test/cpp/end2end/client_lb_end2end_test.cc

   1 /*
   2  *
   3  * Copyright 2016 gRPC authors.
   4  *
   5  * Licensed under the Apache License, Version 2.0 (the "License");
   6  * you may not use this file except in compliance with the License.
   7  * You may obtain a copy of the License at
   8  *
   9  *     http://www.apache.org/licenses/LICENSE-2.0
  10  *
  11  * Unless required by applicable law or agreed to in writing, software
  12  * distributed under the License is distributed on an "AS IS" BASIS,
  13  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  14  * See the License for the specific language governing permissions and
  15  * limitations under the License.
  16  *
  17  */
  18
  19 #include <algorithm>
  20 #include <memory>
  21 #include <mutex>
  22 #include <random>
  23 #include <set>
  24 #include <thread>
  25
  26 #include <grpc/grpc.h>
  27 #include <grpc/support/alloc.h>
  28 #include <grpc/support/atm.h>
  29 #include <grpc/support/log.h>
  30 #include <grpc/support/string_util.h>
  31 #include <grpc/support/time.h>
  32 #include <grpcpp/channel.h>
  33 #include <grpcpp/client_context.h>
  34 #include <grpcpp/create_channel.h>
  35 #include <grpcpp/health_check_service_interface.h>
  36 #include <grpcpp/impl/codegen/sync.h>
  37 #include <grpcpp/server.h>
  38 #include <grpcpp/server_builder.h>
  39
  40 #include "src/core/ext/filters/client_channel/backup_poller.h"
  41 #include "src/core/ext/filters/client_channel/global_subchannel_pool.h"
  42 #include "src/core/ext/filters/client_channel/parse_address.h"
  43 #include "src/core/ext/filters/client_channel/resolver/fake/fake_resolver.h"
  44 #include "src/core/ext/filters/client_channel/server_address.h"
  45 #include "src/core/ext/filters/client_channel/service_config.h"
  46 #include "src/core/lib/backoff/backoff.h"
  47 #include "src/core/lib/channel/channel_args.h"
  48 #include "src/core/lib/gpr/env.h"
  49 #include "src/core/lib/gprpp/debug_location.h"
  50 #include "src/core/lib/gprpp/ref_counted_ptr.h"
  51 #include "src/core/lib/iomgr/tcp_client.h"
  52 #include "src/core/lib/security/credentials/fake/fake_credentials.h"
  53 #include "src/cpp/client/secure_credentials.h"
  54 #include "src/cpp/server/secure_server_credentials.h"
  55
  56 #include "src/proto/grpc/testing/echo.grpc.pb.h"
  57 #include "src/proto/grpc/testing/xds/orca_load_report_for_test.pb.h"
  58 #include "test/core/util/port.h"
  59 #include "test/core/util/test_config.h"
  60 #include "test/core/util/test_lb_policies.h"
  61 #include "test/cpp/end2end/test_service_impl.h"
  62
  63 #include <gmock/gmock.h>
  64 #include <gtest/gtest.h>
  65
  66 using grpc::testing::EchoRequest;
  67 using grpc::testing::EchoResponse;
  68 using std::chrono::system_clock;
  69
  70 // defined in tcp_client.cc
  71 extern grpc_tcp_client_vtable* grpc_tcp_client_impl;
  72
  73 static grpc_tcp_client_vtable* default_client_impl;
  74
  75 namespace grpc {
  76 namespace testing {
  77 namespace {
  78
  79 gpr_atm g_connection_delay_ms;
  80
  81 void tcp_client_connect_with_delay(grpc_closure* closure, grpc_endpoint** ep,
  82                                    grpc_pollset_set* interested_parties,
  83                                    const grpc_channel_args* channel_args,
  84                                    const grpc_resolved_address* addr,
  85                                    grpc_millis deadline) {
  86   const int delay_ms = gpr_atm_acq_load(&g_connection_delay_ms);
  87   if (delay_ms > 0) {
  88     gpr_sleep_until(grpc_timeout_milliseconds_to_deadline(delay_ms));
  89   }
  90   default_client_impl->connect(closure, ep, interested_parties, channel_args,
  91                                addr, deadline + delay_ms);
  92 }
  93
  94 grpc_tcp_client_vtable delayed_connect = {tcp_client_connect_with_delay};
  95
  96 // Subclass of TestServiceImpl that increments a request counter for
  97 // every call to the Echo RPC.
  98 class MyTestServiceImpl : public TestServiceImpl {
  99  public:
 100   Status Echo(ServerContext* context, const EchoRequest* request,
 101               EchoResponse* response) override {
 102     const udpa::data::orca::v1::OrcaLoadReport* load_report = nullptr;
 103     {
 104       grpc::internal::MutexLock lock(&mu_);
 105       ++request_count_;
 106       load_report = load_report_;
 107     }
 108     AddClient(context->peer());
 109     if (load_report != nullptr) {
 110       // TODO(roth): Once we provide a more standard server-side API for
 111       // populating this data, use that API here.
 112       context->AddTrailingMetadata("x-endpoint-load-metrics-bin",
 113                                    load_report->SerializeAsString());
 114     }
 115     return TestServiceImpl::Echo(context, request, response);
 116   }
 117
 118   int request_count() {
 119     grpc::internal::MutexLock lock(&mu_);
 120     return request_count_;
 121   }
 122
 123   void ResetCounters() {
 124     grpc::internal::MutexLock lock(&mu_);
 125     request_count_ = 0;
 126   }
 127
 128   std::set<grpc::string> clients() {
 129     grpc::internal::MutexLock lock(&clients_mu_);
 130     return clients_;
 131   }
 132
 133   void set_load_report(udpa::data::orca::v1::OrcaLoadReport* load_report) {
 134     grpc::internal::MutexLock lock(&mu_);
 135     load_report_ = load_report;
 136   }
 137
 138  private:
 139   void AddClient(const grpc::string& client) {
 140     grpc::internal::MutexLock lock(&clients_mu_);
 141     clients_.insert(client);
 142   }
 143
 144   grpc::internal::Mutex mu_;
 145   int request_count_ = 0;
 146   const udpa::data::orca::v1::OrcaLoadReport* load_report_ = nullptr;
 147   grpc::internal::Mutex clients_mu_;
 148   std::set<grpc::string> clients_;
 149 };
 150
 151 class FakeResolverResponseGeneratorWrapper {
 152  public:
 153   FakeResolverResponseGeneratorWrapper()
 154       : response_generator_(grpc_core::MakeRefCounted<
 155                             grpc_core::FakeResolverResponseGenerator>()) {}
 156
 157   FakeResolverResponseGeneratorWrapper(
 158       FakeResolverResponseGeneratorWrapper&& other) {
 159     response_generator_ = std::move(other.response_generator_);
 160   }
 161
 162   void SetNextResolution(const std::vector<int>& ports,
 163                          const char* service_config_json = nullptr) {
 164     grpc_core::ExecCtx exec_ctx;
 165     response_generator_->SetResponse(
 166         BuildFakeResults(ports, service_config_json));
 167   }
 168
 169   void SetNextResolutionUponError(const std::vector<int>& ports) {
 170     grpc_core::ExecCtx exec_ctx;
 171     response_generator_->SetReresolutionResponse(BuildFakeResults(ports));
 172   }
 173
 174   void SetFailureOnReresolution() {
 175     grpc_core::ExecCtx exec_ctx;
 176     response_generator_->SetFailureOnReresolution();
 177   }
 178
 179   grpc_core::FakeResolverResponseGenerator* Get() const {
 180     return response_generator_.get();
 181   }
 182
 183  private:
 184   static grpc_core::Resolver::Result BuildFakeResults(
 185       const std::vector<int>& ports,
 186       const char* service_config_json = nullptr) {
 187     grpc_core::Resolver::Result result;
 188     for (const int& port : ports) {
 189       char* lb_uri_str;
 190       gpr_asprintf(&lb_uri_str, "ipv4:127.0.0.1:%d", port);
 191       grpc_uri* lb_uri = grpc_uri_parse(lb_uri_str, true);
 192       GPR_ASSERT(lb_uri != nullptr);
 193       grpc_resolved_address address;
 194       GPR_ASSERT(grpc_parse_uri(lb_uri, &address));
 195       result.addresses.emplace_back(address.addr, address.len,
 196                                     nullptr /* args */);
 197       grpc_uri_destroy(lb_uri);
 198       gpr_free(lb_uri_str);
 199     }
 200     if (service_config_json != nullptr) {
 201       result.service_config = grpc_core::ServiceConfig::Create(
 202           service_config_json, &result.service_config_error);
 203       GPR_ASSERT(result.service_config != nullptr);
 204     }
 205     return result;
 206   }
 207
 208   grpc_core::RefCountedPtr<grpc_core::FakeResolverResponseGenerator>
 209       response_generator_;
 210 };
 211
 212 class ClientLbEnd2endTest : public ::testing::Test {
 213  protected:
 214   ClientLbEnd2endTest()
 215       : server_host_("localhost"),
 216         kRequestMessage_("Live long and prosper."),
 217         creds_(new SecureChannelCredentials(
 218             grpc_fake_transport_security_credentials_create())) {}
 219
 220   static void SetUpTestCase() {
 221     // Make the backup poller poll very frequently in order to pick up
 222     // updates from all the subchannels's FDs.
 223     GPR_GLOBAL_CONFIG_SET(grpc_client_channel_backup_poll_interval_ms, 1);
 224 #if TARGET_OS_IPHONE
 225     // Workaround Apple CFStream bug
 226     gpr_setenv("grpc_cfstream", "0");
 227 #endif
 228   }
 229
 230   void SetUp() override { grpc_init(); }
 231
 232   void TearDown() override {
 233     for (size_t i = 0; i < servers_.size(); ++i) {
 234       servers_[i]->Shutdown();
 235     }
 236     // Explicitly destroy all the members so that we can make sure grpc_shutdown
 237     // has finished by the end of this function, and thus all the registered
 238     // LB policy factories are removed.
