPriority::Coordination);
openDatabase.makeWellKnownEndpoint(WLTOKEN_CLIENTLEADERREG_OPENDATABASE, TaskPriority::Coordination);
checkDescriptorMutable.makeWellKnownEndpoint(WLTOKEN_CLIENTLEADERREG_DESCRIPTOR_MUTABLE,
TaskPriority::Coordination);
}
// Nominee is the worker among all workers that are considered as leader by a coordinator
// This function contacts a coordinator coord to ask if the worker is considered as a leader (i.e., if the worker
// is a nominee)
ACTOR Future<Void> monitorNominee(Key key,
ClientLeaderRegInterface coord,
AsyncTrigger* nomineeChange,
Optional<LeaderInfo>* info) {
loop {
state Optional<LeaderInfo> li =
wait(retryBrokenPromise(coord.getLeader,
GetLeaderRequest(key, info->present() ? info->get().changeID : UID()),
TaskPriority::CoordinationReply));
wait(Future<Void>(Void())); // Make sure we weren't cancelled
TraceEvent("GetLeaderReply")
.suppressFor(1.0)
.detail("Coordinator", coord.getLeader.getEndpoint().getPrimaryAddress())
.detail("Nominee", li.present() ? li.get().changeID : UID())
.detail("ClusterKey", key.printable());
if (li != *info) {
*info = li;
nomineeChange->trigger();
if (li.present() && li.get().forward)
wait(Future<Void>(Never()));
wait(Future<Void>(Void()));
}
}
}
// Also used in fdbserver/LeaderElection.actor.cpp!
// bool represents if the LeaderInfo is a majority answer or not.
// This function also masks the first 7 bits of changeId of the nominees and returns the Leader with masked changeId
Optional<std::pair<LeaderInfo, bool>> getLeader(const vector<Optional<LeaderInfo>>& nominees) {
// If any coordinator says that the quorum is forwarded, then it is
for (int i = 0; i < nominees.size(); i++)
if (nominees[i].present() && nominees[i].get().forward)
return std::pair<LeaderInfo, bool>(nominees[i].get(), true);
vector<std::pair<UID, int>> maskedNominees;
maskedNominees.reserve(nominees.size());
for (int i = 0; i < nominees.size(); i++) {
if (nominees[i].present()) {
maskedNominees.emplace_back(
UID(nominees[i].get().changeID.first() & LeaderInfo::changeIDMask, nominees[i].get().changeID.second()),
i);
}
}
if (!maskedNominees.size())
return Optional<std::pair<LeaderInfo, bool>>();
std::sort(maskedNominees.begin(),
maskedNominees.end(),
[](const std::pair<UID, int>& l, const std::pair<UID, int>& r) { return l.first < r.first; });
int bestCount = 0;
int bestIdx = 0;
int currentIdx = 0;
int curCount = 1;
for (int i = 1; i < maskedNominees.size(); i++) {
if (maskedNominees[currentIdx].first == maskedNominees[i].first) {
curCount++;
} else {
if (curCount > bestCount) {
bestIdx = currentIdx;
bestCount = curCount;
}
currentIdx = i;
curCount = 1;
}
}
if (curCount > bestCount) {
bestIdx = currentIdx;
bestCount = curCount;
}
bool majority = bestCount >= nominees.size() / 2 + 1;
return std::pair<LeaderInfo, bool>(nominees[maskedNominees[bestIdx].second].get(), majority);
}
// Leader is the process that will be elected by coordinators as the cluster controller
ACTOR Future<MonitorLeaderInfo> monitorLeaderOneGeneration(Reference<ClusterConnectionFile> connFile,
Reference<AsyncVar<Value>> outSerializedLeaderInfo,
MonitorLeaderInfo info) {
state ClientCoordinators coordinators(info.intermediateConnFile);
state AsyncTrigger nomineeChange;
state std::vector<Optional<LeaderInfo>> nominees;
state Future<Void> allActors;
nominees.resize(coordinators.clientLeaderServers.size());
std::vector<Future<Void>> actors;
// Ask all coordinators if the worker is considered as a leader (leader nominee) by the coordinator.
