Google Git
Sign in
apache / mesos / refs/tags/1.10.0 / . / src / tests / slave_tests.cpp
blob: b46e56142ff5ae4ea46f0cb16d6c0ed193d49623 [file] [log] [blame]
// Licensed to the Apache Software Foundation (ASF) under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you 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.
#ifndef __WINDOWS__
#include <unistd.h>
#endif // __WINDOWS__
#include <algorithm>
#include <map>
#include <memory>
#include <string>
#include <utility>
#include <vector>
#include <gmock/gmock.h>
#include <mesos/executor.hpp>
#include <mesos/scheduler.hpp>
#include <mesos/type_utils.hpp>
#include <mesos/authentication/http/basic_authenticator_factory.hpp>
#include <mesos/v1/mesos.hpp>
#include <mesos/v1/resource_provider/resource_provider.hpp>
#include <process/clock.hpp>
#include <process/collect.hpp>
#include <process/future.hpp>
#include <process/gmock.hpp>
#include <process/gtest.hpp>
#include <process/http.hpp>
#include <process/owned.hpp>
#include <process/pid.hpp>
#include <process/reap.hpp>
#include <process/subprocess.hpp>
#include <stout/hashset.hpp>
#include <stout/json.hpp>
#include <stout/none.hpp>
#include <stout/nothing.hpp>
#include <stout/option.hpp>
#include <stout/os.hpp>
#include <stout/path.hpp>
#include <stout/strings.hpp>
#include <stout/try.hpp>
#ifdef USE_SSL_SOCKET
#include "authentication/executor/jwt_secret_generator.hpp"
#endif // USE_SSL_SOCKET
#include "common/build.hpp"
#include "common/future_tracker.hpp"
#include "common/http.hpp"
#include "common/protobuf_utils.hpp"
#include "master/flags.hpp"
#include "master/master.hpp"
#include "master/registry_operations.hpp"
#include "master/detector/standalone.hpp"
#include "slave/constants.hpp"
#include "slave/gc.hpp"
#include "slave/gc_process.hpp"
#include "slave/flags.hpp"
#include "slave/slave.hpp"
#include "slave/paths.hpp"
#include "slave/containerizer/fetcher.hpp"
#include "slave/containerizer/fetcher_process.hpp"
#include "slave/containerizer/mesos/containerizer.hpp"
#include "slave/containerizer/mesos/isolator_tracker.hpp"
#include "slave/containerizer/mesos/launcher.hpp"
#include "tests/active_user_test_helper.hpp"
#include "tests/containerizer.hpp"
#include "tests/environment.hpp"
#include "tests/flags.hpp"
#include "tests/limiter.hpp"
#include "tests/mesos.hpp"
#include "tests/mock_slave.hpp"
#include "tests/resources_utils.hpp"
#include "tests/utils.hpp"
#include "tests/containerizer/isolator.hpp"
#include "tests/containerizer/mock_containerizer.hpp"
using namespace mesos::internal::slave;
#ifdef USE_SSL_SOCKET
using mesos::authentication::executor::JWTSecretGenerator;
#endif // USE_SSL_SOCKET
using mesos::internal::master::Master;
using mesos::internal::protobuf::createLabel;
using mesos::master::detector::MasterDetector;
using mesos::master::detector::StandaloneMasterDetector;
using mesos::v1::resource_provider::Event;
using mesos::slave::ContainerConfig;
using mesos::slave::ContainerLaunchInfo;
using mesos::slave::ContainerTermination;
using mesos::slave::Isolator;
using mesos::v1::scheduler::Call;
using mesos::v1::scheduler::Mesos;
using process::Clock;
using process::Failure;
using process::Future;
using process::Message;
using process::Owned;
using process::PID;
using process::Promise;
using process::UPID;
using process::filter;
using process::http::InternalServerError;
using process::http::OK;
using process::http::Response;
using process::http::ServiceUnavailable;
using process::http::Unauthorized;
using process::http::authentication::Principal;
using std::list;
using std::map;
using std::shared_ptr;
using std::string;
using std::vector;
using testing::_;
using testing::AtLeast;
using testing::AtMost;
using testing::DoAll;
using testing::Eq;
using testing::Exactly;
using testing::StrEq;
using testing::Invoke;
using testing::InvokeWithoutArgs;
using testing::Return;
using testing::SaveArg;
using testing::WithParamInterface;
namespace mesos {
namespace internal {
namespace tests {
// Those of the overall Mesos master/slave/scheduler/driver tests
// that seem vaguely more slave than master-related are in this file.
// The others are in "master_tests.cpp".
class SlaveTest : public ContainerizerTest<slave::MesosContainerizer>
{
public:
const std::string defaultIsolators{
#ifdef __WINDOWS__
"windows/cpu"
#else
"posix/cpu,posix/mem"
#endif // __WINDOWS__
};
CommandInfo echoAuthorCommand()
{
CommandInfo command;
command.set_shell(false);
#ifdef __WINDOWS__
command.set_value("powershell.exe");
command.add_arguments("powershell.exe");
command.add_arguments("-NoProfile");
command.add_arguments("-Command");
command.add_arguments("echo --author");
#else
command.set_value("/bin/echo");
command.add_arguments("/bin/echo");
command.add_arguments("--author");
#endif // __WINDOWS__
return command;
};
};
// This test ensures that when a slave shuts itself down, it
// unregisters itself and the master notifies the framework
// immediately and rescinds any offers.
TEST_F(SlaveTest, Shutdown)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
EXPECT_EQ(1u, offers->size());
Future<Nothing> offerRescinded;
EXPECT_CALL(sched, offerRescinded(&driver, offers.get()[0].id()))
.WillOnce(FutureSatisfy(&offerRescinded));
Future<Nothing> slaveLost;
EXPECT_CALL(sched, slaveLost(&driver, offers.get()[0].slave_id()))
.WillOnce(FutureSatisfy(&slaveLost));
// Stop the slave with explicit shutdown message so that the slave
// unregisters.
slave.get()->shutdown();
slave->reset();
AWAIT_READY(offerRescinded);
AWAIT_READY(slaveLost);
JSON::Object stats = Metrics();
EXPECT_EQ(1, stats.values["master/slave_removals"]);
EXPECT_EQ(1, stats.values["master/slave_removals/reason_unregistered"]);
EXPECT_EQ(0, stats.values["master/slave_removals/reason_unhealthy"]);
driver.stop();
driver.join();
}
// This test verifies that the slave rejects duplicate terminal
// status updates for tasks before the first terminal update is
// acknowledged.
TEST_F(SlaveTest, DuplicateTerminalUpdateBeforeAck)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
slave::Flags agentFlags = CreateSlaveFlags();
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &containerizer, agentFlags);
ASSERT_SOME(slave);
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_checkpoint(true); // Enable checkpointing.
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, frameworkInfo, master.get()->pid, DEFAULT_CREDENTIAL);
FrameworkID frameworkId;
EXPECT_CALL(sched, registered(_, _, _))
.WillOnce(SaveArg<1>(&frameworkId));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
// Advance the clock to trigger both agent registration and a batch
// allocation.
Clock::advance(agentFlags.registration_backoff_factor);
Clock::advance(masterFlags.allocation_interval);
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
ExecutorDriver* execDriver;
EXPECT_CALL(exec, registered(_, _, _, _))
.WillOnce(SaveArg<0>(&execDriver));
// Send a terminal update right away.
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_FINISHED));
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(_, _))
.WillOnce(FutureArg<1>(&status));
// Drop the first ACK from the scheduler to the slave.
Future<StatusUpdateAcknowledgementMessage> statusUpdateAckMessage =
DROP_PROTOBUF(StatusUpdateAcknowledgementMessage(), _, slave.get()->pid);
Future<Nothing> ___statusUpdate =
FUTURE_DISPATCH(slave.get()->pid, &Slave::___statusUpdate);
TaskInfo task;
task.set_name("test-task");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers->at(0).slave_id());
task.mutable_resources()->MergeFrom(offers->at(0).resources());
task.mutable_executor()->MergeFrom(DEFAULT_EXECUTOR_INFO);
driver.launchTasks(offers->at(0).id(), {task});
AWAIT_READY(status);
EXPECT_EQ(TASK_FINISHED, status->state());
AWAIT_READY(statusUpdateAckMessage);
// At this point the task status update manager has enqueued
// TASK_FINISHED update.