 239     servers_.clear();
 240     creds_.reset();
 241     grpc_shutdown_blocking();
 242   }
 243
 244   void CreateServers(size_t num_servers,
 245                      std::vector<int> ports = std::vector<int>()) {
 246     servers_.clear();
 247     for (size_t i = 0; i < num_servers; ++i) {
 248       int port = 0;
 249       if (ports.size() == num_servers) port = ports[i];
 250       servers_.emplace_back(new ServerData(port));
 251     }
 252   }
 253
 254   void StartServer(size_t index) { servers_[index]->Start(server_host_); }
 255
 256   void StartServers(size_t num_servers,
 257                     std::vector<int> ports = std::vector<int>()) {
 258     CreateServers(num_servers, std::move(ports));
 259     for (size_t i = 0; i < num_servers; ++i) {
 260       StartServer(i);
 261     }
 262   }
 263
 264   std::vector<int> GetServersPorts(size_t start_index = 0) {
 265     std::vector<int> ports;
 266     for (size_t i = start_index; i < servers_.size(); ++i) {
 267       ports.push_back(servers_[i]->port_);
 268     }
 269     return ports;
 270   }
 271
 272   FakeResolverResponseGeneratorWrapper BuildResolverResponseGenerator() {
 273     return FakeResolverResponseGeneratorWrapper();
 274   }
 275
 276   std::unique_ptr<grpc::testing::EchoTestService::Stub> BuildStub(
 277       const std::shared_ptr<Channel>& channel) {
 278     return grpc::testing::EchoTestService::NewStub(channel);
 279   }
 280
 281   std::shared_ptr<Channel> BuildChannel(
 282       const grpc::string& lb_policy_name,
 283       const FakeResolverResponseGeneratorWrapper& response_generator,
 284       ChannelArguments args = ChannelArguments()) {
 285     if (lb_policy_name.size() > 0) {
 286       args.SetLoadBalancingPolicyName(lb_policy_name);
 287     }  // else, default to pick first
 288     args.SetPointer(GRPC_ARG_FAKE_RESOLVER_RESPONSE_GENERATOR,
 289                     response_generator.Get());
 290     return ::grpc::CreateCustomChannel("fake:///", creds_, args);
 291   }
 292
 293   bool SendRpc(
 294       const std::unique_ptr<grpc::testing::EchoTestService::Stub>& stub,
 295       EchoResponse* response = nullptr, int timeout_ms = 1000,
 296       Status* result = nullptr, bool wait_for_ready = false) {
 297     const bool local_response = (response == nullptr);
 298     if (local_response) response = new EchoResponse;
 299     EchoRequest request;
 300     request.set_message(kRequestMessage_);
 301     ClientContext context;
 302     context.set_deadline(grpc_timeout_milliseconds_to_deadline(timeout_ms));
 303     if (wait_for_ready) context.set_wait_for_ready(true);
 304     Status status = stub->Echo(&context, request, response);
 305     if (result != nullptr) *result = status;
 306     if (local_response) delete response;
 307     return status.ok();
 308   }
 309
 310   void CheckRpcSendOk(
 311       const std::unique_ptr<grpc::testing::EchoTestService::Stub>& stub,
 312       const grpc_core::DebugLocation& location, bool wait_for_ready = false) {
 313     EchoResponse response;
 314     Status status;
 315     const bool success =
 316         SendRpc(stub, &response, 2000, &status, wait_for_ready);
 317     ASSERT_TRUE(success) << "From " << location.file() << ":" << location.line()
 318                          << "\n"
 319                          << "Error: " << status.error_message() << " "
 320                          << status.error_details();
 321     ASSERT_EQ(response.message(), kRequestMessage_)
 322         << "From " << location.file() << ":" << location.line();
 323     if (!success) abort();
 324   }
 325
 326   void CheckRpcSendFailure(
 327       const std::unique_ptr<grpc::testing::EchoTestService::Stub>& stub) {
 328     const bool success = SendRpc(stub);
 329     EXPECT_FALSE(success);
 330   }
 331
 332   struct ServerData {
 333     int port_;
 334     std::unique_ptr<Server> server_;
 335     MyTestServiceImpl service_;
 336     std::unique_ptr<std::thread> thread_;
 337     bool server_ready_ = false;
 338     bool started_ = false;
 339
 340     explicit ServerData(int port = 0) {
 341       port_ = port > 0 ? port : grpc_pick_unused_port_or_die();
 342     }
 343
 344     void Start(const grpc::string& server_host) {
 345       gpr_log(GPR_INFO, "starting server on port %d", port_);
 346       started_ = true;
 347       grpc::internal::Mutex mu;
 348       grpc::internal::MutexLock lock(&mu);
 349       grpc::internal::CondVar cond;
 350       thread_.reset(new std::thread(
 351           std::bind(&ServerData::Serve, this, server_host, &mu, &cond)));
 352       cond.WaitUntil(&mu, [this] { return server_ready_; });
 353       server_ready_ = false;
 354       gpr_log(GPR_INFO, "server startup complete");
 355     }
 356
 357     void Serve(const grpc::string& server_host, grpc::internal::Mutex* mu,
 358                grpc::internal::CondVar* cond) {
 359       std::ostringstream server_address;
 360       server_address << server_host << ":" << port_;
 361       ServerBuilder builder;
 362       std::shared_ptr<ServerCredentials> creds(new SecureServerCredentials(
 363           grpc_fake_transport_security_server_credentials_create()));
 364       builder.AddListeningPort(server_address.str(), std::move(creds));
 365       builder.RegisterService(&service_);
 366       server_ = builder.BuildAndStart();
 367       grpc::internal::MutexLock lock(mu);
 368       server_ready_ = true;
 369       cond->Signal();
 370     }
 371
 372     void Shutdown() {
 373       if (!started_) return;
 374       server_->Shutdown(grpc_timeout_milliseconds_to_deadline(0));
 375       thread_->join();
 376       started_ = false;
 377     }
 378
 379     void SetServingStatus(const grpc::string& service, bool serving) {
 380       server_->GetHealthCheckService()->SetServingStatus(service, serving);
 381     }
 382   };
 383
 384   void ResetCounters() {
 385     for (const auto& server : servers_) server->service_.ResetCounters();
 386   }
 387
 388   void WaitForServer(
 389       const std::unique_ptr<grpc::testing::EchoTestService::Stub>& stub,
 390       size_t server_idx, const grpc_core::DebugLocation& location,
 391       bool ignore_failure = false) {
 392     do {
 393       if (ignore_failure) {
 394         SendRpc(stub);
 395       } else {
 396         CheckRpcSendOk(stub, location, true);
 397       }
 398     } while (servers_[server_idx]->service_.request_count() == 0);
 399     ResetCounters();
 400   }
 401
 402   bool WaitForChannelState(
 403       Channel* channel, std::function<bool(grpc_connectivity_state)> predicate,
 404       bool try_to_connect = false, int timeout_seconds = 5) {
 405     const gpr_timespec deadline =
 406         grpc_timeout_seconds_to_deadline(timeout_seconds);
 407     while (true) {
 408       grpc_connectivity_state state = channel->GetState(try_to_connect);
 409       if (predicate(state)) break;
 410       if (!channel->WaitForStateChange(state, deadline)) return false;
 411     }
 412     return true;
 413   }
 414
 415   bool WaitForChannelNotReady(Channel* channel, int timeout_seconds = 5) {
 416     auto predicate = [](grpc_connectivity_state state) {
 417       return state != GRPC_CHANNEL_READY;
 418     };
 419     return WaitForChannelState(channel, predicate, false, timeout_seconds);
 420   }
 421
 422   bool WaitForChannelReady(Channel* channel, int timeout_seconds = 5) {
 423     auto predicate = [](grpc_connectivity_state state) {
 424       return state == GRPC_CHANNEL_READY;
 425     };
 426     return WaitForChannelState(channel, predicate, true, timeout_seconds);
 427   }
 428
 429   bool SeenAllServers() {
 430     for (const auto& server : servers_) {
 431       if (server->service_.request_count() == 0) return false;
 432     }
 433     return true;
 434   }
 435
 436   // Updates \a connection_order by appending to it the index of the newly
 437   // connected server. Must be called after every single RPC.
 438   void UpdateConnectionOrder(
 439       const std::vector<std::unique_ptr<ServerData>>& servers,
 440       std::vector<int>* connection_order) {
 441     for (size_t i = 0; i < servers.size(); ++i) {
 442       if (servers[i]->service_.request_count() == 1) {
 443         // Was the server index known? If not, update connection_order.
 444         const auto it =
 445             std::find(connection_order->begin(), connection_order->end(), i);
 446         if (it == connection_order->end()) {
 447           connection_order->push_back(i);
 448           return;
 449         }
 450       }
 451     }
 452   }
 453
 454   const grpc::string server_host_;
 455   std::vector<std::unique_ptr<ServerData>> servers_;
 456   const grpc::string kRequestMessage_;
 457   std::shared_ptr<ChannelCredentials> creds_;
 458 };
 459
 460 TEST_F(ClientLbEnd2endTest, ChannelStateConnectingWhenResolving) {
 461   const int kNumServers = 3;
 462   StartServers(kNumServers);
 463   auto response_generator = BuildResolverResponseGenerator();
 464   auto channel = BuildChannel("", response_generator);
 465   auto stub = BuildStub(channel);
 466   // Initial state should be IDLE.
 467   EXPECT_EQ(channel->GetState(false /* try_to_connect */), GRPC_CHANNEL_IDLE);
 468   // Tell the channel to try to connect.
 469   // Note that this call also returns IDLE, since the state change has
 470   // not yet occurred; it just gets triggered by this call.
 471   EXPECT_EQ(channel->GetState(true /* try_to_connect */), GRPC_CHANNEL_IDLE);
 472   // Now that the channel is trying to connect, we should be in state
 473   // CONNECTING.
 474   EXPECT_EQ(channel->GetState(false /* try_to_connect */),
 475             GRPC_CHANNEL_CONNECTING);
 476   // Return a resolver result, which allows the connection attempt to proceed.
 477   response_generator.SetNextResolution(GetServersPorts());
 478   // We should eventually transition into state READY.
 479   EXPECT_TRUE(WaitForChannelReady(channel.get()));
 480 }
 481
 482 TEST_F(ClientLbEnd2endTest, PickFirst) {
 483   // Start servers and send one RPC per server.
 484   const int kNumServers = 3;
 485   StartServers(kNumServers);
 486   auto response_generator = BuildResolverResponseGenerator();
 487   auto channel = BuildChannel(
 488       "", response_generator);  // test that pick first is the default.
 489   auto stub = BuildStub(channel);
 490   response_generator.SetNextResolution(GetServersPorts());
 491   for (size_t i = 0; i < servers_.size(); ++i) {
 492     CheckRpcSendOk(stub, DEBUG_LOCATION);
 493   }
 494   // All requests should have gone to a single server.
 495   bool found = false;
 496   for (size_t i = 0; i < servers_.size(); ++i) {
 497     const int request_count = servers_[i]->service_.request_count();
 498     if (request_count == kNumServers) {
 499       found = true;
 500     } else {
 501       EXPECT_EQ(0, request_count);
 502     }
 503   }
 504   EXPECT_TRUE(found);
 505   // Check LB policy name for the channel.
 506   EXPECT_EQ("pick_first", channel->GetLoadBalancingPolicyName());
 507 }
 508
 509 TEST_F(ClientLbEnd2endTest, PickFirstProcessPending) {
 510   StartServers(1);  // Single server
 511   auto response_generator = BuildResolverResponseGenerator();
 512   auto channel = BuildChannel(
 513       "", response_generator);  // test that pick first is the default.
 514   auto stub = BuildStub(channel);
 515   response_generator.SetNextResolution({servers_[0]->port_});
 516   WaitForServer(stub, 0, DEBUG_LOCATION);
 517   // Create a new channel and its corresponding PF LB policy, which will pick
 518   // the subchannels in READY state from the previous RPC against the same
 519   // target (even if it happened over a different channel, because subchannels
 520   // are globally reused). Progress should happen without any transition from
 521   // this READY state.
 522   auto second_response_generator = BuildResolverResponseGenerator();
 523   auto second_channel = BuildChannel("", second_response_generator);
 524   auto second_stub = BuildStub(second_channel);
 525   second_response_generator.SetNextResolution({servers_[0]->port_});
 526   CheckRpcSendOk(second_stub, DEBUG_LOCATION);
 527 }
 528
 529 TEST_F(ClientLbEnd2endTest, PickFirstSelectsReadyAtStartup) {
 530   ChannelArguments args;
 531   constexpr int kInitialBackOffMs = 5000;
 532   args.SetInt(GRPC_ARG_INITIAL_RECONNECT_BACKOFF_MS, kInitialBackOffMs);
 533   // Create 2 servers, but start only the second one.
 534   std::vector<int> ports = {grpc_pick_unused_port_or_die(),
 535                             grpc_pick_unused_port_or_die()};
 536   CreateServers(2, ports);
 537   StartServer(1);
 538   auto response_generator1 = BuildResolverResponseGenerator();
 539   auto channel1 = BuildChannel("pick_first", response_generator1, args);
 540   auto stub1 = BuildStub(channel1);
 541   response_generator1.SetNextResolution(ports);
 542   // Wait for second server to be ready.