actors.reserve(coordinators.clientLeaderServers.size());
for (int i = 0; i < coordinators.clientLeaderServers.size(); i++)
actors.push_back(
monitorNominee(coordinators.clusterKey, coordinators.clientLeaderServers[i], &nomineeChange, &nominees[i]));
allActors = waitForAll(actors);
loop {
Optional<std::pair<LeaderInfo, bool>> leader = getLeader(nominees);
TraceEvent("MonitorLeaderChange")
.detail("NewLeader", leader.present() ? leader.get().first.changeID : UID(1, 1));
if (leader.present()) {
if (leader.get().first.forward) {
TraceEvent("MonitorLeaderForwarding")
.detail("NewConnStr", leader.get().first.serializedInfo.toString())
.detail("OldConnStr", info.intermediateConnFile->getConnectionString().toString())
.trackLatest("MonitorLeaderForwarding");
info.intermediateConnFile = makeReference<ClusterConnectionFile>(
connFile->getFilename(), ClusterConnectionString(leader.get().first.serializedInfo.toString()));
return info;
}
if (connFile != info.intermediateConnFile) {
if (!info.hasConnected) {
TraceEvent(SevWarnAlways, "IncorrectClusterFileContentsAtConnection")
.detail("Filename", connFile->getFilename())
.detail("ConnectionStringFromFile", connFile->getConnectionString().toString())
.detail("CurrentConnectionString", info.intermediateConnFile->getConnectionString().toString());
}
connFile->setConnectionString(info.intermediateConnFile->getConnectionString());
info.intermediateConnFile = connFile;
}
info.hasConnected = true;
connFile->notifyConnected();
outSerializedLeaderInfo->set(leader.get().first.serializedInfo);
}
wait(nomineeChange.onTrigger() || allActors);
}
}
Future<Void> monitorLeaderRemotelyInternal(Reference<ClusterConnectionFile> const& connFile,
Reference<AsyncVar<Value>> const& outSerializedLeaderInfo);
template <class LeaderInterface>
Future<Void> monitorLeaderRemotely(Reference<ClusterConnectionFile> const& connFile,
Reference<AsyncVar<Optional<LeaderInterface>>> const& outKnownLeader) {
LeaderDeserializer<LeaderInterface> deserializer;
auto serializedInfo = makeReference<AsyncVar<Value>>();
Future<Void> m = monitorLeaderRemotelyInternal(connFile, serializedInfo);
return m || deserializer(serializedInfo, outKnownLeader);
}
ACTOR Future<Void> monitorLeaderInternal(Reference<ClusterConnectionFile> connFile,
Reference<AsyncVar<Value>> outSerializedLeaderInfo) {
state MonitorLeaderInfo info(connFile);
loop {
MonitorLeaderInfo _info = wait(monitorLeaderOneGeneration(connFile, outSerializedLeaderInfo, info));
info = _info;
}
}
ACTOR Future<Void> asyncDeserializeClusterInterface(Reference<AsyncVar<Value>> serializedInfo,
Reference<AsyncVar<Optional<ClusterInterface>>> outKnownLeader) {
state Reference<AsyncVar<Optional<ClusterControllerClientInterface>>> knownLeader(
new AsyncVar<Optional<ClusterControllerClientInterface>>{});
state Future<Void> deserializer = asyncDeserialize(serializedInfo, knownLeader);
loop {
choose {
when(wait(deserializer)) { UNSTOPPABLE_ASSERT(false); }
when(wait(knownLeader->onChange())) {
if (knownLeader->get().present()) {
outKnownLeader->set(knownLeader->get().get().clientInterface);
} else {
outKnownLeader->set(Optional<ClusterInterface>{});
}
}
}
}
}
struct ClientStatusStats {
int count;
std::vector<std::pair<NetworkAddress, Key>> examples;
ClientStatusStats() : count(0) { examples.reserve(CLIENT_KNOBS->CLIENT_EXAMPLE_AMOUNT); }
};
OpenDatabaseRequest ClientData::getRequest() {
OpenDatabaseRequest req;
std::map<StringRef, ClientStatusStats> issueMap;
std::map<ClientVersionRef, ClientStatusStats> versionMap;
std::map<StringRef, ClientStatusStats> maxProtocolMap;