AWAIT_READY(___statusUpdate);
Future<Nothing> _statusUpdate2 =
FUTURE_DISPATCH(slave.get()->pid, &Slave::_statusUpdate);
// Now send a TASK_KILLED update for the same task.
TaskStatus status2 = status.get();
status2.set_state(TASK_KILLED);
execDriver->sendStatusUpdate(status2);
// At this point the slave has handled the TASK_KILLED update.
AWAIT_READY(_statusUpdate2);
// After we advance the clock, the scheduler should receive
// the retried TASK_FINISHED update and acknowledge it.
Future<TaskStatus> update;
EXPECT_CALL(sched, statusUpdate(_, _))
.WillOnce(FutureArg<1>(&update));
Clock::advance(slave::STATUS_UPDATE_RETRY_INTERVAL_MIN);
Clock::settle();
// Ensure the scheduler receives TASK_FINISHED.
AWAIT_READY(update);
EXPECT_EQ(TASK_FINISHED, update->state());
// Settle the clock to ensure that TASK_KILLED is not sent.
Clock::settle();
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
TEST_F(SlaveTest, ShutdownUnregisteredExecutor)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
// Need flags for 'executor_registration_timeout'.
slave::Flags flags = CreateSlaveFlags();
// Set the isolation flag so we know a MesosContainerizer will
// be created.
flags.isolation = defaultIsolators;
Fetcher fetcher(flags);
Try<MesosContainerizer*> _containerizer =
MesosContainerizer::create(flags, false, &fetcher);
ASSERT_SOME(_containerizer);
Owned<MesosContainerizer> containerizer(_containerizer.get());
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), containerizer.get());
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
// Launch a task with the command executor.
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers.get()[0].slave_id());
task.mutable_resources()->MergeFrom(offers.get()[0].resources());
CommandInfo command;
command.set_value(SLEEP_COMMAND(10));
task.mutable_command()->MergeFrom(command);
// Drop the registration message from the executor to the slave.
Future<Message> registerExecutor =
DROP_MESSAGE(Eq(RegisterExecutorMessage().GetTypeName()), _, _);
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(registerExecutor);
Clock::pause();
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
// Ensure that the slave times out and kills the executor.
Future<Nothing> destroyExecutor =
FUTURE_DISPATCH(_, &MesosContainerizerProcess::destroy);
Clock::advance(flags.executor_registration_timeout);
AWAIT_READY(destroyExecutor);
Clock::settle(); // Wait for Containerizer::destroy to complete.
Clock::resume();
AWAIT_READY(status);
ASSERT_EQ(TASK_FAILED, status->state());
EXPECT_EQ(TaskStatus::SOURCE_SLAVE, status->source());
EXPECT_EQ(TaskStatus::REASON_EXECUTOR_REGISTRATION_TIMEOUT,
status->reason());
driver.stop();
driver.join();
}
#ifndef __WINDOWS__
// This test verifies that mesos agent gets notified of task
// launch failure triggered by the executor register timeout
// caused by slow URI fetching.
TEST_F(SlaveTest, ExecutorTimeoutCausedBySlowFetch)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
string hadoopPath = os::getcwd();
string hadoopBinPath = path::join(hadoopPath, "bin");
ASSERT_SOME(os::mkdir(hadoopBinPath));
ASSERT_SOME(os::chmod(hadoopBinPath, S_IRWXU | S_IRWXG | S_IRWXO));
// A spurious "hadoop" script that sleeps forever.
string mockHadoopScript = "#!/usr/bin/env bash\n"
"sleep 1000";
string hadoopCommand = path::join(hadoopBinPath, "hadoop");
ASSERT_SOME(os::write(hadoopCommand, mockHadoopScript));
ASSERT_SOME(os::chmod(hadoopCommand,
S_IRWXU | S_IRGRP | S_IXGRP | S_IROTH | S_IXOTH));
slave::Flags flags = CreateSlaveFlags();
flags.hadoop_home = hadoopPath;
Fetcher fetcher(flags);
Try<MesosContainerizer*> _containerizer = MesosContainerizer::create(
flags, true, &fetcher);
ASSERT_SOME(_containerizer);
Owned<MesosContainerizer> containerizer(_containerizer.get());
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(
detector.get(),
containerizer.get(),
flags);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched,
DEFAULT_FRAMEWORK_INFO,
master.get()->pid,
DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
// Launch a task with the command executor.
// The task uses a URI that needs to be fetched by the HDFS client
// and will be blocked until the executor registrartion times out.
CommandInfo commandInfo;
CommandInfo::URI* uri = commandInfo.add_uris();
uri->set_value(path::join("hdfs://dummyhost/dummypath", "test"));
// Using a dummy command value as it's a required field. The
// command won't be invoked.
commandInfo.set_value(SLEEP_COMMAND(10));
ExecutorID executorId;
executorId.set_value("test-executor-staging");
TaskInfo task = createTask(
offers.get()[0].slave_id(),
offers.get()[0].resources(),
commandInfo,
executorId,
"test-task-staging");
Future<Nothing> fetch = FUTURE_DISPATCH(
_, &FetcherProcess::fetch);
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
Clock::pause();
driver.launchTasks(offers.get()[0].id(), {task});
Future<Nothing> executorLost;
EXPECT_CALL(sched, executorLost(&driver, executorId, _, _))
.WillOnce(FutureSatisfy(&executorLost));
// Ensure that the slave times out and kills the executor.
Future<Nothing> destroyExecutor = FUTURE_DISPATCH(
_, &MesosContainerizerProcess::destroy);
AWAIT_READY(fetch);
Clock::advance(flags.executor_registration_timeout);
AWAIT_READY(destroyExecutor);
Clock::settle(); // Wait for Containerizer::destroy to complete.
Clock::resume();
AWAIT_READY(executorLost);
AWAIT_READY(status);
ASSERT_EQ(TASK_FAILED, status->state());
EXPECT_EQ(TaskStatus::SOURCE_SLAVE, status->source());
EXPECT_EQ(TaskStatus::REASON_CONTAINER_LAUNCH_FAILED, status->reason());
driver.stop();
driver.join();
}
#endif // __WINDOWS__
// This test verifies that when an executor terminates before
// registering with slave, it is properly cleaned up.
TEST_F(SlaveTest, RemoveUnregisteredTerminatedExecutor)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers.get()[0].slave_id());
task.mutable_resources()->MergeFrom(offers.get()[0].resources());
task.mutable_executor()->MergeFrom(DEFAULT_EXECUTOR_INFO);
// Drop the registration message from the executor to the slave.
Future<Message> registerExecutorMessage =
DROP_MESSAGE(Eq(RegisterExecutorMessage().GetTypeName()), _, _);
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(registerExecutorMessage);
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
Future<Nothing> schedule =
FUTURE_DISPATCH(_, &GarbageCollectorProcess::schedule);
EXPECT_CALL(sched, executorLost(&driver, DEFAULT_EXECUTOR_ID, _, _));
// Now kill the executor.
containerizer.destroy(offers.get()[0].framework_id(), DEFAULT_EXECUTOR_ID);
AWAIT_READY(status);
EXPECT_EQ(TASK_FAILED, status->state());
EXPECT_EQ(TaskStatus::SOURCE_SLAVE, status->source());
EXPECT_EQ(TaskStatus::REASON_EXECUTOR_TERMINATED, status->reason());
// We use 'gc.schedule' as a signal for the executor being cleaned
// up by the slave.
AWAIT_READY(schedule);
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// Test that we don't let task arguments bleed over as
// mesos-executor args. For more details of this see MESOS-1873.
//
// This assumes the ability to execute '/bin/echo --author'.
TEST_F(SlaveTest, CommandTaskWithArguments)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
// Need flags for 'executor_registration_timeout'.
slave::Flags flags = CreateSlaveFlags();
flags.isolation = defaultIsolators;
Fetcher fetcher(flags);
Try<MesosContainerizer*> _containerizer =
MesosContainerizer::create(flags, false, &fetcher);
ASSERT_SOME(_containerizer);
Owned<MesosContainerizer> containerizer(_containerizer.get());
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), containerizer.get());
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
// Launch a task with the command executor.