 543   WaitForServer(stub1, 1, DEBUG_LOCATION);
 544   // Create a second channel with the same addresses.  Its PF instance
 545   // should immediately pick the second subchannel, since it's already
 546   // in READY state.
 547   auto response_generator2 = BuildResolverResponseGenerator();
 548   auto channel2 = BuildChannel("pick_first", response_generator2, args);
 549   response_generator2.SetNextResolution(ports);
 550   // Check that the channel reports READY without waiting for the
 551   // initial backoff.
 552   EXPECT_TRUE(WaitForChannelReady(channel2.get(), 1 /* timeout_seconds */));
 553 }
 554
 555 TEST_F(ClientLbEnd2endTest, PickFirstBackOffInitialReconnect) {
 556   ChannelArguments args;
 557   constexpr int kInitialBackOffMs = 100;
 558   args.SetInt(GRPC_ARG_INITIAL_RECONNECT_BACKOFF_MS, kInitialBackOffMs);
 559   const std::vector<int> ports = {grpc_pick_unused_port_or_die()};
 560   const gpr_timespec t0 = gpr_now(GPR_CLOCK_MONOTONIC);
 561   auto response_generator = BuildResolverResponseGenerator();
 562   auto channel = BuildChannel("pick_first", response_generator, args);
 563   auto stub = BuildStub(channel);
 564   response_generator.SetNextResolution(ports);
 565   // The channel won't become connected (there's no server).
 566   ASSERT_FALSE(channel->WaitForConnected(
 567       grpc_timeout_milliseconds_to_deadline(kInitialBackOffMs * 2)));
 568   // Bring up a server on the chosen port.
 569   StartServers(1, ports);
 570   // Now it will.
 571   ASSERT_TRUE(channel->WaitForConnected(
 572       grpc_timeout_milliseconds_to_deadline(kInitialBackOffMs * 2)));
 573   const gpr_timespec t1 = gpr_now(GPR_CLOCK_MONOTONIC);
 574   const grpc_millis waited_ms = gpr_time_to_millis(gpr_time_sub(t1, t0));
 575   gpr_log(GPR_DEBUG, "Waited %" PRId64 " milliseconds", waited_ms);
 576   // We should have waited at least kInitialBackOffMs. We substract one to
 577   // account for test and precision accuracy drift.
 578   EXPECT_GE(waited_ms, kInitialBackOffMs - 1);
 579   // But not much more.
 580   EXPECT_GT(
 581       gpr_time_cmp(
 582           grpc_timeout_milliseconds_to_deadline(kInitialBackOffMs * 1.10), t1),
 583       0);
 584 }
 585
 586 TEST_F(ClientLbEnd2endTest, PickFirstBackOffMinReconnect) {
 587   ChannelArguments args;
 588   constexpr int kMinReconnectBackOffMs = 1000;
 589   args.SetInt(GRPC_ARG_MIN_RECONNECT_BACKOFF_MS, kMinReconnectBackOffMs);
 590   const std::vector<int> ports = {grpc_pick_unused_port_or_die()};
 591   auto response_generator = BuildResolverResponseGenerator();
 592   auto channel = BuildChannel("pick_first", response_generator, args);
 593   auto stub = BuildStub(channel);
 594   response_generator.SetNextResolution(ports);
 595   // Make connection delay a 10% longer than it's willing to in order to make
 596   // sure we are hitting the codepath that waits for the min reconnect backoff.
 597   gpr_atm_rel_store(&g_connection_delay_ms, kMinReconnectBackOffMs * 1.10);
 598   default_client_impl = grpc_tcp_client_impl;
 599   grpc_set_tcp_client_impl(&delayed_connect);
 600   const gpr_timespec t0 = gpr_now(GPR_CLOCK_MONOTONIC);
 601   channel->WaitForConnected(
 602       grpc_timeout_milliseconds_to_deadline(kMinReconnectBackOffMs * 2));
 603   const gpr_timespec t1 = gpr_now(GPR_CLOCK_MONOTONIC);
 604   const grpc_millis waited_ms = gpr_time_to_millis(gpr_time_sub(t1, t0));
 605   gpr_log(GPR_DEBUG, "Waited %" PRId64 " ms", waited_ms);
 606   // We should have waited at least kMinReconnectBackOffMs. We substract one to
 607   // account for test and precision accuracy drift.
 608   EXPECT_GE(waited_ms, kMinReconnectBackOffMs - 1);
 609   gpr_atm_rel_store(&g_connection_delay_ms, 0);
 610 }
 611
 612 TEST_F(ClientLbEnd2endTest, PickFirstResetConnectionBackoff) {
 613   ChannelArguments args;
 614   constexpr int kInitialBackOffMs = 1000;
 615   args.SetInt(GRPC_ARG_INITIAL_RECONNECT_BACKOFF_MS, kInitialBackOffMs);
 616   const std::vector<int> ports = {grpc_pick_unused_port_or_die()};
 617   auto response_generator = BuildResolverResponseGenerator();
 618   auto channel = BuildChannel("pick_first", response_generator, args);
 619   auto stub = BuildStub(channel);
 620   response_generator.SetNextResolution(ports);
 621   // The channel won't become connected (there's no server).
 622   EXPECT_FALSE(
 623       channel->WaitForConnected(grpc_timeout_milliseconds_to_deadline(10)));
 624   // Bring up a server on the chosen port.
 625   StartServers(1, ports);
 626   const gpr_timespec t0 = gpr_now(GPR_CLOCK_MONOTONIC);
 627   // Wait for connect, but not long enough.  This proves that we're
 628   // being throttled by initial backoff.
 629   EXPECT_FALSE(
 630       channel->WaitForConnected(grpc_timeout_milliseconds_to_deadline(10)));
 631   // Reset connection backoff.
 632   experimental::ChannelResetConnectionBackoff(channel.get());
 633   // Wait for connect.  Should happen ~immediately.
 634   EXPECT_TRUE(
 635       channel->WaitForConnected(grpc_timeout_milliseconds_to_deadline(10)));
 636   const gpr_timespec t1 = gpr_now(GPR_CLOCK_MONOTONIC);
 637   const grpc_millis waited_ms = gpr_time_to_millis(gpr_time_sub(t1, t0));
 638   gpr_log(GPR_DEBUG, "Waited %" PRId64 " milliseconds", waited_ms);
 639   // We should have waited less than kInitialBackOffMs.
 640   EXPECT_LT(waited_ms, kInitialBackOffMs);
 641 }
 642
 643 TEST_F(ClientLbEnd2endTest,
 644        PickFirstResetConnectionBackoffNextAttemptStartsImmediately) {
 645   ChannelArguments args;
 646   constexpr int kInitialBackOffMs = 1000;
 647   args.SetInt(GRPC_ARG_INITIAL_RECONNECT_BACKOFF_MS, kInitialBackOffMs);
 648   const std::vector<int> ports = {grpc_pick_unused_port_or_die()};
 649   auto response_generator = BuildResolverResponseGenerator();
 650   auto channel = BuildChannel("pick_first", response_generator, args);
 651   auto stub = BuildStub(channel);
 652   response_generator.SetNextResolution(ports);
 653   // Wait for connect, which should fail ~immediately, because the server
 654   // is not up.
 655   gpr_log(GPR_INFO, "=== INITIAL CONNECTION ATTEMPT");
 656   EXPECT_FALSE(
 657       channel->WaitForConnected(grpc_timeout_milliseconds_to_deadline(10)));
 658   // Reset connection backoff.
 659   // Note that the time at which the third attempt will be started is
 660   // actually computed at this point, so we record the start time here.
 661   gpr_log(GPR_INFO, "=== RESETTING BACKOFF");
 662   const gpr_timespec t0 = gpr_now(GPR_CLOCK_MONOTONIC);
 663   experimental::ChannelResetConnectionBackoff(channel.get());
 664   // Trigger a second connection attempt.  This should also fail
 665   // ~immediately, but the retry should be scheduled for
 666   // kInitialBackOffMs instead of applying the multiplier.
 667   gpr_log(GPR_INFO, "=== POLLING FOR SECOND CONNECTION ATTEMPT");
 668   EXPECT_FALSE(
 669       channel->WaitForConnected(grpc_timeout_milliseconds_to_deadline(10)));
 670   // Bring up a server on the chosen port.
 671   gpr_log(GPR_INFO, "=== STARTING BACKEND");
 672   StartServers(1, ports);
 673   // Wait for connect.  Should happen within kInitialBackOffMs.
 674   // Give an extra 100ms to account for the time spent in the second and
 675   // third connection attempts themselves (since what we really want to
 676   // measure is the time between the two).  As long as this is less than
 677   // the 1.6x increase we would see if the backoff state was not reset
 678   // properly, the test is still proving that the backoff was reset.
 679   constexpr int kWaitMs = kInitialBackOffMs + 100;
 680   gpr_log(GPR_INFO, "=== POLLING FOR THIRD CONNECTION ATTEMPT");
 681   EXPECT_TRUE(channel->WaitForConnected(
 682       grpc_timeout_milliseconds_to_deadline(kWaitMs)));
 683   const gpr_timespec t1 = gpr_now(GPR_CLOCK_MONOTONIC);
 684   const grpc_millis waited_ms = gpr_time_to_millis(gpr_time_sub(t1, t0));
 685   gpr_log(GPR_DEBUG, "Waited %" PRId64 " milliseconds", waited_ms);
 686   EXPECT_LT(waited_ms, kWaitMs);
 687 }
 688
 689 TEST_F(ClientLbEnd2endTest, PickFirstUpdates) {
 690   // Start servers and send one RPC per server.
 691   const int kNumServers = 3;
 692   StartServers(kNumServers);
 693   auto response_generator = BuildResolverResponseGenerator();
 694   auto channel = BuildChannel("pick_first", response_generator);
 695   auto stub = BuildStub(channel);
 696
 697   std::vector<int> ports;
 698
 699   // Perform one RPC against the first server.
 700   ports.emplace_back(servers_[0]->port_);
 701   response_generator.SetNextResolution(ports);
 702   gpr_log(GPR_INFO, "****** SET [0] *******");
 703   CheckRpcSendOk(stub, DEBUG_LOCATION);
 704   EXPECT_EQ(servers_[0]->service_.request_count(), 1);
 705
 706   // An empty update will result in the channel going into TRANSIENT_FAILURE.
 707   ports.clear();
 708   response_generator.SetNextResolution(ports);
 709   gpr_log(GPR_INFO, "****** SET none *******");
 710   grpc_connectivity_state channel_state;
 711   do {
 712     channel_state = channel->GetState(true /* try to connect */);
 713   } while (channel_state == GRPC_CHANNEL_READY);
 714   ASSERT_NE(channel_state, GRPC_CHANNEL_READY);
 715   servers_[0]->service_.ResetCounters();
 716
 717   // Next update introduces servers_[1], making the channel recover.
 718   ports.clear();
 719   ports.emplace_back(servers_[1]->port_);
 720   response_generator.SetNextResolution(ports);
 721   gpr_log(GPR_INFO, "****** SET [1] *******");
 722   WaitForServer(stub, 1, DEBUG_LOCATION);
 723   EXPECT_EQ(servers_[0]->service_.request_count(), 0);
 724
 725   // And again for servers_[2]
 726   ports.clear();
 727   ports.emplace_back(servers_[2]->port_);
 728   response_generator.SetNextResolution(ports);
 729   gpr_log(GPR_INFO, "****** SET [2] *******");
 730   WaitForServer(stub, 2, DEBUG_LOCATION);
 731   EXPECT_EQ(servers_[0]->service_.request_count(), 0);
 732   EXPECT_EQ(servers_[1]->service_.request_count(), 0);
 733
 734   // Check LB policy name for the channel.