int clientCount = 0;
// SOMEDAY: add a yield in this loop
for (auto& ci : clientStatusInfoMap) {
for (auto& it : ci.second.issues) {
auto& entry = issueMap[it];
entry.count++;
if (entry.examples.size() < CLIENT_KNOBS->CLIENT_EXAMPLE_AMOUNT) {
entry.examples.emplace_back(ci.first, ci.second.traceLogGroup);
}
}
if (ci.second.versions.size()) {
clientCount++;
StringRef maxProtocol;
for (auto& it : ci.second.versions) {
maxProtocol = std::max(maxProtocol, it.protocolVersion);
auto& entry = versionMap[it];
entry.count++;
if (entry.examples.size() < CLIENT_KNOBS->CLIENT_EXAMPLE_AMOUNT) {
entry.examples.emplace_back(ci.first, ci.second.traceLogGroup);
}
}
auto& maxEntry = maxProtocolMap[maxProtocol];
maxEntry.count++;
if (maxEntry.examples.size() < CLIENT_KNOBS->CLIENT_EXAMPLE_AMOUNT) {
maxEntry.examples.emplace_back(ci.first, ci.second.traceLogGroup);
}
} else {
auto& entry = versionMap[ClientVersionRef()];
entry.count++;
if (entry.examples.size() < CLIENT_KNOBS->CLIENT_EXAMPLE_AMOUNT) {
entry.examples.emplace_back(ci.first, ci.second.traceLogGroup);
}
}
}
req.issues.reserve(issueMap.size());
for (auto& it : issueMap) {
req.issues.push_back(ItemWithExamples<Key>(it.first, it.second.count, it.second.examples));
}
req.supportedVersions.reserve(versionMap.size());
for (auto& it : versionMap) {
req.supportedVersions.push_back(
ItemWithExamples<Standalone<ClientVersionRef>>(it.first, it.second.count, it.second.examples));
}
req.maxProtocolSupported.reserve(maxProtocolMap.size());
for (auto& it : maxProtocolMap) {
req.maxProtocolSupported.push_back(ItemWithExamples<Key>(it.first, it.second.count, it.second.examples));
}
req.clientCount = clientCount;
return req;
}
ACTOR Future<Void> getClientInfoFromLeader(Reference<AsyncVar<Optional<ClusterControllerClientInterface>>> knownLeader,
ClientData* clientData) {
while (!knownLeader->get().present()) {
wait(knownLeader->onChange());
}
state double lastRequestTime = now();
state OpenDatabaseRequest req = clientData->getRequest();
loop {
if (now() - lastRequestTime > CLIENT_KNOBS->MAX_CLIENT_STATUS_AGE) {
lastRequestTime = now();
req = clientData->getRequest();
} else {
resetReply(req);
}
req.knownClientInfoID = clientData->clientInfo->get().read().id;
choose {
when(ClientDBInfo ni =
wait(brokenPromiseToNever(knownLeader->get().get().clientInterface.openDatabase.getReply(req)))) {
TraceEvent("MonitorLeaderForProxiesGotClientInfo", knownLeader->get().get().clientInterface.id())
.detail("CommitProxy0", ni.commitProxies.size() ? ni.commitProxies[0].id() : UID())
.detail("GrvProxy0", ni.grvProxies.size() ? ni.grvProxies[0].id() : UID())
.detail("ClientID", ni.id);
clientData->clientInfo->set(CachedSerialization<ClientDBInfo>(ni));
}
when(wait(knownLeader->onChange())) {}
}
}
}
ACTOR Future<Void> monitorLeaderForProxies(Key clusterKey,
vector<NetworkAddress> coordinators,
ClientData* clientData,
Reference<AsyncVar<Optional<LeaderInfo>>> leaderInfo) {
state vector<ClientLeaderRegInterface> clientLeaderServers;
state AsyncTrigger nomineeChange;
state std::vector<Optional<LeaderInfo>> nominees;
state Future<Void> allActors;
state Reference<AsyncVar<Optional<ClusterControllerClientInterface>>> knownLeader(
new AsyncVar<Optional<ClusterControllerClientInterface>>{});
for (auto s = coordinators.begin(); s != coordinators.end(); ++s) {
clientLeaderServers.push_back(ClientLeaderRegInterface(*s));
}
nominees.resize(clientLeaderServers.size());
std::vector<Future<Void>> actors;
// Ask all coordinators if the worker is considered as a leader (leader nominee) by the coordinator.