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers.get()[0].slave_id());
task.mutable_resources()->MergeFrom(offers.get()[0].resources());
// Command executor will run as user running test.
task.mutable_command()->MergeFrom(echoAuthorCommand());
Future<TaskStatus> statusStarting;
Future<TaskStatus> statusRunning;
Future<TaskStatus> statusFinished;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusStarting))
.WillOnce(FutureArg<1>(&statusRunning))
.WillOnce(FutureArg<1>(&statusFinished));
driver.launchTasks(offers.get()[0].id(), {task});
// Scheduler should first receive TASK_STARTING, followed by
// TASK_RUNNING and TASK_FINISHED from the executor.
AWAIT_READY(statusStarting);
EXPECT_EQ(TASK_STARTING, statusStarting->state());
EXPECT_EQ(TaskStatus::SOURCE_EXECUTOR, statusStarting->source());
AWAIT_READY(statusRunning);
EXPECT_EQ(TASK_RUNNING, statusRunning->state());
EXPECT_EQ(TaskStatus::SOURCE_EXECUTOR, statusRunning->source());
AWAIT_READY(statusFinished);
EXPECT_EQ(TASK_FINISHED, statusFinished->state());
EXPECT_EQ(TaskStatus::SOURCE_EXECUTOR, statusFinished->source());
driver.stop();
driver.join();
}
// Tests that task's kill policy grace period does not extend the time
// a task responsive to SIGTERM needs to exit and the terminal status
// to be delivered to the master.
TEST_F(SlaveTest, CommandTaskWithKillPolicy)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers));
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
Offer offer = offers.get()[0];
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offer.slave_id());
task.mutable_resources()->MergeFrom(offer.resources());
CommandInfo command;
command.set_value(SLEEP_COMMAND(1000));
task.mutable_command()->MergeFrom(command);
// Set task's kill policy grace period to a large value.
Duration gracePeriod = Seconds(100);
task.mutable_kill_policy()->mutable_grace_period()->set_nanoseconds(
gracePeriod.ns());
Future<TaskStatus> statusStarting;
Future<TaskStatus> statusRunning;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusStarting))
.WillOnce(FutureArg<1>(&statusRunning));
driver.launchTasks(offer.id(), {task});
AWAIT_READY(statusStarting);
EXPECT_EQ(TASK_STARTING, statusStarting->state());
AWAIT_READY(statusRunning);
EXPECT_EQ(TASK_RUNNING, statusRunning->state());
// Kill the task.
Future<TaskStatus> statusKilled;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusKilled));
driver.killTask(statusRunning->task_id());
// Since "sleep 1000" task is responsive to SIGTERM, we should
// observe TASK_KILLED update sooner than after `gracePeriod`
// elapses. This indicates that extended grace period does not
// influence the time a task and its command executor need to
// exit. We add a small buffer for a task to clean up and the
// update to be processed by the master.
AWAIT_READY_FOR(statusKilled, Seconds(1) + process::MAX_REAP_INTERVAL());
EXPECT_EQ(TASK_KILLED, statusKilled->state());
EXPECT_EQ(TaskStatus::SOURCE_EXECUTOR,
statusKilled->source());
driver.stop();
driver.join();
}
// Don't let args from the CommandInfo struct bleed over into
// mesos-executor forking. For more details of this see MESOS-1873.
TEST_F(SlaveTest, GetExecutorInfo)
{
TestContainerizer containerizer;
StandaloneMasterDetector detector;
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
None(),
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
FrameworkID frameworkId;
frameworkId.set_value("20141010-221431-251662764-60288-32120-0000");
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
frameworkInfo.mutable_id()->CopyFrom(frameworkId);
// Launch a task with the command executor.
Resources taskResources = Resources::parse("cpus:0.1;mem:32").get();
taskResources.allocate(frameworkInfo.roles(0));
TaskInfo task;
task.set_name("task");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->set_value(
"20141010-221431-251662764-60288-32120-0001");
task.mutable_resources()->MergeFrom(taskResources);
task.mutable_command()->MergeFrom(echoAuthorCommand());
DiscoveryInfo* info = task.mutable_discovery();
info->set_visibility(DiscoveryInfo::EXTERNAL);
info->set_name("mytask");
info->set_environment("mytest");
info->set_location("mylocation");
info->set_version("v0.1.1");
Labels* labels = task.mutable_labels();
labels->add_labels()->CopyFrom(createLabel("label1", "key1"));
labels->add_labels()->CopyFrom(createLabel("label2", "key2"));
const ExecutorInfo& executor =
slave.get()->mock()->getExecutorInfo(frameworkInfo, task);
// Now assert that it actually is running mesos-executor without any
// bleedover from the command we intend on running.
EXPECT_FALSE(executor.command().shell());
EXPECT_EQ(2, executor.command().arguments_size());
ASSERT_TRUE(executor.has_labels());
EXPECT_EQ(2, executor.labels().labels_size());
ASSERT_TRUE(executor.has_discovery());
ASSERT_TRUE(executor.discovery().has_name());
EXPECT_EQ("mytask", executor.discovery().name());
EXPECT_NE(string::npos, executor.command().value().find("mesos-executor"));
}
// Ensure getExecutorInfo for mesos-executor gets the ContainerInfo,
// if present. This ensures the MesosContainerizer can get the
// NetworkInfo even when using the command executor.
TEST_F(SlaveTest, GetExecutorInfoForTaskWithContainer)
{
TestContainerizer containerizer;
StandaloneMasterDetector detector;
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
None(),
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
frameworkInfo.mutable_id()->set_value(
"20141010-221431-251662764-60288-12345-0000");
// Launch a task with the command executor and ContainerInfo with
// NetworkInfo.
Resources taskResources = Resources::parse("cpus:0.1;mem:32").get();
taskResources.allocate(frameworkInfo.roles(0));
TaskInfo task;
task.set_name("task");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->set_value(
"20141010-221431-251662764-60288-12345-0001");
task.mutable_resources()->MergeFrom(taskResources);
task.mutable_command()->MergeFrom(echoAuthorCommand());
ContainerInfo* container = task.mutable_container();
container->set_type(ContainerInfo::MESOS);
NetworkInfo* network = container->add_network_infos();
network->add_ip_addresses()->set_ip_address("4.3.2.1");
network->add_groups("public");
const ExecutorInfo& executor =
slave.get()->mock()->getExecutorInfo(frameworkInfo, task);
// Now assert that the executor has both the command and ContainerInfo
EXPECT_FALSE(executor.command().shell());
// CommandInfo.container is not included. In this test the ContainerInfo
// must be included in Executor.container (copied from TaskInfo.container).
EXPECT_TRUE(executor.has_container());
EXPECT_EQ(1, executor.container().network_infos(0).ip_addresses_size());
NetworkInfo::IPAddress ipAddress =
executor.container().network_infos(0).ip_addresses(0);
EXPECT_EQ("4.3.2.1", ipAddress.ip_address());
EXPECT_EQ(1, executor.container().network_infos(0).groups_size());
EXPECT_EQ("public", executor.container().network_infos(0).groups(0));
}
// This test runs a command without the command user field set. The
// command will verify the assumption that the command is run as the
// slave user (in this case, root).
//
// TODO(andschwa): Enable when user impersonation works on Windows.
TEST_F_TEMP_DISABLED_ON_WINDOWS(
SlaveTest, ROOT_RunTaskWithCommandInfoWithoutUser)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
// Need flags for 'executor_registration_timeout'.
slave::Flags flags = CreateSlaveFlags();
flags.isolation = "posix/cpu,posix/mem";
Fetcher fetcher(flags);
Try<MesosContainerizer*> _containerizer =
MesosContainerizer::create(flags, false, &fetcher);
ASSERT_SOME(_containerizer);
Owned<MesosContainerizer> containerizer(_containerizer.get());
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), containerizer.get());
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
// Launch a task with the command executor.
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers.get()[0].slave_id());
task.mutable_resources()->MergeFrom(offers.get()[0].resources());
Result<string> user = os::user();
ASSERT_SOME(user) << "Failed to get current user name"
<< (user.isError() ? ": " + user.error() : "");
const string helper = getTestHelperPath("test-helper");
// Command executor will run as user running test.
CommandInfo command;
command.set_shell(false);
command.set_value(helper);
command.add_arguments(helper);
command.add_arguments(ActiveUserTestHelper::NAME);
command.add_arguments("--user=" + user.get());
task.mutable_command()->MergeFrom(command);
Future<TaskStatus> statusStarting;
Future<TaskStatus> statusRunning;
Future<TaskStatus> statusFinished;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusStarting))
.WillOnce(FutureArg<1>(&statusRunning))
.WillOnce(FutureArg<1>(&statusFinished));
driver.launchTasks(offers.get()[0].id(), {task});
// Scheduler should first receive TASK_STARTING followed by
// TASK_RUNNING and TASK_FINISHED from the executor.