 735   EXPECT_EQ("pick_first", channel->GetLoadBalancingPolicyName());
 736 }
 737
 738 TEST_F(ClientLbEnd2endTest, PickFirstUpdateSuperset) {
 739   // Start servers and send one RPC per server.
 740   const int kNumServers = 3;
 741   StartServers(kNumServers);
 742   auto response_generator = BuildResolverResponseGenerator();
 743   auto channel = BuildChannel("pick_first", response_generator);
 744   auto stub = BuildStub(channel);
 745
 746   std::vector<int> ports;
 747
 748   // Perform one RPC against the first server.
 749   ports.emplace_back(servers_[0]->port_);
 750   response_generator.SetNextResolution(ports);
 751   gpr_log(GPR_INFO, "****** SET [0] *******");
 752   CheckRpcSendOk(stub, DEBUG_LOCATION);
 753   EXPECT_EQ(servers_[0]->service_.request_count(), 1);
 754   servers_[0]->service_.ResetCounters();
 755
 756   // Send and superset update
 757   ports.clear();
 758   ports.emplace_back(servers_[1]->port_);
 759   ports.emplace_back(servers_[0]->port_);
 760   response_generator.SetNextResolution(ports);
 761   gpr_log(GPR_INFO, "****** SET superset *******");
 762   CheckRpcSendOk(stub, DEBUG_LOCATION);
 763   // We stick to the previously connected server.
 764   WaitForServer(stub, 0, DEBUG_LOCATION);
 765   EXPECT_EQ(0, servers_[1]->service_.request_count());
 766
 767   // Check LB policy name for the channel.
 768   EXPECT_EQ("pick_first", channel->GetLoadBalancingPolicyName());
 769 }
 770
 771 TEST_F(ClientLbEnd2endTest, PickFirstGlobalSubchannelPool) {
 772   // Start one server.
 773   const int kNumServers = 1;
 774   StartServers(kNumServers);
 775   std::vector<int> ports = GetServersPorts();
 776   // Create two channels that (by default) use the global subchannel pool.
 777   auto response_generator1 = BuildResolverResponseGenerator();
 778   auto channel1 = BuildChannel("pick_first", response_generator1);
 779   auto stub1 = BuildStub(channel1);
 780   response_generator1.SetNextResolution(ports);
 781   auto response_generator2 = BuildResolverResponseGenerator();
 782   auto channel2 = BuildChannel("pick_first", response_generator2);
 783   auto stub2 = BuildStub(channel2);
 784   response_generator2.SetNextResolution(ports);
 785   WaitForServer(stub1, 0, DEBUG_LOCATION);
 786   // Send one RPC on each channel.
 787   CheckRpcSendOk(stub1, DEBUG_LOCATION);
 788   CheckRpcSendOk(stub2, DEBUG_LOCATION);
 789   // The server receives two requests.
 790   EXPECT_EQ(2, servers_[0]->service_.request_count());
 791   // The two requests are from the same client port, because the two channels
 792   // share subchannels via the global subchannel pool.
 793   EXPECT_EQ(1UL, servers_[0]->service_.clients().size());
 794 }
 795
 796 TEST_F(ClientLbEnd2endTest, PickFirstLocalSubchannelPool) {
 797   // Start one server.
 798   const int kNumServers = 1;
 799   StartServers(kNumServers);
 800   std::vector<int> ports = GetServersPorts();
 801   // Create two channels that use local subchannel pool.
 802   ChannelArguments args;
 803   args.SetInt(GRPC_ARG_USE_LOCAL_SUBCHANNEL_POOL, 1);
 804   auto response_generator1 = BuildResolverResponseGenerator();
 805   auto channel1 = BuildChannel("pick_first", response_generator1, args);
 806   auto stub1 = BuildStub(channel1);
 807   response_generator1.SetNextResolution(ports);
 808   auto response_generator2 = BuildResolverResponseGenerator();
 809   auto channel2 = BuildChannel("pick_first", response_generator2, args);
 810   auto stub2 = BuildStub(channel2);
 811   response_generator2.SetNextResolution(ports);
 812   WaitForServer(stub1, 0, DEBUG_LOCATION);
 813   // Send one RPC on each channel.
 814   CheckRpcSendOk(stub1, DEBUG_LOCATION);
 815   CheckRpcSendOk(stub2, DEBUG_LOCATION);
 816   // The server receives two requests.
 817   EXPECT_EQ(2, servers_[0]->service_.request_count());
 818   // The two requests are from two client ports, because the two channels didn't
 819   // share subchannels with each other.
 820   EXPECT_EQ(2UL, servers_[0]->service_.clients().size());
 821 }
 822
 823 TEST_F(ClientLbEnd2endTest, PickFirstManyUpdates) {
 824   const int kNumUpdates = 1000;
 825   const int kNumServers = 3;
 826   StartServers(kNumServers);
 827   auto response_generator = BuildResolverResponseGenerator();
 828   auto channel = BuildChannel("pick_first", response_generator);
 829   auto stub = BuildStub(channel);
 830   std::vector<int> ports = GetServersPorts();
 831   for (size_t i = 0; i < kNumUpdates; ++i) {
 832     std::shuffle(ports.begin(), ports.end(),
 833                  std::mt19937(std::random_device()()));
 834     response_generator.SetNextResolution(ports);
 835     // We should re-enter core at the end of the loop to give the resolution
 836     // setting closure a chance to run.
 837     if ((i + 1) % 10 == 0) CheckRpcSendOk(stub, DEBUG_LOCATION);
 838   }
 839   // Check LB policy name for the channel.
 840   EXPECT_EQ("pick_first", channel->GetLoadBalancingPolicyName());
 841 }
 842
 843 TEST_F(ClientLbEnd2endTest, PickFirstReresolutionNoSelected) {
 844   // Prepare the ports for up servers and down servers.
 845   const int kNumServers = 3;
 846   const int kNumAliveServers = 1;
 847   StartServers(kNumAliveServers);
 848   std::vector<int> alive_ports, dead_ports;
 849   for (size_t i = 0; i < kNumServers; ++i) {
 850     if (i < kNumAliveServers) {
 851       alive_ports.emplace_back(servers_[i]->port_);
 852     } else {
 853       dead_ports.emplace_back(grpc_pick_unused_port_or_die());
 854     }
 855   }
 856   auto response_generator = BuildResolverResponseGenerator();
 857   auto channel = BuildChannel("pick_first", response_generator);
 858   auto stub = BuildStub(channel);
 859   // The initial resolution only contains dead ports. There won't be any
 860   // selected subchannel. Re-resolution will return the same result.
 861   response_generator.SetNextResolution(dead_ports);
 862   gpr_log(GPR_INFO, "****** INITIAL RESOLUTION SET *******");
 863   for (size_t i = 0; i < 10; ++i) CheckRpcSendFailure(stub);
 864   // Set a re-resolution result that contains reachable ports, so that the
 865   // pick_first LB policy can recover soon.
 866   response_generator.SetNextResolutionUponError(alive_ports);
 867   gpr_log(GPR_INFO, "****** RE-RESOLUTION SET *******");
 868   WaitForServer(stub, 0, DEBUG_LOCATION, true /* ignore_failure */);
 869   CheckRpcSendOk(stub, DEBUG_LOCATION);
 870   EXPECT_EQ(servers_[0]->service_.request_count(), 1);
 871   // Check LB policy name for the channel.
 872   EXPECT_EQ("pick_first", channel->GetLoadBalancingPolicyName());
 873 }
 874
 875 TEST_F(ClientLbEnd2endTest, PickFirstReconnectWithoutNewResolverResult) {
 876   std::vector<int> ports = {grpc_pick_unused_port_or_die()};
 877   StartServers(1, ports);
 878   auto response_generator = BuildResolverResponseGenerator();
 879   auto channel = BuildChannel("pick_first", response_generator);
 880   auto stub = BuildStub(channel);
 881   response_generator.SetNextResolution(ports);
 882   gpr_log(GPR_INFO, "****** INITIAL CONNECTION *******");
 883   WaitForServer(stub, 0, DEBUG_LOCATION);
 884   gpr_log(GPR_INFO, "****** STOPPING SERVER ******");
 885   servers_[0]->Shutdown();
 886   EXPECT_TRUE(WaitForChannelNotReady(channel.get()));
 887   gpr_log(GPR_INFO, "****** RESTARTING SERVER ******");
 888   StartServers(1, ports);
 889   WaitForServer(stub, 0, DEBUG_LOCATION);
 890 }
 891
 892 TEST_F(ClientLbEnd2endTest,
 893        PickFirstReconnectWithoutNewResolverResultStartsFromTopOfList) {
 894   std::vector<int> ports = {grpc_pick_unused_port_or_die(),
 895                             grpc_pick_unused_port_or_die()};
 896   CreateServers(2, ports);
 897   StartServer(1);
 898   auto response_generator = BuildResolverResponseGenerator();
 899   auto channel = BuildChannel("pick_first", response_generator);
 900   auto stub = BuildStub(channel);
 901   response_generator.SetNextResolution(ports);
 902   gpr_log(GPR_INFO, "****** INITIAL CONNECTION *******");
 903   WaitForServer(stub, 1, DEBUG_LOCATION);
 904   gpr_log(GPR_INFO, "****** STOPPING SERVER ******");
 905   servers_[1]->Shutdown();
 906   EXPECT_TRUE(WaitForChannelNotReady(channel.get()));
 907   gpr_log(GPR_INFO, "****** STARTING BOTH SERVERS ******");
 908   StartServers(2, ports);
 909   WaitForServer(stub, 0, DEBUG_LOCATION);
 910 }
 911
 912 TEST_F(ClientLbEnd2endTest, PickFirstCheckStateBeforeStartWatch) {
 913   std::vector<int> ports = {grpc_pick_unused_port_or_die()};
 914   StartServers(1, ports);
 915   auto response_generator = BuildResolverResponseGenerator();
 916   auto channel_1 = BuildChannel("pick_first", response_generator);
 917   auto stub_1 = BuildStub(channel_1);
 918   response_generator.SetNextResolution(ports);
 919   gpr_log(GPR_INFO, "****** RESOLUTION SET FOR CHANNEL 1 *******");
 920   WaitForServer(stub_1, 0, DEBUG_LOCATION);
 921   gpr_log(GPR_INFO, "****** CHANNEL 1 CONNECTED *******");
 922   servers_[0]->Shutdown();
 923   // Channel 1 will receive a re-resolution containing the same server. It will
 924   // create a new subchannel and hold a ref to it.
 925   StartServers(1, ports);
 926   gpr_log(GPR_INFO, "****** SERVER RESTARTED *******");
 927   auto response_generator_2 = BuildResolverResponseGenerator();
 928   auto channel_2 = BuildChannel("pick_first", response_generator_2);
 929   auto stub_2 = BuildStub(channel_2);
 930   response_generator_2.SetNextResolution(ports);
 931   gpr_log(GPR_INFO, "****** RESOLUTION SET FOR CHANNEL 2 *******");
 932   WaitForServer(stub_2, 0, DEBUG_LOCATION, true);
 933   gpr_log(GPR_INFO, "****** CHANNEL 2 CONNECTED *******");
 934   servers_[0]->Shutdown();
 935   // Wait until the disconnection has triggered the connectivity notification.
 936   // Otherwise, the subchannel may be picked for next call but will fail soon.
 937   EXPECT_TRUE(WaitForChannelNotReady(channel_2.get()));
 938   // Channel 2 will also receive a re-resolution containing the same server.
 939   // Both channels will ref the same subchannel that failed.