actors.reserve(clientLeaderServers.size());
for (int i = 0; i < clientLeaderServers.size(); i++) {
actors.push_back(monitorNominee(clusterKey, clientLeaderServers[i], &nomineeChange, &nominees[i]));
}
actors.push_back(getClientInfoFromLeader(knownLeader, clientData));
allActors = waitForAll(actors);
loop {
Optional<std::pair<LeaderInfo, bool>> leader = getLeader(nominees);
TraceEvent("MonitorLeaderForProxiesChange")
.detail("NewLeader", leader.present() ? leader.get().first.changeID : UID(1, 1))
.detail("Key", clusterKey.printable());
if (leader.present()) {
if (leader.get().first.forward) {
ClientDBInfo outInfo;
outInfo.id = deterministicRandom()->randomUniqueID();
outInfo.forward = leader.get().first.serializedInfo;
clientData->clientInfo->set(CachedSerialization<ClientDBInfo>(outInfo));
leaderInfo->set(leader.get().first);
TraceEvent("MonitorLeaderForProxiesForwarding")
.detail("NewConnStr", leader.get().first.serializedInfo.toString());
return Void();
}
if (leader.get().first.serializedInfo.size()) {
ObjectReader reader(leader.get().first.serializedInfo.begin(), IncludeVersion());
ClusterControllerClientInterface res;
reader.deserialize(res);
knownLeader->set(res);
leaderInfo->set(leader.get().first);
}
}
wait(nomineeChange.onTrigger() || allActors);
}
}
void shrinkProxyList(ClientDBInfo& ni,
std::vector<UID>& lastCommitProxyUIDs,
std::vector<CommitProxyInterface>& lastCommitProxies,
std::vector<UID>& lastGrvProxyUIDs,
std::vector<GrvProxyInterface>& lastGrvProxies) {
if (ni.commitProxies.size() > CLIENT_KNOBS->MAX_COMMIT_PROXY_CONNECTIONS) {
std::vector<UID> commitProxyUIDs;
for (auto& commitProxy : ni.commitProxies) {
commitProxyUIDs.push_back(commitProxy.id());
}
if (commitProxyUIDs != lastCommitProxyUIDs) {
lastCommitProxyUIDs.swap(commitProxyUIDs);
lastCommitProxies = ni.commitProxies;
deterministicRandom()->randomShuffle(lastCommitProxies);
lastCommitProxies.resize(CLIENT_KNOBS->MAX_COMMIT_PROXY_CONNECTIONS);
for (int i = 0; i < lastCommitProxies.size(); i++) {
TraceEvent("ConnectedCommitProxy").detail("CommitProxy", lastCommitProxies[i].id());
}
}
ni.firstCommitProxy = ni.commitProxies[0];
ni.commitProxies = lastCommitProxies;
}
if (ni.grvProxies.size() > CLIENT_KNOBS->MAX_GRV_PROXY_CONNECTIONS) {
std::vector<UID> grvProxyUIDs;
for (auto& grvProxy : ni.grvProxies) {
grvProxyUIDs.push_back(grvProxy.id());
}
if (grvProxyUIDs != lastGrvProxyUIDs) {
lastGrvProxyUIDs.swap(grvProxyUIDs);
lastGrvProxies = ni.grvProxies;
deterministicRandom()->randomShuffle(lastGrvProxies);
lastGrvProxies.resize(CLIENT_KNOBS->MAX_GRV_PROXY_CONNECTIONS);
for (int i = 0; i < lastGrvProxies.size(); i++) {
TraceEvent("ConnectedGrvProxy").detail("GrvProxy", lastGrvProxies[i].id());
}
}
ni.grvProxies = lastGrvProxies;
}
}
// Leader is the process that will be elected by coordinators as the cluster controller
ACTOR Future<MonitorLeaderInfo> monitorProxiesOneGeneration(
Reference<ClusterConnectionFile> connFile,
Reference<AsyncVar<ClientDBInfo>> clientInfo,
Reference<AsyncVar<Optional<ClientLeaderRegInterface>>> coordinator,
MonitorLeaderInfo info,
Reference<ReferencedObject<Standalone<VectorRef<ClientVersionRef>>>> supportedVersions,
Key traceLogGroup) {
state ClusterConnectionString cs = info.intermediateConnFile->getConnectionString();
state vector<NetworkAddress> addrs = cs.coordinators();
state int idx = 0;
state int successIdx = 0;
state Optional<double> incorrectTime;
state std::vector<UID> lastCommitProxyUIDs;
state std::vector<CommitProxyInterface> lastCommitProxies;
state std::vector<UID> lastGrvProxyUIDs;