AWAIT_READY(statusStarting);
EXPECT_EQ(TASK_STARTING, statusStarting->state());
EXPECT_EQ(TaskStatus::SOURCE_EXECUTOR, statusStarting->source());
AWAIT_READY(statusRunning);
EXPECT_EQ(TASK_RUNNING, statusRunning->state());
EXPECT_EQ(TaskStatus::SOURCE_EXECUTOR, statusRunning->source());
AWAIT_READY(statusFinished);
EXPECT_EQ(TASK_FINISHED, statusFinished->state());
EXPECT_EQ(TaskStatus::SOURCE_EXECUTOR, statusFinished->source());
driver.stop();
driver.join();
}
#ifndef __WINDOWS__
// This test runs a command _with_ the command user field set. The
// command will verify the assumption that the command is run as the
// specified user.
TEST_F(SlaveTest, ROOT_UNPRIVILEGED_USER_RunTaskWithCommandInfoWithUser)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched,
DEFAULT_FRAMEWORK_INFO,
master.get()->pid,
DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
Option<string> user = os::getenv("SUDO_USER");
ASSERT_SOME(user);
Result<uid_t> uid = os::getuid(user.get());
ASSERT_SOME(uid);
TaskInfo task = createTask(
offers->at(0),
"test `id -u` -eq " + stringify(uid.get()));
task.mutable_command()->set_user(user.get());
Future<TaskStatus> statusStarting;
Future<TaskStatus> statusRunning;
Future<TaskStatus> statusFinished;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusStarting))
.WillOnce(FutureArg<1>(&statusRunning))
.WillOnce(FutureArg<1>(&statusFinished));
driver.launchTasks(offers->at(0).id(), {task});
AWAIT_READY(statusStarting);
EXPECT_EQ(TASK_STARTING, statusStarting->state());
AWAIT_READY(statusRunning);
EXPECT_EQ(TASK_RUNNING, statusRunning->state());
AWAIT_READY(statusFinished);
EXPECT_EQ(TASK_FINISHED, statusFinished->state());
driver.stop();
driver.join();
}
// This test verifies that the agent gracefully drops tasks when
// a scheduler launches as a user that is not present on the agent.
//
// TODO(andschwa): Enable after `flags.switch_user` is added.
TEST_F(SlaveTest, ROOT_RunTaskWithCommandInfoWithInvalidUser)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
slave::Flags flags = CreateSlaveFlags();
// Enable `switch_user` so the agent is forced to
// evaluate the provided user name.
flags.switch_user = true;
Fetcher fetcher(flags);
Try<MesosContainerizer*> _containerizer =
MesosContainerizer::create(flags, false, &fetcher);
ASSERT_SOME(_containerizer);
Owned<MesosContainerizer> containerizer(_containerizer.get());
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), containerizer.get());
ASSERT_SOME(slave);
// Enable partition awareness so that we can expect `TASK_DROPPED`
// rather than `TASK_LOST`.
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
frameworkInfo.add_capabilities()->set_type(
FrameworkInfo::Capability::PARTITION_AWARE);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, frameworkInfo, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
const string taskUser = id::UUID::random().toString();
// Create a command that would trivially succeed if only
// the user was valid.
CommandInfo command;
command.set_user(taskUser);
command.set_shell(true);
command.set_value("true");
TaskInfo task = createTask(
offers->front().slave_id(),
offers->front().resources(),
command);
Future<TaskStatus> statusDropped;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusDropped));
driver.launchTasks(offers->front().id(), {task});
AWAIT_READY(statusDropped);
EXPECT_EQ(TASK_DROPPED, statusDropped->state());
EXPECT_EQ(TaskStatus::SOURCE_SLAVE, statusDropped->source());
// Since we expect the task to fail because the task user didn't
// exist, it's reasonable to check that the user was mentioned in
// the status message.
EXPECT_TRUE(strings::contains(statusDropped->message(), taskUser))
<< statusDropped->message();
driver.stop();
driver.join();
}
#endif // __WINDOWS__
// This test ensures that a status update acknowledgement from a
// non-leading master is ignored.
TEST_F(SlaveTest, IgnoreNonLeaderStatusUpdateAcknowledgement)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver schedDriver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&schedDriver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&schedDriver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
// We need to grab this message to get the scheduler's pid.
Future<Message> frameworkRegisteredMessage = FUTURE_MESSAGE(
Eq(FrameworkRegisteredMessage().GetTypeName()), master.get()->pid, _);
schedDriver.start();
AWAIT_READY(frameworkRegisteredMessage);
const UPID schedulerPid = frameworkRegisteredMessage->to;
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
TaskInfo task = createTask(offers.get()[0], "", DEFAULT_EXECUTOR_ID);
EXPECT_CALL(exec, registered(_, _, _, _));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> update;
EXPECT_CALL(sched, statusUpdate(&schedDriver, _))
.WillOnce(FutureArg<1>(&update));
// Pause the clock to prevent status update retries on the slave.
Clock::pause();
// Intercept the acknowledgement sent to the slave so that we can
// spoof the master's pid.
Future<StatusUpdateAcknowledgementMessage> acknowledgementMessage =
DROP_PROTOBUF(StatusUpdateAcknowledgementMessage(),
master.get()->pid,
slave.get()->pid);
Future<Nothing> _statusUpdateAcknowledgement =
FUTURE_DISPATCH(slave.get()->pid, &Slave::_statusUpdateAcknowledgement);
schedDriver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(update);
EXPECT_EQ(TASK_RUNNING, update->state());
AWAIT_READY(acknowledgementMessage);
// Send the acknowledgement to the slave with a non-leading master.
post(process::UPID("master@localhost:1"),
slave.get()->pid,
acknowledgementMessage.get());
// Make sure the acknowledgement was ignored.
Clock::settle();
ASSERT_TRUE(_statusUpdateAcknowledgement.isPending());
// Make sure the status update gets retried because the slave
// ignored the acknowledgement.
Future<TaskStatus> retriedUpdate;
EXPECT_CALL(sched, statusUpdate(&schedDriver, _))
.WillOnce(FutureArg<1>(&retriedUpdate));
Clock::advance(slave::STATUS_UPDATE_RETRY_INTERVAL_MIN);
AWAIT_READY(retriedUpdate);
// Ensure the slave receives and properly handles the ACK.
// Clock::settle() ensures that the slave successfully
// executes Slave::_statusUpdateAcknowledgement().
AWAIT_READY(_statusUpdateAcknowledgement);
Clock::settle();
Clock::resume();
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
schedDriver.stop();
schedDriver.join();
}
TEST_F(SlaveTest, MetricsInMetricsEndpoint)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
// Make sure slave finishes recovery.