 940   StartServers(1, ports);
 941   gpr_log(GPR_INFO, "****** SERVER RESTARTED AGAIN *******");
 942   gpr_log(GPR_INFO, "****** CHANNEL 2 STARTING A CALL *******");
 943   // The first call after the server restart will succeed.
 944   CheckRpcSendOk(stub_2, DEBUG_LOCATION);
 945   gpr_log(GPR_INFO, "****** CHANNEL 2 FINISHED A CALL *******");
 946   // Check LB policy name for the channel.
 947   EXPECT_EQ("pick_first", channel_1->GetLoadBalancingPolicyName());
 948   // Check LB policy name for the channel.
 949   EXPECT_EQ("pick_first", channel_2->GetLoadBalancingPolicyName());
 950 }
 951
 952 TEST_F(ClientLbEnd2endTest, PickFirstIdleOnDisconnect) {
 953   // Start server, send RPC, and make sure channel is READY.
 954   const int kNumServers = 1;
 955   StartServers(kNumServers);
 956   auto response_generator = BuildResolverResponseGenerator();
 957   auto channel =
 958       BuildChannel("", response_generator);  // pick_first is the default.
 959   auto stub = BuildStub(channel);
 960   response_generator.SetNextResolution(GetServersPorts());
 961   CheckRpcSendOk(stub, DEBUG_LOCATION);
 962   EXPECT_EQ(channel->GetState(false), GRPC_CHANNEL_READY);
 963   // Stop server.  Channel should go into state IDLE.
 964   response_generator.SetFailureOnReresolution();
 965   servers_[0]->Shutdown();
 966   EXPECT_TRUE(WaitForChannelNotReady(channel.get()));
 967   EXPECT_EQ(channel->GetState(false), GRPC_CHANNEL_IDLE);
 968   servers_.clear();
 969 }
 970
 971 TEST_F(ClientLbEnd2endTest, PickFirstPendingUpdateAndSelectedSubchannelFails) {
 972   auto response_generator = BuildResolverResponseGenerator();
 973   auto channel =
 974       BuildChannel("", response_generator);  // pick_first is the default.
 975   auto stub = BuildStub(channel);
 976   // Create a number of servers, but only start 1 of them.
 977   CreateServers(10);
 978   StartServer(0);
 979   // Initially resolve to first server and make sure it connects.
 980   gpr_log(GPR_INFO, "Phase 1: Connect to first server.");
 981   response_generator.SetNextResolution({servers_[0]->port_});
 982   CheckRpcSendOk(stub, DEBUG_LOCATION, true /* wait_for_ready */);
 983   EXPECT_EQ(channel->GetState(false), GRPC_CHANNEL_READY);
 984   // Send a resolution update with the remaining servers, none of which are
 985   // running yet, so the update will stay pending.  Note that it's important
 986   // to have multiple servers here, or else the test will be flaky; with only
 987   // one server, the pending subchannel list has already gone into
 988   // TRANSIENT_FAILURE due to hitting the end of the list by the time we
 989   // check the state.
 990   gpr_log(GPR_INFO,
 991           "Phase 2: Resolver update pointing to remaining "
 992           "(not started) servers.");
 993   response_generator.SetNextResolution(GetServersPorts(1 /* start_index */));
 994   // RPCs will continue to be sent to the first server.
 995   CheckRpcSendOk(stub, DEBUG_LOCATION);
 996   // Now stop the first server, so that the current subchannel list
 997   // fails.  This should cause us to immediately swap over to the
 998   // pending list, even though it's not yet connected.  The state should
 999   // be set to CONNECTING, since that's what the pending subchannel list
1000   // was doing when we swapped over.
1001   gpr_log(GPR_INFO, "Phase 3: Stopping first server.");
1002   servers_[0]->Shutdown();
1003   WaitForChannelNotReady(channel.get());
1004   // TODO(roth): This should always return CONNECTING, but it's flaky
1005   // between that and TRANSIENT_FAILURE.  I suspect that this problem
1006   // will go away once we move the backoff code out of the subchannel
1007   // and into the LB policies.
1008   EXPECT_THAT(channel->GetState(false),
1009               ::testing::AnyOf(GRPC_CHANNEL_CONNECTING,
1010                                GRPC_CHANNEL_TRANSIENT_FAILURE));
1011   // Now start the second server.
1012   gpr_log(GPR_INFO, "Phase 4: Starting second server.");
1013   StartServer(1);
1014   // The channel should go to READY state and RPCs should go to the
1015   // second server.
1016   WaitForChannelReady(channel.get());
1017   WaitForServer(stub, 1, DEBUG_LOCATION, true /* ignore_failure */);
1018 }
1019
1020 TEST_F(ClientLbEnd2endTest, PickFirstStaysIdleUponEmptyUpdate) {
1021   // Start server, send RPC, and make sure channel is READY.
1022   const int kNumServers = 1;
1023   StartServers(kNumServers);
1024   auto response_generator = BuildResolverResponseGenerator();
1025   auto channel =
1026       BuildChannel("", response_generator);  // pick_first is the default.
1027   auto stub = BuildStub(channel);
1028   response_generator.SetNextResolution(GetServersPorts());
1029   CheckRpcSendOk(stub, DEBUG_LOCATION);
1030   EXPECT_EQ(channel->GetState(false), GRPC_CHANNEL_READY);
1031   // Stop server.  Channel should go into state IDLE.
1032   servers_[0]->Shutdown();
1033   EXPECT_TRUE(WaitForChannelNotReady(channel.get()));
1034   EXPECT_EQ(channel->GetState(false), GRPC_CHANNEL_IDLE);
1035   // Now send resolver update that includes no addresses.  Channel
1036   // should stay in state IDLE.
1037   response_generator.SetNextResolution({});
1038   EXPECT_FALSE(channel->WaitForStateChange(
1039       GRPC_CHANNEL_IDLE, grpc_timeout_seconds_to_deadline(3)));
1040   // Now bring the backend back up and send a non-empty resolver update,
1041   // and then try to send an RPC.  Channel should go back into state READY.
1042   StartServer(0);
1043   response_generator.SetNextResolution(GetServersPorts());
1044   CheckRpcSendOk(stub, DEBUG_LOCATION);
1045   EXPECT_EQ(channel->GetState(false), GRPC_CHANNEL_READY);
1046 }
1047
1048 TEST_F(ClientLbEnd2endTest, RoundRobin) {
1049   // Start servers and send one RPC per server.
1050   const int kNumServers = 3;
1051   StartServers(kNumServers);
1052   auto response_generator = BuildResolverResponseGenerator();
1053   auto channel = BuildChannel("round_robin", response_generator);
1054   auto stub = BuildStub(channel);
1055   response_generator.SetNextResolution(GetServersPorts());
1056   // Wait until all backends are ready.
1057   do {
1058     CheckRpcSendOk(stub, DEBUG_LOCATION);
1059   } while (!SeenAllServers());
1060   ResetCounters();
1061   // "Sync" to the end of the list. Next sequence of picks will start at the
1062   // first server (index 0).
1063   WaitForServer(stub, servers_.size() - 1, DEBUG_LOCATION);
1064   std::vector<int> connection_order;
1065   for (size_t i = 0; i < servers_.size(); ++i) {
1066     CheckRpcSendOk(stub, DEBUG_LOCATION);
1067     UpdateConnectionOrder(servers_, &connection_order);
1068   }
1069   // Backends should be iterated over in the order in which the addresses were
1070   // given.
1071   const auto expected = std::vector<int>{0, 1, 2};
1072   EXPECT_EQ(expected, connection_order);
1073   // Check LB policy name for the channel.
1074   EXPECT_EQ("round_robin", channel->GetLoadBalancingPolicyName());
1075 }
1076
1077 TEST_F(ClientLbEnd2endTest, RoundRobinProcessPending) {
1078   StartServers(1);  // Single server
1079   auto response_generator = BuildResolverResponseGenerator();
1080   auto channel = BuildChannel("round_robin", response_generator);
1081   auto stub = BuildStub(channel);
1082   response_generator.SetNextResolution({servers_[0]->port_});
1083   WaitForServer(stub, 0, DEBUG_LOCATION);
1084   // Create a new channel and its corresponding RR LB policy, which will pick
1085   // the subchannels in READY state from the previous RPC against the same
1086   // target (even if it happened over a different channel, because subchannels
1087   // are globally reused). Progress should happen without any transition from
1088   // this READY state.
1089   auto second_response_generator = BuildResolverResponseGenerator();
1090   auto second_channel = BuildChannel("round_robin", second_response_generator);
1091   auto second_stub = BuildStub(second_channel);
1092   second_response_generator.SetNextResolution({servers_[0]->port_});
1093   CheckRpcSendOk(second_stub, DEBUG_LOCATION);
1094 }
1095
1096 TEST_F(ClientLbEnd2endTest, RoundRobinUpdates) {
1097   // Start servers and send one RPC per server.
1098   const int kNumServers = 3;
1099   StartServers(kNumServers);
1100   auto response_generator = BuildResolverResponseGenerator();
1101   auto channel = BuildChannel("round_robin", response_generator);
1102   auto stub = BuildStub(channel);
1103   std::vector<int> ports;
1104   // Start with a single server.
1105   gpr_log(GPR_INFO, "*** FIRST BACKEND ***");
1106   ports.emplace_back(servers_[0]->port_);
1107   response_generator.SetNextResolution(ports);
1108   WaitForServer(stub, 0, DEBUG_LOCATION);
1109   // Send RPCs. They should all go servers_[0]
1110   for (size_t i = 0; i < 10; ++i) CheckRpcSendOk(stub, DEBUG_LOCATION);
1111   EXPECT_EQ(10, servers_[0]->service_.request_count());
1112   EXPECT_EQ(0, servers_[1]->service_.request_count());
1113   EXPECT_EQ(0, servers_[2]->service_.request_count());
1114   servers_[0]->service_.ResetCounters();
1115   // And now for the second server.
1116   gpr_log(GPR_INFO, "*** SECOND BACKEND ***");
1117   ports.clear();
1118   ports.emplace_back(servers_[1]->port_);
1119   response_generator.SetNextResolution(ports);
1120   // Wait until update has been processed, as signaled by the second backend
1121   // receiving a request.
1122   EXPECT_EQ(0, servers_[1]->service_.request_count());
1123   WaitForServer(stub, 1, DEBUG_LOCATION);
1124   for (size_t i = 0; i < 10; ++i) CheckRpcSendOk(stub, DEBUG_LOCATION);
1125   EXPECT_EQ(0, servers_[0]->service_.request_count());
1126   EXPECT_EQ(10, servers_[1]->service_.request_count());
1127   EXPECT_EQ(0, servers_[2]->service_.request_count());
1128   servers_[1]->service_.ResetCounters();
1129   // ... and for the last server.
1130   gpr_log(GPR_INFO, "*** THIRD BACKEND ***");
1131   ports.clear();
1132   ports.emplace_back(servers_[2]->port_);
1133   response_generator.SetNextResolution(ports);
1134   WaitForServer(stub, 2, DEBUG_LOCATION);
1135   for (size_t i = 0; i < 10; ++i) CheckRpcSendOk(stub, DEBUG_LOCATION);
1136   EXPECT_EQ(0, servers_[0]->service_.request_count());
1137   EXPECT_EQ(0, servers_[1]->service_.request_count());
1138   EXPECT_EQ(10, servers_[2]->service_.request_count());
1139   servers_[2]->service_.ResetCounters();
1140   // Back to all servers.