state std::vector<GrvProxyInterface> lastGrvProxies;
deterministicRandom()->randomShuffle(addrs);
loop {
state ClientLeaderRegInterface clientLeaderServer(addrs[idx]);
state OpenDatabaseCoordRequest req;
coordinator->set(clientLeaderServer);
req.clusterKey = cs.clusterKey();
req.coordinators = cs.coordinators();
req.knownClientInfoID = clientInfo->get().id;
req.supportedVersions = supportedVersions->get();
req.traceLogGroup = traceLogGroup;
ClusterConnectionString fileConnectionString;
if (connFile && !connFile->fileContentsUpToDate(fileConnectionString)) {
req.issues.push_back_deep(req.issues.arena(), LiteralStringRef("incorrect_cluster_file_contents"));
std::string connectionString = connFile->getConnectionString().toString();
if (!incorrectTime.present()) {
incorrectTime = now();
}
if (connFile->canGetFilename()) {
// Don't log a SevWarnAlways initially to account for transient issues (e.g. someone else changing the
// file right before us)
TraceEvent(now() - incorrectTime.get() > 300 ? SevWarnAlways : SevWarn, "IncorrectClusterFileContents")
.detail("Filename", connFile->getFilename())
.detail("ConnectionStringFromFile", fileConnectionString.toString())
.detail("CurrentConnectionString", connectionString);
}
} else {
incorrectTime = Optional<double>();
}
state ErrorOr<CachedSerialization<ClientDBInfo>> rep =
wait(clientLeaderServer.openDatabase.tryGetReply(req, TaskPriority::CoordinationReply));
if (rep.present()) {
if (rep.get().read().forward.present()) {
TraceEvent("MonitorProxiesForwarding")
.detail("NewConnStr", rep.get().read().forward.get().toString())
.detail("OldConnStr", info.intermediateConnFile->getConnectionString().toString());
info.intermediateConnFile = Reference<ClusterConnectionFile>(new ClusterConnectionFile(
connFile->getFilename(), ClusterConnectionString(rep.get().read().forward.get().toString())));
return info;
}
if (connFile != info.intermediateConnFile) {
if (!info.hasConnected) {
TraceEvent(SevWarnAlways, "IncorrectClusterFileContentsAtConnection")
.detail("Filename", connFile->getFilename())
.detail("ConnectionStringFromFile", connFile->getConnectionString().toString())
.detail("CurrentConnectionString", info.intermediateConnFile->getConnectionString().toString());
}
connFile->setConnectionString(info.intermediateConnFile->getConnectionString());
info.intermediateConnFile = connFile;
}
info.hasConnected = true;
connFile->notifyConnected();
auto& ni = rep.get().mutate();
shrinkProxyList(ni, lastCommitProxyUIDs, lastCommitProxies, lastGrvProxyUIDs, lastGrvProxies);
clientInfo->set(ni);
successIdx = idx;
} else {
TEST(rep.getError().code() == error_code_failed_to_progress); // Coordinator cannot talk to cluster controller
idx = (idx + 1) % addrs.size();
if (idx == successIdx) {
wait(delay(CLIENT_KNOBS->COORDINATOR_RECONNECTION_DELAY));
}
}
}
}
ACTOR Future<Void> monitorProxies(
Reference<AsyncVar<Reference<ClusterConnectionFile>>> connFile,
Reference<AsyncVar<ClientDBInfo>> clientInfo,
Reference<AsyncVar<Optional<ClientLeaderRegInterface>>> coordinator,
Reference<ReferencedObject<Standalone<VectorRef<ClientVersionRef>>>> supportedVersions,
Key traceLogGroup) {
state MonitorLeaderInfo info(connFile->get());
loop {
choose {
when(MonitorLeaderInfo _info = wait(monitorProxiesOneGeneration(
connFile->get(), clientInfo, coordinator, info, supportedVersions, traceLogGroup))) {
info = _info;
}
when(wait(connFile->onChange())) {
info.hasConnected = false;
info.intermediateConnFile = connFile->get();
}
}
}
}
Reference in New Issue View Git Blame Copy Permalink