Future<RegisterSlaveMessage> registerSlave = FUTURE_PROTOBUF(
RegisterSlaveMessage(), slave.get()->pid, master.get()->pid);
AWAIT_READY(registerSlave);
JSON::Object snapshot = Metrics();
EXPECT_EQ(1u, snapshot.values.count("slave/uptime_secs"));
EXPECT_EQ(1u, snapshot.values.count("slave/registered"));
EXPECT_EQ(1u, snapshot.values.count("slave/recovery_errors"));
EXPECT_EQ(1u, snapshot.values.count("slave/recovery_time_secs"));
EXPECT_EQ(1u, snapshot.values.count("slave/frameworks_active"));
EXPECT_EQ(1u, snapshot.values.count("slave/tasks_staging"));
EXPECT_EQ(1u, snapshot.values.count("slave/tasks_starting"));
EXPECT_EQ(1u, snapshot.values.count("slave/tasks_running"));
EXPECT_EQ(1u, snapshot.values.count("slave/tasks_killing"));
EXPECT_EQ(1u, snapshot.values.count("slave/tasks_finished"));
EXPECT_EQ(1u, snapshot.values.count("slave/tasks_failed"));
EXPECT_EQ(1u, snapshot.values.count("slave/tasks_killed"));
EXPECT_EQ(1u, snapshot.values.count("slave/tasks_lost"));
EXPECT_EQ(1u, snapshot.values.count("slave/tasks_gone"));
EXPECT_EQ(1u, snapshot.values.count("slave/executors_registering"));
EXPECT_EQ(1u, snapshot.values.count("slave/executors_running"));
EXPECT_EQ(1u, snapshot.values.count("slave/executors_terminating"));
EXPECT_EQ(1u, snapshot.values.count("slave/executors_terminated"));
EXPECT_EQ(1u, snapshot.values.count("slave/executors_preempted"));
EXPECT_EQ(1u, snapshot.values.count("slave/valid_status_updates"));
EXPECT_EQ(1u, snapshot.values.count("slave/invalid_status_updates"));
EXPECT_EQ(1u, snapshot.values.count("slave/valid_framework_messages"));
EXPECT_EQ(1u, snapshot.values.count("slave/invalid_framework_messages"));
EXPECT_EQ(1u, snapshot.values.count(
"slave/executor_directory_max_allowed_age_secs"));
EXPECT_EQ(1u, snapshot.values.count("slave/container_launch_errors"));
EXPECT_EQ(1u, snapshot.values.count("slave/cpus_total"));
EXPECT_EQ(1u, snapshot.values.count("slave/cpus_used"));
EXPECT_EQ(1u, snapshot.values.count("slave/cpus_percent"));
EXPECT_EQ(1u, snapshot.values.count("slave/cpus_revocable_total"));
EXPECT_EQ(1u, snapshot.values.count("slave/cpus_revocable_used"));
EXPECT_EQ(1u, snapshot.values.count("slave/cpus_revocable_percent"));
EXPECT_EQ(1u, snapshot.values.count("slave/gpus_total"));
EXPECT_EQ(1u, snapshot.values.count("slave/gpus_used"));
EXPECT_EQ(1u, snapshot.values.count("slave/gpus_percent"));
EXPECT_EQ(1u, snapshot.values.count("slave/gpus_revocable_total"));
EXPECT_EQ(1u, snapshot.values.count("slave/gpus_revocable_used"));
EXPECT_EQ(1u, snapshot.values.count("slave/gpus_revocable_percent"));
EXPECT_EQ(1u, snapshot.values.count("slave/mem_total"));
EXPECT_EQ(1u, snapshot.values.count("slave/mem_used"));
EXPECT_EQ(1u, snapshot.values.count("slave/mem_percent"));
EXPECT_EQ(1u, snapshot.values.count("slave/mem_revocable_total"));
EXPECT_EQ(1u, snapshot.values.count("slave/mem_revocable_used"));
EXPECT_EQ(1u, snapshot.values.count("slave/mem_revocable_percent"));
EXPECT_EQ(1u, snapshot.values.count("slave/disk_total"));
EXPECT_EQ(1u, snapshot.values.count("slave/disk_used"));
EXPECT_EQ(1u, snapshot.values.count("slave/disk_percent"));
EXPECT_EQ(1u, snapshot.values.count("slave/disk_revocable_total"));
EXPECT_EQ(1u, snapshot.values.count("slave/disk_revocable_used"));
EXPECT_EQ(1u, snapshot.values.count("slave/disk_revocable_percent"));
}
// Test to verify that we increment the container launch errors metric
// when we fail to launch a container.
TEST_F(SlaveTest, MetricsSlaveLaunchErrors)
{
// Start a master.
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
TestContainerizer containerizer;
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a slave.
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
const Offer offer = offers.get()[0];
// Verify that we start with no launch failures.
JSON::Object snapshot = Metrics();
EXPECT_EQ(0, snapshot.values["slave/container_launch_errors"]);
EXPECT_CALL(containerizer, launch(_, _, _, _))
.WillOnce(Return(Failure("Injected failure")));
Future<TaskStatus> failureUpdate;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&failureUpdate));
// The above injected containerizer failure also triggers executorLost.
EXPECT_CALL(sched, executorLost(&driver, DEFAULT_EXECUTOR_ID, _, _));
// Try to start a task
TaskInfo task = createTask(
offer.slave_id(),
Resources::parse("cpus:1;mem:32").get(),
SLEEP_COMMAND(1000),
DEFAULT_EXECUTOR_ID);
driver.launchTasks(offer.id(), {task});
AWAIT_READY(failureUpdate);
ASSERT_EQ(TASK_FAILED, failureUpdate->state());
// After failure injection, metrics should report a single failure.
snapshot = Metrics();
EXPECT_EQ(1, snapshot.values["slave/container_launch_errors"]);
driver.stop();
driver.join();
}
TEST_F(SlaveTest, StateEndpoint)
{
master::Flags masterFlags = this->CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
slave::Flags agentFlags = this->CreateSlaveFlags();
agentFlags.hostname = "localhost";
agentFlags.resources = "cpus:4;gpus:0;mem:2048;disk:512;ports:[33000-34000]";
agentFlags.attributes = "rack:abc;host:myhost";
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
// Capture the start time deterministically.
Clock::pause();
Future<Nothing> __recover = FUTURE_DISPATCH(_, &Slave::__recover);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &containerizer, agentFlags);
ASSERT_SOME(slave);
// Ensure slave has finished recovery.
AWAIT_READY(__recover);
Clock::settle();
Future<Response> response = process::http::get(
slave.get()->pid,
"state",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
Try<JSON::Object> parse = JSON::parse<JSON::Object>(response->body);
ASSERT_SOME(parse);
JSON::Object state = parse.get();
EXPECT_EQ(MESOS_VERSION, state.values["version"]);
if (build::GIT_SHA.isSome()) {
EXPECT_EQ(build::GIT_SHA.get(), state.values["git_sha"]);
}
if (build::GIT_BRANCH.isSome()) {
EXPECT_EQ(build::GIT_BRANCH.get(), state.values["git_branch"]);
}
if (build::GIT_TAG.isSome()) {
EXPECT_EQ(build::GIT_TAG.get(), state.values["git_tag"]);
}
EXPECT_EQ(build::DATE, state.values["build_date"]);
EXPECT_EQ(build::TIME, state.values["build_time"]);
EXPECT_EQ(build::USER, state.values["build_user"]);
// Even with a paused clock, the value of `start_time` from the
// state endpoint can differ slightly from the actual start time
// since the value went through a number of conversions (`double` to
// `string` to `JSON::Value`). Since `Clock::now` is a floating
// point value, the actual maximal possible difference between the
// real and observed value depends on both the mantissa and the
// exponent of the compared values; for simplicity we compare with
// an epsilon of `1` which allows for e.g., changes in the integer
// part of values close to an integer value.
ASSERT_TRUE(state.values["start_time"].is<JSON::Number>());
EXPECT_NEAR(
Clock::now().secs(),
state.values["start_time"].as<JSON::Number>().as<double>(),
1);
// TODO(bmahler): The slave must register for the 'id'
// to be non-empty.
ASSERT_TRUE(state.values["id"].is<JSON::String>());
EXPECT_EQ(stringify(slave.get()->pid), state.values["pid"]);
EXPECT_EQ(agentFlags.hostname.get(), state.values["hostname"]);
ASSERT_TRUE(state.values["capabilities"].is<JSON::Array>());
EXPECT_FALSE(state.values["capabilities"].as<JSON::Array>().values.empty());
JSON::Value slaveCapabilities = state.values.at("capabilities");
// Agents should have the following capabilities in the current
// implementation.
Try<JSON::Value> expectedCapabilities = JSON::parse(
R"~(
[
"MULTI_ROLE",
"HIERARCHICAL_ROLE",
"RESERVATION_REFINEMENT",
"RESOURCE_PROVIDER",
"RESIZE_VOLUME",
"AGENT_OPERATION_FEEDBACK",
"AGENT_DRAINING",
"TASK_RESOURCE_LIMITS"
]
)~");
ASSERT_SOME(expectedCapabilities);
EXPECT_TRUE(slaveCapabilities.contains(expectedCapabilities.get()));
Try<Resources> resources = Resources::parse(
agentFlags.resources.get(), agentFlags.default_role);
ASSERT_SOME(resources);
EXPECT_EQ(model(resources.get()), state.values["resources"]);
Attributes attributes = Attributes::parse(agentFlags.attributes.get());
EXPECT_EQ(model(attributes), state.values["attributes"]);
// TODO(bmahler): Test "master_hostname", "log_dir",
// "external_log_file".
ASSERT_TRUE(state.values["frameworks"].is<JSON::Array>());
EXPECT_TRUE(state.values["frameworks"].as<JSON::Array>().values.empty());
ASSERT_TRUE(
state.values["completed_frameworks"].is<JSON::Array>());
EXPECT_TRUE(
state.values["completed_frameworks"].as<JSON::Array>().values.empty());
// TODO(bmahler): Ensure this contains all the agentFlags.