1141   gpr_log(GPR_INFO, "*** ALL BACKENDS ***");
1142   ports.clear();
1143   ports.emplace_back(servers_[0]->port_);
1144   ports.emplace_back(servers_[1]->port_);
1145   ports.emplace_back(servers_[2]->port_);
1146   response_generator.SetNextResolution(ports);
1147   WaitForServer(stub, 0, DEBUG_LOCATION);
1148   WaitForServer(stub, 1, DEBUG_LOCATION);
1149   WaitForServer(stub, 2, DEBUG_LOCATION);
1150   // Send three RPCs, one per server.
1151   for (size_t i = 0; i < 3; ++i) CheckRpcSendOk(stub, DEBUG_LOCATION);
1152   EXPECT_EQ(1, servers_[0]->service_.request_count());
1153   EXPECT_EQ(1, servers_[1]->service_.request_count());
1154   EXPECT_EQ(1, servers_[2]->service_.request_count());
1155   // An empty update will result in the channel going into TRANSIENT_FAILURE.
1156   gpr_log(GPR_INFO, "*** NO BACKENDS ***");
1157   ports.clear();
1158   response_generator.SetNextResolution(ports);
1159   grpc_connectivity_state channel_state;
1160   do {
1161     channel_state = channel->GetState(true /* try to connect */);
1162   } while (channel_state == GRPC_CHANNEL_READY);
1163   ASSERT_NE(channel_state, GRPC_CHANNEL_READY);
1164   servers_[0]->service_.ResetCounters();
1165   // Next update introduces servers_[1], making the channel recover.
1166   gpr_log(GPR_INFO, "*** BACK TO SECOND BACKEND ***");
1167   ports.clear();
1168   ports.emplace_back(servers_[1]->port_);
1169   response_generator.SetNextResolution(ports);
1170   WaitForServer(stub, 1, DEBUG_LOCATION);
1171   channel_state = channel->GetState(false /* try to connect */);
1172   ASSERT_EQ(channel_state, GRPC_CHANNEL_READY);
1173   // Check LB policy name for the channel.
1174   EXPECT_EQ("round_robin", channel->GetLoadBalancingPolicyName());
1175 }
1176
1177 TEST_F(ClientLbEnd2endTest, RoundRobinUpdateInError) {
1178   const int kNumServers = 3;
1179   StartServers(kNumServers);
1180   auto response_generator = BuildResolverResponseGenerator();
1181   auto channel = BuildChannel("round_robin", response_generator);
1182   auto stub = BuildStub(channel);
1183   std::vector<int> ports;
1184   // Start with a single server.
1185   ports.emplace_back(servers_[0]->port_);
1186   response_generator.SetNextResolution(ports);
1187   WaitForServer(stub, 0, DEBUG_LOCATION);
1188   // Send RPCs. They should all go to servers_[0]
1189   for (size_t i = 0; i < 10; ++i) SendRpc(stub);
1190   EXPECT_EQ(10, servers_[0]->service_.request_count());
1191   EXPECT_EQ(0, servers_[1]->service_.request_count());
1192   EXPECT_EQ(0, servers_[2]->service_.request_count());
1193   servers_[0]->service_.ResetCounters();
1194   // Shutdown one of the servers to be sent in the update.
1195   servers_[1]->Shutdown();
1196   ports.emplace_back(servers_[1]->port_);
1197   ports.emplace_back(servers_[2]->port_);
1198   response_generator.SetNextResolution(ports);
1199   WaitForServer(stub, 0, DEBUG_LOCATION);
1200   WaitForServer(stub, 2, DEBUG_LOCATION);
1201   // Send three RPCs, one per server.
1202   for (size_t i = 0; i < kNumServers; ++i) SendRpc(stub);
1203   // The server in shutdown shouldn't receive any.
1204   EXPECT_EQ(0, servers_[1]->service_.request_count());
1205 }
1206
1207 TEST_F(ClientLbEnd2endTest, RoundRobinManyUpdates) {
1208   // Start servers and send one RPC per server.
1209   const int kNumServers = 3;
1210   StartServers(kNumServers);
1211   auto response_generator = BuildResolverResponseGenerator();
1212   auto channel = BuildChannel("round_robin", response_generator);
1213   auto stub = BuildStub(channel);
1214   std::vector<int> ports = GetServersPorts();
1215   for (size_t i = 0; i < 1000; ++i) {
1216     std::shuffle(ports.begin(), ports.end(),
1217                  std::mt19937(std::random_device()()));
1218     response_generator.SetNextResolution(ports);
1219     if (i % 10 == 0) CheckRpcSendOk(stub, DEBUG_LOCATION);
1220   }
1221   // Check LB policy name for the channel.
1222   EXPECT_EQ("round_robin", channel->GetLoadBalancingPolicyName());
1223 }
1224
1225 TEST_F(ClientLbEnd2endTest, RoundRobinConcurrentUpdates) {
1226   // TODO(dgq): replicate the way internal testing exercises the concurrent
1227   // update provisions of RR.
1228 }
1229
1230 TEST_F(ClientLbEnd2endTest, RoundRobinReresolve) {
1231   // Start servers and send one RPC per server.
1232   const int kNumServers = 3;
1233   std::vector<int> first_ports;
1234   std::vector<int> second_ports;
1235   first_ports.reserve(kNumServers);
1236   for (int i = 0; i < kNumServers; ++i) {
1237     first_ports.push_back(grpc_pick_unused_port_or_die());
1238   }
1239   second_ports.reserve(kNumServers);
1240   for (int i = 0; i < kNumServers; ++i) {
1241     second_ports.push_back(grpc_pick_unused_port_or_die());
1242   }
1243   StartServers(kNumServers, first_ports);
1244   auto response_generator = BuildResolverResponseGenerator();
1245   auto channel = BuildChannel("round_robin", response_generator);
1246   auto stub = BuildStub(channel);
1247   response_generator.SetNextResolution(first_ports);
1248   // Send a number of RPCs, which succeed.
1249   for (size_t i = 0; i < 100; ++i) {
1250     CheckRpcSendOk(stub, DEBUG_LOCATION);
1251   }
1252   // Kill all servers
1253   gpr_log(GPR_INFO, "****** ABOUT TO KILL SERVERS *******");
1254   for (size_t i = 0; i < servers_.size(); ++i) {
1255     servers_[i]->Shutdown();
1256   }
1257   gpr_log(GPR_INFO, "****** SERVERS KILLED *******");
1258   gpr_log(GPR_INFO, "****** SENDING DOOMED REQUESTS *******");
1259   // Client requests should fail. Send enough to tickle all subchannels.
1260   for (size_t i = 0; i < servers_.size(); ++i) CheckRpcSendFailure(stub);
1261   gpr_log(GPR_INFO, "****** DOOMED REQUESTS SENT *******");
1262   // Bring servers back up on a different set of ports. We need to do this to be
1263   // sure that the eventual success is *not* due to subchannel reconnection
1264   // attempts and that an actual re-resolution has happened as a result of the
1265   // RR policy going into transient failure when all its subchannels become
1266   // unavailable (in transient failure as well).
1267   gpr_log(GPR_INFO, "****** RESTARTING SERVERS *******");
1268   StartServers(kNumServers, second_ports);
1269   // Don't notify of the update. Wait for the LB policy's re-resolution to
1270   // "pull" the new ports.
1271   response_generator.SetNextResolutionUponError(second_ports);
1272   gpr_log(GPR_INFO, "****** SERVERS RESTARTED *******");
1273   gpr_log(GPR_INFO, "****** SENDING REQUEST TO SUCCEED *******");
1274   // Client request should eventually (but still fairly soon) succeed.
1275   const gpr_timespec deadline = grpc_timeout_seconds_to_deadline(5);
1276   gpr_timespec now = gpr_now(GPR_CLOCK_MONOTONIC);
1277   while (gpr_time_cmp(deadline, now) > 0) {
1278     if (SendRpc(stub)) break;
1279     now = gpr_now(GPR_CLOCK_MONOTONIC);
1280   }
1281   ASSERT_GT(gpr_time_cmp(deadline, now), 0);
1282 }
1283
1284 TEST_F(ClientLbEnd2endTest, RoundRobinTransientFailure) {
1285   // Start servers and create channel.  Channel should go to READY state.
1286   const int kNumServers = 3;
1287   StartServers(kNumServers);
1288   auto response_generator = BuildResolverResponseGenerator();
1289   auto channel = BuildChannel("round_robin", response_generator);
1290   auto stub = BuildStub(channel);
1291   response_generator.SetNextResolution(GetServersPorts());
1292   EXPECT_TRUE(WaitForChannelReady(channel.get()));
1293   // Now kill the servers.  The channel should transition to TRANSIENT_FAILURE.
1294   // TODO(roth): This test should ideally check that even when the
1295   // subchannels are in state CONNECTING for an extended period of time,
1296   // we will still report TRANSIENT_FAILURE.  Unfortunately, we don't
1297   // currently have a good way to get a subchannel to report CONNECTING
1298   // for a long period of time, since the servers in this test framework
1299   // are on the loopback interface, which will immediately return a
1300   // "Connection refused" error, so the subchannels will only be in
1301   // CONNECTING state very briefly.  When we have time, see if we can
1302   // find a way to fix this.
1303   for (size_t i = 0; i < servers_.size(); ++i) {
1304     servers_[i]->Shutdown();
1305   }
1306   auto predicate = [](grpc_connectivity_state state) {
1307     return state == GRPC_CHANNEL_TRANSIENT_FAILURE;
1308   };
1309   EXPECT_TRUE(WaitForChannelState(channel.get(), predicate));
1310 }
1311
1312 TEST_F(ClientLbEnd2endTest, RoundRobinTransientFailureAtStartup) {
1313   // Create channel and return servers that don't exist.  Channel should
1314   // quickly transition into TRANSIENT_FAILURE.
1315   // TODO(roth): This test should ideally check that even when the
1316   // subchannels are in state CONNECTING for an extended period of time,
1317   // we will still report TRANSIENT_FAILURE.  Unfortunately, we don't
1318   // currently have a good way to get a subchannel to report CONNECTING
1319   // for a long period of time, since the servers in this test framework
1320   // are on the loopback interface, which will immediately return a
1321   // "Connection refused" error, so the subchannels will only be in
1322   // CONNECTING state very briefly.  When we have time, see if we can
1323   // find a way to fix this.
1324   auto response_generator = BuildResolverResponseGenerator();
1325   auto channel = BuildChannel("round_robin", response_generator);
1326   auto stub = BuildStub(channel);
1327   response_generator.SetNextResolution({
1328       grpc_pick_unused_port_or_die(),
1329       grpc_pick_unused_port_or_die(),
1330       grpc_pick_unused_port_or_die(),
1331   });
1332   for (size_t i = 0; i < servers_.size(); ++i) {
1333     servers_[i]->Shutdown();
1334   }
1335   auto predicate = [](grpc_connectivity_state state) {
1336     return state == GRPC_CHANNEL_TRANSIENT_FAILURE;
1337   };
1338   EXPECT_TRUE(WaitForChannelState(channel.get(), predicate, true));
1339 }
1340
1341 TEST_F(ClientLbEnd2endTest, RoundRobinSingleReconnect) {
1342   const int kNumServers = 3;
1343   StartServers(kNumServers);
1344   const auto ports = GetServersPorts();
1345   auto response_generator = BuildResolverResponseGenerator();
1346   auto channel = BuildChannel("round_robin", response_generator);
1347   auto stub = BuildStub(channel);
1348   response_generator.SetNextResolution(ports);
1349   for (size_t i = 0; i < kNumServers; ++i) {
1350     WaitForServer(stub, i, DEBUG_LOCATION);
1351   }
1352   for (size_t i = 0; i < servers_.size(); ++i) {
1353     CheckRpcSendOk(stub, DEBUG_LOCATION);
1354     EXPECT_EQ(1, servers_[i]->service_.request_count()) << "for backend #" << i;
1355   }
1356   // One request should have gone to each server.
1357   for (size_t i = 0; i < servers_.size(); ++i) {
1358     EXPECT_EQ(1, servers_[i]->service_.request_count());
1359   }
1360   const auto pre_death = servers_[0]->service_.request_count();
1361   // Kill the first server.
1362   servers_[0]->Shutdown();
1363   // Client request still succeed. May need retrying if RR had returned a pick
1364   // before noticing the change in the server's connectivity.
1365   while (!SendRpc(stub)) {
1366   }  // Retry until success.
1367   // Send a bunch of RPCs that should succeed.
1368   for (int i = 0; i < 10 * kNumServers; ++i) {
1369     CheckRpcSendOk(stub, DEBUG_LOCATION);
1370   }
1371   const auto post_death = servers_[0]->service_.request_count();
1372   // No requests have gone to the deceased server.
1373   EXPECT_EQ(pre_death, post_death);
1374   // Bring the first server back up.
1375   StartServer(0);
1376   // Requests should start arriving at the first server either right away (if
1377   // the server managed to start before the RR policy retried the subchannel) or
1378   // after the subchannel retry delay otherwise (RR's subchannel retried before
1379   // the server was fully back up).
1380   WaitForServer(stub, 0, DEBUG_LOCATION);
1381 }
1382
1383 // If health checking is required by client but health checking service
1384 // is not running on the server, the channel should be treated as healthy.
1385 TEST_F(ClientLbEnd2endTest,
1386        RoundRobinServersHealthCheckingUnimplementedTreatedAsHealthy) {
1387   StartServers(1);  // Single server
1388   ChannelArguments args;
1389   args.SetServiceConfigJSON(
1390       "{\"healthCheckConfig\": "
1391       "{\"serviceName\": \"health_check_service_name\"}}");
1392   auto response_generator = BuildResolverResponseGenerator();
1393   auto channel = BuildChannel("round_robin", response_generator, args);
1394   auto stub = BuildStub(channel);
1395   response_generator.SetNextResolution({servers_[0]->port_});
1396   EXPECT_TRUE(WaitForChannelReady(channel.get()));
1397   CheckRpcSendOk(stub, DEBUG_LOCATION);
1398 }
1399
1400 TEST_F(ClientLbEnd2endTest, RoundRobinWithHealthChecking) {
1401   EnableDefaultHealthCheckService(true);
1402   // Start servers.
1403   const int kNumServers = 3;
1404   StartServers(kNumServers);
1405   ChannelArguments args;
1406   args.SetServiceConfigJSON(
1407       "{\"healthCheckConfig\": "
1408       "{\"serviceName\": \"health_check_service_name\"}}");
1409   auto response_generator = BuildResolverResponseGenerator();
1410   auto channel = BuildChannel("round_robin", response_generator, args);
1411   auto stub = BuildStub(channel);
1412   response_generator.SetNextResolution(GetServersPorts());
1413   // Channel should not become READY, because health checks should be failing.
1414   gpr_log(GPR_INFO,
1415           "*** initial state: unknown health check service name for "
1416           "all servers");
1417   EXPECT_FALSE(WaitForChannelReady(channel.get(), 1));
1418   // Now set one of the servers to be healthy.
1419   // The channel should become healthy and all requests should go to
1420   // the healthy server.
1421   gpr_log(GPR_INFO, "*** server 0 healthy");
1422   servers_[0]->SetServingStatus("health_check_service_name", true);
1423   EXPECT_TRUE(WaitForChannelReady(channel.get()));
1424   for (int i = 0; i < 10; ++i) {
1425     CheckRpcSendOk(stub, DEBUG_LOCATION);
1426   }
1427   EXPECT_EQ(10, servers_[0]->service_.request_count());
1428   EXPECT_EQ(0, servers_[1]->service_.request_count());
1429   EXPECT_EQ(0, servers_[2]->service_.request_count());
1430   // Now set a second server to be healthy.
1431   gpr_log(GPR_INFO, "*** server 2 healthy");
1432   servers_[2]->SetServingStatus("health_check_service_name", true);
1433   WaitForServer(stub, 2, DEBUG_LOCATION);
1434   for (int i = 0; i < 10; ++i) {
1435     CheckRpcSendOk(stub, DEBUG_LOCATION);
1436   }
1437   EXPECT_EQ(5, servers_[0]->service_.request_count());
1438   EXPECT_EQ(0, servers_[1]->service_.request_count());
1439   EXPECT_EQ(5, servers_[2]->service_.request_count());
1440   // Now set the remaining server to be healthy.
1441   gpr_log(GPR_INFO, "*** server 1 healthy");
1442   servers_[1]->SetServingStatus("health_check_service_name", true);
1443   WaitForServer(stub, 1, DEBUG_LOCATION);
1444   for (int i = 0; i < 9; ++i) {
1445     CheckRpcSendOk(stub, DEBUG_LOCATION);
1446   }
1447   EXPECT_EQ(3, servers_[0]->service_.request_count());
1448   EXPECT_EQ(3, servers_[1]->service_.request_count());
1449   EXPECT_EQ(3, servers_[2]->service_.request_count());
1450   // Now set one server to be unhealthy again.  Then wait until the
1451   // unhealthiness has hit the client.  We know that the client will see
1452   // this when we send kNumServers requests and one of the remaining servers
1453   // sees two of the requests.
1454   gpr_log(GPR_INFO, "*** server 0 unhealthy");
1455   servers_[0]->SetServingStatus("health_check_service_name", false);
1456   do {
1457     ResetCounters();
1458     for (int i = 0; i < kNumServers; ++i) {
1459       CheckRpcSendOk(stub, DEBUG_LOCATION);
1460     }
1461   } while (servers_[1]->service_.request_count() != 2 &&
1462            servers_[2]->service_.request_count() != 2);
1463   // Now set the remaining two servers to be unhealthy.  Make sure the
1464   // channel leaves READY state and that RPCs fail.
1465   gpr_log(GPR_INFO, "*** all servers unhealthy");
1466   servers_[1]->SetServingStatus("health_check_service_name", false);
1467   servers_[2]->SetServingStatus("health_check_service_name", false);
1468   EXPECT_TRUE(WaitForChannelNotReady(channel.get()));
1469   CheckRpcSendFailure(stub);
1470   // Clean up.
1471   EnableDefaultHealthCheckService(false);
1472 }
1473
1474 TEST_F(ClientLbEnd2endTest,
1475        RoundRobinWithHealthCheckingHandlesSubchannelFailure) {
1476   EnableDefaultHealthCheckService(true);
1477   // Start servers.
1478   const int kNumServers = 3;
1479   StartServers(kNumServers);
1480   servers_[0]->SetServingStatus("health_check_service_name", true);
1481   servers_[1]->SetServingStatus("health_check_service_name", true);
1482   servers_[2]->SetServingStatus("health_check_service_name", true);
1483   ChannelArguments args;
1484   args.SetServiceConfigJSON(
1485       "{\"healthCheckConfig\": "
1486       "{\"serviceName\": \"health_check_service_name\"}}");
1487   auto response_generator = BuildResolverResponseGenerator();
1488   auto channel = BuildChannel("round_robin", response_generator, args);
1489   auto stub = BuildStub(channel);
1490   response_generator.SetNextResolution(GetServersPorts());
1491   WaitForServer(stub, 0, DEBUG_LOCATION);
1492   // Stop server 0 and send a new resolver result to ensure that RR
1493   // checks each subchannel's state.
1494   servers_[0]->Shutdown();
1495   response_generator.SetNextResolution(GetServersPorts());
1496   // Send a bunch more RPCs.
1497   for (size_t i = 0; i < 100; i++) {
1498     SendRpc(stub);
1499   }
1500 }
1501
1502 TEST_F(ClientLbEnd2endTest, RoundRobinWithHealthCheckingInhibitPerChannel) {
1503   EnableDefaultHealthCheckService(true);
1504   // Start server.
1505   const int kNumServers = 1;
1506   StartServers(kNumServers);
1507   // Create a channel with health-checking enabled.
1508   ChannelArguments args;
1509   args.SetServiceConfigJSON(
1510       "{\"healthCheckConfig\": "
1511       "{\"serviceName\": \"health_check_service_name\"}}");
1512   auto response_generator1 = BuildResolverResponseGenerator();
1513   auto channel1 = BuildChannel("round_robin", response_generator1, args);
1514   auto stub1 = BuildStub(channel1);
1515   std::vector<int> ports = GetServersPorts();
1516   response_generator1.SetNextResolution(ports);
1517   // Create a channel with health checking enabled but inhibited.
1518   args.SetInt(GRPC_ARG_INHIBIT_HEALTH_CHECKING, 1);
1519   auto response_generator2 = BuildResolverResponseGenerator();
1520   auto channel2 = BuildChannel("round_robin", response_generator2, args);
1521   auto stub2 = BuildStub(channel2);
1522   response_generator2.SetNextResolution(ports);
1523   // First channel should not become READY, because health checks should be
1524   // failing.
1525   EXPECT_FALSE(WaitForChannelReady(channel1.get(), 1));
1526   CheckRpcSendFailure(stub1);
1527   // Second channel should be READY.
1528   EXPECT_TRUE(WaitForChannelReady(channel2.get(), 1));
1529   CheckRpcSendOk(stub2, DEBUG_LOCATION);
1530   // Enable health checks on the backend and wait for channel 1 to succeed.
1531   servers_[0]->SetServingStatus("health_check_service_name", true);
1532   CheckRpcSendOk(stub1, DEBUG_LOCATION, true /* wait_for_ready */);
1533   // Check that we created only one subchannel to the backend.
1534   EXPECT_EQ(1UL, servers_[0]->service_.clients().size());
1535   // Clean up.
1536   EnableDefaultHealthCheckService(false);
1537 }
1538
1539 TEST_F(ClientLbEnd2endTest, RoundRobinWithHealthCheckingServiceNamePerChannel) {
1540   EnableDefaultHealthCheckService(true);
1541   // Start server.
1542   const int kNumServers = 1;
1543   StartServers(kNumServers);
1544   // Create a channel with health-checking enabled.
1545   ChannelArguments args;
1546   args.SetServiceConfigJSON(
1547       "{\"healthCheckConfig\": "
1548       "{\"serviceName\": \"health_check_service_name\"}}");
1549   auto response_generator1 = BuildResolverResponseGenerator();
1550   auto channel1 = BuildChannel("round_robin", response_generator1, args);
1551   auto stub1 = BuildStub(channel1);
1552   std::vector<int> ports = GetServersPorts();
1553   response_generator1.SetNextResolution(ports);
1554   // Create a channel with health-checking enabled with a different
1555   // service name.
1556   ChannelArguments args2;
1557   args2.SetServiceConfigJSON(
1558       "{\"healthCheckConfig\": "
1559       "{\"serviceName\": \"health_check_service_name2\"}}");
1560   auto response_generator2 = BuildResolverResponseGenerator();
1561   auto channel2 = BuildChannel("round_robin", response_generator2, args2);
1562   auto stub2 = BuildStub(channel2);
1563   response_generator2.SetNextResolution(ports);
1564   // Allow health checks from channel 2 to succeed.
1565   servers_[0]->SetServingStatus("health_check_service_name2", true);
1566   // First channel should not become READY, because health checks should be
1567   // failing.
1568   EXPECT_FALSE(WaitForChannelReady(channel1.get(), 1));
1569   CheckRpcSendFailure(stub1);
1570   // Second channel should be READY.