ASSERT_TRUE(state.values["flags"].is<JSON::Object>());
EXPECT_FALSE(state.values["flags"].as<JSON::Object>().values.empty());
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
// Advance the clock to trigger both agent registration and a batch
// allocation.
Clock::advance(agentFlags.registration_backoff_factor);
Clock::advance(masterFlags.allocation_interval);
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
Resources executorResources = Resources::parse("cpus:0.1;mem:32").get();
executorResources.allocate("*");
TaskID taskId;
taskId.set_value("1");
TaskInfo task;
task.set_name("");
task.mutable_task_id()->MergeFrom(taskId);
task.mutable_slave_id()->MergeFrom(offers.get()[0].slave_id());
task.mutable_resources()->MergeFrom(
Resources(offers.get()[0].resources()) - executorResources);
task.mutable_executor()->MergeFrom(DEFAULT_EXECUTOR_INFO);
task.mutable_executor()->mutable_resources()->CopyFrom(executorResources);
EXPECT_CALL(exec, registered(_, _, _, _));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(status);
EXPECT_EQ(TASK_RUNNING, status->state());
response = process::http::get(
slave.get()->pid,
"state",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
parse = JSON::parse<JSON::Object>(response->body);
ASSERT_SOME(parse);
state = parse.get();
ASSERT_TRUE(state.values["frameworks"].is<JSON::Array>());
JSON::Array frameworks = state.values["frameworks"].as<JSON::Array>();
EXPECT_EQ(1u, frameworks.values.size());
ASSERT_TRUE(frameworks.values[0].is<JSON::Object>());
JSON::Object roles = {
{ "roles", JSON::Array { DEFAULT_FRAMEWORK_INFO.roles(0) } }
};
EXPECT_TRUE(frameworks.values[0].contains(roles));
JSON::Object framework = frameworks.values[0].as<JSON::Object>();
EXPECT_EQ("default", framework.values["name"]);
EXPECT_EQ(model(resources.get()), state.values["resources"]);
ASSERT_TRUE(framework.values["executors"].is<JSON::Array>());
JSON::Array executors = framework.values["executors"].as<JSON::Array>();
EXPECT_EQ(1u, executors.values.size());
ASSERT_TRUE(executors.values[0].is<JSON::Object>());
JSON::Object executor = executors.values[0].as<JSON::Object>();
EXPECT_EQ("default", executor.values["id"]);
EXPECT_EQ("", executor.values["source"]);
EXPECT_EQ("*", executor.values["role"]);
EXPECT_EQ(
model(Resources(task.resources()) +
Resources(task.executor().resources())),
executor.values["resources"]);
Result<JSON::Array> tasks = executor.find<JSON::Array>("tasks");
ASSERT_SOME(tasks);
EXPECT_EQ(1u, tasks->values.size());
JSON::Object taskJSON = tasks->values[0].as<JSON::Object>();
EXPECT_EQ("default", taskJSON.values["executor_id"]);
EXPECT_EQ("", taskJSON.values["name"]);
EXPECT_EQ(taskId.value(), taskJSON.values["id"]);
EXPECT_EQ("TASK_RUNNING", taskJSON.values["state"]);
EXPECT_EQ("*", taskJSON.values["role"]);
EXPECT_EQ(model(task.resources()), taskJSON.values["resources"]);
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// Verifies that requests to the agent's '/state' endpoint are successful when
// there are pending tasks from a task group. This test was used to confirm the
// fix for MESOS-7871.
TEST_F(SlaveTest, GetStateTaskGroupPending)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
auto executor = std::make_shared<v1::MockHTTPExecutor>();
Resources resources = Resources::parse("cpus:0.1;mem:32;disk:32").get();
ExecutorInfo executorInfo;
executorInfo.set_type(ExecutorInfo::DEFAULT);
executorInfo.mutable_executor_id()->CopyFrom(DEFAULT_EXECUTOR_ID);
executorInfo.mutable_resources()->CopyFrom(resources);
const ExecutorID& executorId = executorInfo.executor_id();
TestContainerizer containerizer(executorId, executor);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(
detector.get(),
&containerizer,
None(),
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
process::PID<Slave> slavePid = slave.get()->pid;
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers));
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(v1::DEFAULT_FRAMEWORK_INFO);
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(devolve(frameworkId));
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const SlaveID slaveId = devolve(offer.agent_id());
// Override the default expectation, which forwards calls to the agent's
// unmocked `_run()` method. Instead, we return a pending future to pause
// the original continuation so that tasks remain in the framework's
// 'pending' list.
Promise<Nothing> promise;
Future<Nothing> _run;
EXPECT_CALL(*slave.get()->mock(), _run(_, _, _, _, _, _))
.WillOnce(DoAll(FutureSatisfy(&_run),
Return(promise.future())));
// The executor should not be launched.
EXPECT_CALL(*executor, connected(_))
.Times(0);
v1::TaskInfo task1 = evolve(createTask(slaveId, resources, ""));
v1::TaskInfo task2 = evolve(createTask(slaveId, resources, ""));
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(task1);
taskGroup.add_tasks()->CopyFrom(task2);
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACCEPT);
Call::Accept* accept = call.mutable_accept();
accept->add_offer_ids()->CopyFrom(offer.id());
v1::Offer::Operation* operation = accept->add_operations();
operation->set_type(v1::Offer::Operation::LAUNCH_GROUP);
v1::Offer::Operation::LaunchGroup* launchGroup =
operation->mutable_launch_group();
launchGroup->mutable_executor()->CopyFrom(evolve(executorInfo));
launchGroup->mutable_task_group()->CopyFrom(taskGroup);
mesos.send(call);
}
// Wait for the tasks to be placed in 'pending'.
AWAIT_READY(_run);
Future<Response> response = process::http::get(
slavePid,
"state",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
// To confirm the fix for MESOS-7871, we simply verify that the
// agent doesn't crash when this request is made.
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
}
// This test checks that when a slave is in RECOVERING state it responds
// to HTTP requests for "/state" endpoint with ServiceUnavailable.
TEST_F(SlaveTest, StateEndpointUnavailableDuringRecovery)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer1(&exec);
TestContainerizer containerizer2;
slave::Flags flags = CreateSlaveFlags();
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &containerizer1, flags);
ASSERT_SOME(slave);
// Launch a task so that slave has something to recover after restart.
MockScheduler sched;
// Enable checkpointing for the framework.
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_checkpoint(true);
MesosSchedulerDriver driver(
&sched, frameworkInfo, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(LaunchTasks(DEFAULT_EXECUTOR_INFO, 1, 1, 512, "*"))
.WillRepeatedly(Return()); // Ignore subsequent offers.
EXPECT_CALL(exec, registered(_, _, _, _));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status))
.WillRepeatedly(Return()); // Ignore subsequent updates.
driver.start();
AWAIT_READY(status);
EXPECT_EQ(TASK_RUNNING, status->state());
// Need this expectation here because `TestContainerizer` doesn't do recovery
// and hence sets `MESOS_RECOVERY_TIMEOUT` as '0s' causing the executor driver
// to exit immediately after slave exit.
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
// Restart the slave.
slave.get()->terminate();
// Pause the clock to keep slave in RECOVERING state.
Clock::pause();
Future<Nothing> _recover = FUTURE_DISPATCH(_, &Slave::_recover);
slave = StartSlave(detector.get(), &containerizer2, flags);
ASSERT_SOME(slave);
// Ensure slave has setup the route for "/state".
AWAIT_READY(_recover);
Future<Response> response = process::http::get(
slave.get()->pid,
"state",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(ServiceUnavailable().status, response);
driver.stop();
driver.join();
}
// Tests that a client will receive an `Unauthorized` response when agent HTTP
// authentication is enabled and requests for the `/state` and `/flags`
// endpoints include invalid credentials or no credentials at all.
TEST_F(SlaveTest, HTTPEndpointsBadAuthentication)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
// A credential that will not be accepted by the agent.
Credential badCredential;
badCredential.set_principal("bad-principal");
badCredential.set_secret("bad-secret");
// Capture the start time deterministically.
Clock::pause();
Future<Nothing> recover = FUTURE_DISPATCH(_, &Slave::__recover);
Owned<MasterDetector> detector = master.get()->createDetector();
// HTTP authentication is enabled by default in `StartSlave`.