1571   EXPECT_TRUE(WaitForChannelReady(channel2.get(), 1));
1572   CheckRpcSendOk(stub2, DEBUG_LOCATION);
1573   // Enable health checks for channel 1 and wait for it to succeed.
1574   servers_[0]->SetServingStatus("health_check_service_name", true);
1575   CheckRpcSendOk(stub1, DEBUG_LOCATION, true /* wait_for_ready */);
1576   // Check that we created only one subchannel to the backend.
1577   EXPECT_EQ(1UL, servers_[0]->service_.clients().size());
1578   // Clean up.
1579   EnableDefaultHealthCheckService(false);
1580 }
1581
1582 TEST_F(ClientLbEnd2endTest,
1583        RoundRobinWithHealthCheckingServiceNameChangesAfterSubchannelsCreated) {
1584   EnableDefaultHealthCheckService(true);
1585   // Start server.
1586   const int kNumServers = 1;
1587   StartServers(kNumServers);
1588   // Create a channel with health-checking enabled.
1589   const char* kServiceConfigJson =
1590       "{\"healthCheckConfig\": "
1591       "{\"serviceName\": \"health_check_service_name\"}}";
1592   auto response_generator = BuildResolverResponseGenerator();
1593   auto channel = BuildChannel("round_robin", response_generator);
1594   auto stub = BuildStub(channel);
1595   std::vector<int> ports = GetServersPorts();
1596   response_generator.SetNextResolution(ports, kServiceConfigJson);
1597   servers_[0]->SetServingStatus("health_check_service_name", true);
1598   EXPECT_TRUE(WaitForChannelReady(channel.get(), 1 /* timeout_seconds */));
1599   // Send an update on the channel to change it to use a health checking
1600   // service name that is not being reported as healthy.
1601   const char* kServiceConfigJson2 =
1602       "{\"healthCheckConfig\": "
1603       "{\"serviceName\": \"health_check_service_name2\"}}";
1604   response_generator.SetNextResolution(ports, kServiceConfigJson2);
1605   EXPECT_TRUE(WaitForChannelNotReady(channel.get()));
1606   // Clean up.
1607   EnableDefaultHealthCheckService(false);
1608 }
1609
1610 TEST_F(ClientLbEnd2endTest, ChannelIdleness) {
1611   // Start server.
1612   const int kNumServers = 1;
1613   StartServers(kNumServers);
1614   // Set max idle time and build the channel.
1615   ChannelArguments args;
1616   args.SetInt(GRPC_ARG_CLIENT_IDLE_TIMEOUT_MS, 1000);
1617   auto response_generator = BuildResolverResponseGenerator();
1618   auto channel = BuildChannel("", response_generator, args);
1619   auto stub = BuildStub(channel);
1620   // The initial channel state should be IDLE.
1621   EXPECT_EQ(channel->GetState(false), GRPC_CHANNEL_IDLE);
1622   // After sending RPC, channel state should be READY.
1623   response_generator.SetNextResolution(GetServersPorts());
1624   CheckRpcSendOk(stub, DEBUG_LOCATION);
1625   EXPECT_EQ(channel->GetState(false), GRPC_CHANNEL_READY);
1626   // After a period time not using the channel, the channel state should switch
1627   // to IDLE.
1628   gpr_sleep_until(grpc_timeout_milliseconds_to_deadline(1200));
1629   EXPECT_EQ(channel->GetState(false), GRPC_CHANNEL_IDLE);
1630   // Sending a new RPC should awake the IDLE channel.
1631   response_generator.SetNextResolution(GetServersPorts());
1632   CheckRpcSendOk(stub, DEBUG_LOCATION);
1633   EXPECT_EQ(channel->GetState(false), GRPC_CHANNEL_READY);
1634 }
1635
1636 class ClientLbInterceptTrailingMetadataTest : public ClientLbEnd2endTest {
1637  protected:
1638   void SetUp() override {
1639     ClientLbEnd2endTest::SetUp();
1640     grpc_core::RegisterInterceptRecvTrailingMetadataLoadBalancingPolicy(
1641         ReportTrailerIntercepted, this);
1642   }
1643
1644   void TearDown() override { ClientLbEnd2endTest::TearDown(); }
1645
1646   int trailers_intercepted() {
1647     grpc::internal::MutexLock lock(&mu_);
1648     return trailers_intercepted_;
1649   }
1650
1651   const udpa::data::orca::v1::OrcaLoadReport* backend_load_report() {
1652     grpc::internal::MutexLock lock(&mu_);
1653     return load_report_.get();
1654   }
1655
1656  private:
1657   static void ReportTrailerIntercepted(
1658       void* arg, const grpc_core::LoadBalancingPolicy::BackendMetricData*
1659                      backend_metric_data) {
1660     ClientLbInterceptTrailingMetadataTest* self =
1661         static_cast<ClientLbInterceptTrailingMetadataTest*>(arg);
1662     grpc::internal::MutexLock lock(&self->mu_);
1663     self->trailers_intercepted_++;
1664     if (backend_metric_data != nullptr) {
1665       self->load_report_.reset(new udpa::data::orca::v1::OrcaLoadReport);
1666       self->load_report_->set_cpu_utilization(
1667           backend_metric_data->cpu_utilization);
1668       self->load_report_->set_mem_utilization(
1669           backend_metric_data->mem_utilization);
1670       self->load_report_->set_rps(backend_metric_data->requests_per_second);
1671       for (const auto& p : backend_metric_data->request_cost) {
1672         grpc_core::UniquePtr<char> name = p.first.dup();
1673         (*self->load_report_->mutable_request_cost())[name.get()] = p.second;
1674       }
1675       for (const auto& p : backend_metric_data->utilization) {
1676         grpc_core::UniquePtr<char> name = p.first.dup();
1677         (*self->load_report_->mutable_utilization())[name.get()] = p.second;
1678       }
1679     }
1680   }
1681
1682   grpc::internal::Mutex mu_;
1683   int trailers_intercepted_ = 0;
1684   std::unique_ptr<udpa::data::orca::v1::OrcaLoadReport> load_report_;
1685 };
1686
1687 TEST_F(ClientLbInterceptTrailingMetadataTest, InterceptsRetriesDisabled) {
1688   const int kNumServers = 1;
1689   const int kNumRpcs = 10;
1690   StartServers(kNumServers);
1691   auto response_generator = BuildResolverResponseGenerator();
1692   auto channel =
1693       BuildChannel("intercept_trailing_metadata_lb", response_generator);
1694   auto stub = BuildStub(channel);
1695   response_generator.SetNextResolution(GetServersPorts());
1696   for (size_t i = 0; i < kNumRpcs; ++i) {
1697     CheckRpcSendOk(stub, DEBUG_LOCATION);
1698   }
1699   // Check LB policy name for the channel.
1700   EXPECT_EQ("intercept_trailing_metadata_lb",
1701             channel->GetLoadBalancingPolicyName());
1702   EXPECT_EQ(kNumRpcs, trailers_intercepted());
1703   EXPECT_EQ(nullptr, backend_load_report());
1704 }
1705
1706 TEST_F(ClientLbInterceptTrailingMetadataTest, InterceptsRetriesEnabled) {
1707   const int kNumServers = 1;
1708   const int kNumRpcs = 10;
1709   StartServers(kNumServers);
1710   ChannelArguments args;
1711   args.SetServiceConfigJSON(
1712       "{\n"
1713       "  \"methodConfig\": [ {\n"
1714       "    \"name\": [\n"
1715       "      { \"service\": \"grpc.testing.EchoTestService\" }\n"
1716       "    ],\n"
1717       "    \"retryPolicy\": {\n"
1718       "      \"maxAttempts\": 3,\n"
1719       "      \"initialBackoff\": \"1s\",\n"
1720       "      \"maxBackoff\": \"120s\",\n"
1721       "      \"backoffMultiplier\": 1.6,\n"
1722       "      \"retryableStatusCodes\": [ \"ABORTED\" ]\n"
1723       "    }\n"
1724       "  } ]\n"
1725       "}");
1726   auto response_generator = BuildResolverResponseGenerator();
1727   auto channel =
1728       BuildChannel("intercept_trailing_metadata_lb", response_generator, args);
1729   auto stub = BuildStub(channel);
1730   response_generator.SetNextResolution(GetServersPorts());
1731   for (size_t i = 0; i < kNumRpcs; ++i) {
1732     CheckRpcSendOk(stub, DEBUG_LOCATION);
1733   }
1734   // Check LB policy name for the channel.
1735   EXPECT_EQ("intercept_trailing_metadata_lb",
1736             channel->GetLoadBalancingPolicyName());
1737   EXPECT_EQ(kNumRpcs, trailers_intercepted());
1738   EXPECT_EQ(nullptr, backend_load_report());
1739 }
1740
1741 TEST_F(ClientLbInterceptTrailingMetadataTest, BackendMetricData) {
1742   const int kNumServers = 1;
1743   const int kNumRpcs = 10;
1744   StartServers(kNumServers);
1745   udpa::data::orca::v1::OrcaLoadReport load_report;
1746   load_report.set_cpu_utilization(0.5);
1747   load_report.set_mem_utilization(0.75);
1748   load_report.set_rps(25);
1749   auto* request_cost = load_report.mutable_request_cost();
1750   (*request_cost)["foo"] = 0.8;
1751   (*request_cost)["bar"] = 1.4;
1752   auto* utilization = load_report.mutable_utilization();
1753   (*utilization)["baz"] = 1.1;
1754   (*utilization)["quux"] = 0.9;
1755   for (const auto& server : servers_) {
1756     server->service_.set_load_report(&load_report);
1757   }
1758   auto response_generator = BuildResolverResponseGenerator();
1759   auto channel =
1760       BuildChannel("intercept_trailing_metadata_lb", response_generator);
1761   auto stub = BuildStub(channel);
1762   response_generator.SetNextResolution(GetServersPorts());
1763   for (size_t i = 0; i < kNumRpcs; ++i) {
1764     CheckRpcSendOk(stub, DEBUG_LOCATION);
1765     auto* actual = backend_load_report();
1766     ASSERT_NE(actual, nullptr);
1767     // TODO(roth): Change this to use EqualsProto() once that becomes
1768     // available in OSS.
1769     EXPECT_EQ(actual->cpu_utilization(), load_report.cpu_utilization());
1770     EXPECT_EQ(actual->mem_utilization(), load_report.mem_utilization());
1771     EXPECT_EQ(actual->rps(), load_report.rps());
1772     EXPECT_EQ(actual->request_cost().size(), load_report.request_cost().size());
1773     for (const auto& p : actual->request_cost()) {
1774       auto it = load_report.request_cost().find(p.first);
1775       ASSERT_NE(it, load_report.request_cost().end());
1776       EXPECT_EQ(it->second, p.second);
1777     }
1778     EXPECT_EQ(actual->utilization().size(), load_report.utilization().size());
1779     for (const auto& p : actual->utilization()) {
1780       auto it = load_report.utilization().find(p.first);
1781       ASSERT_NE(it, load_report.utilization().end());
1782       EXPECT_EQ(it->second, p.second);
1783     }
1784   }
1785   // Check LB policy name for the channel.
1786   EXPECT_EQ("intercept_trailing_metadata_lb",
1787             channel->GetLoadBalancingPolicyName());
1788   EXPECT_EQ(kNumRpcs, trailers_intercepted());
1789 }
1790
1791 }  // namespace
1792 }  // namespace testing
1793 }  // namespace grpc
1794
1795 int main(int argc, char** argv) {
1796   ::testing::InitGoogleTest(&argc, argv);
1797   grpc::testing::TestEnvironment env(argc, argv);
1798   const auto result = RUN_ALL_TESTS();
1799   return result;
1800 }