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
// Ensure slave has finished recovery.
AWAIT_READY(recover);
Clock::settle();
// Requests containing invalid credentials.
{
Future<Response> response = process::http::get(
slave.get()->pid,
"state",
None(),
createBasicAuthHeaders(badCredential));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(Unauthorized({}).status, response);
response = process::http::get(
slave.get()->pid,
"flags",
None(),
createBasicAuthHeaders(badCredential));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(Unauthorized({}).status, response);
}
// Requests containing no authentication headers.
{
Future<Response> response = process::http::get(slave.get()->pid, "state");
AWAIT_EXPECT_RESPONSE_STATUS_EQ(Unauthorized({}).status, response);
response = process::http::get(slave.get()->pid, "flags");
AWAIT_EXPECT_RESPONSE_STATUS_EQ(Unauthorized({}).status, response);
}
}
// Tests that a client can talk to read-only endpoints when read-only
// authentication is disabled.
TEST_F(SlaveTest, ReadonlyHTTPEndpointsNoAuthentication)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
// Capture the start time deterministically.
Clock::pause();
Future<Nothing> recover = FUTURE_DISPATCH(_, &Slave::__recover);
Owned<MasterDetector> detector = master.get()->createDetector();
slave::Flags flags = CreateSlaveFlags();
flags.authenticate_http_readonly = false;
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), flags);
ASSERT_SOME(slave);
// Ensure slave has finished recovery.
AWAIT_READY(recover);
Clock::settle();
// Requests containing no authentication headers.
{
Future<Response> response = process::http::get(slave.get()->pid, "state");
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
response = process::http::get(slave.get()->pid, "flags");
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
response = process::http::get(slave.get()->pid, "containers");
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
}
}
// Since executor authentication currently has SSL as a dependency, we cannot
// test executor authentication when Mesos has not been built with SSL.
#ifdef USE_SSL_SOCKET
// This test verifies that HTTP executor SUBSCRIBE and LAUNCH_NESTED_CONTAINER
// calls fail if the executor provides an incorrectly-signed authentication
// token with valid claims.
TEST_F(SlaveTest, HTTPExecutorBadAuthentication)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
auto executor = std::make_shared<v1::MockHTTPExecutor>();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::ExecutorInfo executorInfo;
executorInfo.set_type(v1::ExecutorInfo::DEFAULT);
executorInfo.mutable_executor_id()->CopyFrom(v1::DEFAULT_EXECUTOR_ID);
executorInfo.mutable_resources()->CopyFrom(resources);
Owned<TestContainerizer> containerizer(
new TestContainerizer(devolve(executorInfo.executor_id()), executor));
Owned<MockSecretGenerator> mockSecretGenerator(new MockSecretGenerator());
Try<Owned<cluster::Slave>> slave = StartSlave(
detector.get(),
containerizer.get(),
mockSecretGenerator.get());
ASSERT_SOME(slave);
process::PID<Slave> slavePid = slave.get()->pid;
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
mesos.send(v1::createCallSubscribe(v1::DEFAULT_FRAMEWORK_INFO));
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
executorInfo.mutable_framework_id()->CopyFrom(frameworkId);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
Future<v1::executor::Mesos*> executorLib;
EXPECT_CALL(*executor, connected(_))
.WillOnce(FutureArg<0>(&executorLib));
Promise<Secret> secret;
Future<Principal> principal;
EXPECT_CALL(*mockSecretGenerator, generate(_))
.WillOnce(DoAll(FutureArg<0>(&principal),
Return(secret.future())));
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
{
v1::TaskInfo taskInfo =
v1::createTask(agentId, resources, SLEEP_COMMAND(1000));
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(taskInfo);
v1::scheduler::Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(v1::scheduler::Call::ACCEPT);
v1::scheduler::Call::Accept* accept = call.mutable_accept();
accept->add_offer_ids()->CopyFrom(offer.id());
v1::Offer::Operation* operation = accept->add_operations();
operation->set_type(v1::Offer::Operation::LAUNCH_GROUP);
v1::Offer::Operation::LaunchGroup* launchGroup =
operation->mutable_launch_group();
launchGroup->mutable_executor()->CopyFrom(executorInfo);
launchGroup->mutable_task_group()->CopyFrom(taskGroup);
mesos.send(call);
}
AWAIT_READY(principal);
// Create a secret generator initialized with an incorrect key.
Owned<JWTSecretGenerator> jwtSecretGenerator(
new JWTSecretGenerator("incorrect_key"));
Future<Secret> authenticationToken =
jwtSecretGenerator->generate(principal.get());
AWAIT_READY(authenticationToken);
secret.set(authenticationToken.get());
{
AWAIT_READY(executorLib);
v1::executor::Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.mutable_executor_id()->CopyFrom(v1::DEFAULT_EXECUTOR_ID);
call.set_type(v1::executor::Call::SUBSCRIBE);
call.mutable_subscribe();
executorLib.get()->send(call);
Future<v1::executor::Event::Error> error;
EXPECT_CALL(*executor, error(_, _))
.WillOnce(FutureArg<1>(&error));
AWAIT_READY(error);
EXPECT_EQ(
error->message(),
"Received unexpected '401 Unauthorized' () for SUBSCRIBE");
}
{
ASSERT_TRUE(principal->claims.contains("cid"));
v1::ContainerID parentContainerId;
parentContainerId.set_value(principal->claims.at("cid"));
v1::ContainerID containerId;
containerId.set_value(id::UUID::random().toString());
containerId.mutable_parent()->CopyFrom(parentContainerId);
v1::agent::Call call;
call.set_type(v1::agent::Call::LAUNCH_NESTED_CONTAINER);
call.mutable_launch_nested_container()->mutable_container_id()
->CopyFrom(containerId);
process::http::Headers headers;
headers["Authorization"] = "Bearer " + authenticationToken->value().data();
Future<Response> response = process::http::post(
slavePid,
"api/v1",
headers,
serialize(ContentType::PROTOBUF, call),
stringify(ContentType::PROTOBUF));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(Unauthorized({}).status, response);
ASSERT_TRUE(response->headers.contains("WWW-Authenticate"));
ASSERT_TRUE(strings::contains(
response->headers.at("WWW-Authenticate"),
"Invalid JWT: Token signature does not match"));
}
}
#endif // USE_SSL_SOCKET
// This test verifies correct handling of statistics endpoint when
// there is no exeuctor running.
TEST_F(SlaveTest, StatisticsEndpointNoExecutor)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
Future<Response> response = process::http::get(
slave.get()->pid,
"/monitor/statistics",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
AWAIT_EXPECT_RESPONSE_BODY_EQ("[]", response);
}
// This test verifies the correct handling of the statistics
// endpoint when statistics is missing in ResourceUsage.
TEST_F(SlaveTest, StatisticsEndpointMissingStatistics)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
StandaloneMasterDetector detector(master.get()->pid);
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
None());
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
EXPECT_CALL(exec, registered(_, _, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
const Offer& offer = offers.get()[0];
TaskInfo task = createTask(
offer.slave_id(),
Resources::parse("cpus:0.1;mem:32").get(),
SLEEP_COMMAND(1000),
exec.id);
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
driver.launchTasks(offer.id(), {task});
AWAIT_READY(status);
EXPECT_EQ(TASK_RUNNING, status->state());
// Set up the containerizer so the next usage() will fail.
EXPECT_CALL(containerizer, usage(_))
.WillOnce(Return(Failure("Injected failure")));
Future<Response> response = process::http::get(
slave.get()->pid,
"monitor/statistics",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_READY(response);
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
AWAIT_EXPECT_RESPONSE_BODY_EQ("[]", response);
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// This test verifies the correct response of /monitor/statistics endpoint
// when ResourceUsage collection fails.
TEST_F(SlaveTest, StatisticsEndpointGetResourceUsageFailed)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
StandaloneMasterDetector detector(master.get()->pid);
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
None(),
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
EXPECT_CALL(*slave.get()->mock(), usage())
.WillOnce(Return(Failure("Resource Collection Failure")));
slave.get()->start();
Future<Response> response = process::http::get(
slave.get()->pid,
"monitor/statistics",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_READY(response);
AWAIT_EXPECT_RESPONSE_STATUS_EQ(InternalServerError().status, response);
}
// This is an end-to-end test that verifies that the slave returns the
// correct ResourceUsage based on the currently running executors, and
// the values returned by the /monitor/statistics endpoint are as expected.
TEST_F(SlaveTest, StatisticsEndpointRunningExecutor)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
const Offer& offer = offers.get()[0];
// Launch a task and wait until it is in RUNNING status.
TaskInfo task = createTask(
offer.slave_id(),
Resources::parse("cpus:1;mem:32").get(),
SLEEP_COMMAND(1000));
Future<TaskStatus> statusStarting;
Future<TaskStatus> statusRunning;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusStarting))
.WillOnce(FutureArg<1>(&statusRunning));
driver.launchTasks(offer.id(), {task});
AWAIT_READY(statusStarting);
EXPECT_EQ(task.task_id(), statusStarting->task_id());
EXPECT_EQ(TASK_STARTING, statusStarting->state());
AWAIT_READY(statusRunning);
EXPECT_EQ(task.task_id(), statusRunning->task_id());
EXPECT_EQ(TASK_RUNNING, statusRunning->state());
// Hit the statistics endpoint and expect the response contains the
// resource statistics for the running container.
Future<Response> response = process::http::get(
slave.get()->pid,
"monitor/statistics",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
// Verify that the statistics in the response contains the proper
// resource limits for the container.
Try<JSON::Value> value = JSON::parse(response->body);
ASSERT_SOME(value);
Try<JSON::Value> expected = JSON::parse(strings::format(
"[{"
"\"statistics\":{"
"\"cpus_soft_limit\":%g,"
"\"mem_limit_bytes\":%lu,"
"\"mem_soft_limit_bytes\":%lu"
"}"
"}]",
1 + slave::DEFAULT_EXECUTOR_CPUS,
(Megabytes(32) + slave::DEFAULT_EXECUTOR_MEM).bytes(),
(Megabytes(32) + slave::DEFAULT_EXECUTOR_MEM).bytes()).get());
ASSERT_SOME(expected);
EXPECT_TRUE(value->contains(expected.get()));
driver.stop();
driver.join();
}
// This test confirms that an agent's statistics endpoint is
// authenticated. We rely on the agent implicitly having HTTP
// authentication enabled.
TEST_F(SlaveTest, StatisticsEndpointAuthentication)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> agent = StartSlave(detector.get());
ASSERT_SOME(agent);
const string statisticsEndpoint = "monitor/statistics";
// Unauthenticated requests are rejected.
{
Future<Response> response = process::http::get(
agent.get()->pid,
statisticsEndpoint);
AWAIT_EXPECT_RESPONSE_STATUS_EQ(Unauthorized({}).status, response);
}
// Incorrectly authenticated requests are rejected.
{
Credential badCredential;
badCredential.set_principal("badPrincipal");
badCredential.set_secret("badSecret");
Future<Response> response = process::http::get(
agent.get()->pid,
statisticsEndpoint,
None(),
createBasicAuthHeaders(badCredential));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(Unauthorized({}).status, response);
}
// Correctly authenticated requests succeed.
{
Future<Response> response = process::http::get(
agent.get()->pid,
statisticsEndpoint,
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
}
}
// This test verifies correct handling of containers endpoint when
// there is no exeuctor running.
TEST_F(SlaveTest, ContainersEndpointNoExecutor)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
Future<Response> response = process::http::get(
slave.get()->pid,
"containers",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
AWAIT_EXPECT_RESPONSE_BODY_EQ("[]", response);
}
// This is an end-to-end test that verifies that the slave returns the
// correct container status and resource statistics based on the currently
// running executors, and ensures that '/containers' endpoint returns the
// correct container when it is provided a container ID query parameter.
TEST_F(SlaveTest, ContainersEndpoint)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
// Create two executors so that we can launch tasks in two separate
// containers.
ExecutorInfo executor1 = createExecutorInfo("executor-1", "exit 1");
ExecutorInfo executor2 = createExecutorInfo("executor-2", "exit 1");
MockExecutor exec1(executor1.executor_id());
MockExecutor exec2(executor2.executor_id());
hashmap<ExecutorID, Executor*> execs;
execs[executor1.executor_id()] = &exec1;
execs[executor2.executor_id()] = &exec2;
TestContainerizer containerizer(execs);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
// Launch two tasks, each under a different executor.
vector<TaskInfo> tasks;
TaskInfo task1;
{
task1.set_name("");
task1.mutable_task_id()->set_value("1");
task1.mutable_slave_id()->MergeFrom(offers->front().slave_id());
task1.mutable_resources()->MergeFrom(
Resources::parse("cpus:1;mem:512").get());
task1.mutable_executor()->MergeFrom(executor1);
tasks.push_back(task1);
}
TaskInfo task2;
{
task2.set_name("");
task2.mutable_task_id()->set_value("2");
task2.mutable_slave_id()->MergeFrom(offers->front().slave_id());
task2.mutable_resources()->MergeFrom(
Resources::parse("cpus:1;mem:512").get());
task2.mutable_executor()->MergeFrom(executor2);
tasks.push_back(task2);
}
EXPECT_CALL(exec1, registered(_, _, _, _));
Future<TaskInfo> launchedTask1;
EXPECT_CALL(exec1, launchTask(_, _))
.WillOnce(DoAll(SendStatusUpdateFromTask(TASK_RUNNING),
FutureArg<1>(&launchedTask1)));
EXPECT_CALL(exec2, registered(_, _, _, _));
Future<TaskInfo> launchedTask2;
EXPECT_CALL(exec2, launchTask(_, _))
.WillOnce(DoAll(SendStatusUpdateFromTask(TASK_RUNNING),
FutureArg<1>(&launchedTask2)));
Future<TaskStatus> status1, status2;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status1))
.WillOnce(FutureArg<1>(&status2));
driver.launchTasks(offers->front().id(), tasks);
AWAIT_READY(launchedTask1);
EXPECT_EQ(task1.task_id(), launchedTask1->task_id());
AWAIT_READY(launchedTask2);
EXPECT_EQ(task2.task_id(), launchedTask2->task_id());
AWAIT_READY(status1);
EXPECT_EQ(TASK_RUNNING, status1->state());
AWAIT_READY(status2);
EXPECT_EQ(TASK_RUNNING, status2->state());
// Prepare container statistics.
ResourceStatistics statistics1;
statistics1.set_mem_limit_bytes(2048);
ResourceStatistics statistics2;
statistics2.set_mem_limit_bytes(2048);
// Get the container ID and return simulated statistics.
Future<ContainerID> containerId1;
Future<ContainerID> containerId2;
// Will be called twice during the first request. We extract the assigned
// container IDs for use when requesting information on a single container.
EXPECT_CALL(containerizer, usage(_))
.WillOnce(DoAll(FutureArg<0>(&containerId1), Return(statistics1)))
.WillOnce(DoAll(FutureArg<0>(&containerId2), Return(statistics2)));
// Construct the container statuses to be returned. Note that
// these container IDs will be different than the actual container
// IDs assigned by the agent, but creating them here allows us to
// easily confirm the output of '/containers'.
ContainerStatus containerStatus1;
ContainerStatus containerStatus2;
ContainerID parent;
parent.set_value("parent");
{
ContainerID child;
child.set_value("child1");
child.mutable_parent()->CopyFrom(parent);
containerStatus1.mutable_container_id()->CopyFrom(child);
CgroupInfo* cgroupInfo = containerStatus1.mutable_cgroup_info();
CgroupInfo::NetCls* netCls = cgroupInfo->mutable_net_cls();
netCls->set_classid(42);
NetworkInfo* networkInfo = containerStatus1.add_network_infos();
NetworkInfo::IPAddress* ipAddr = networkInfo->add_ip_addresses();
ipAddr->set_ip_address("192.168.1.20");
}
{
ContainerID child;
child.set_value("child2");
child.mutable_parent()->CopyFrom(parent);
containerStatus2.mutable_container_id()->CopyFrom(child);
CgroupInfo* cgroupInfo = containerStatus2.mutable_cgroup_info();
CgroupInfo::NetCls* netCls = cgroupInfo->mutable_net_cls();
netCls->set_classid(42);
NetworkInfo* networkInfo = containerStatus2.add_network_infos();
NetworkInfo::IPAddress* ipAddr = networkInfo->add_ip_addresses();
ipAddr->set_ip_address("192.168.1.21");
}
// Will be called twice during the first request.
EXPECT_CALL(containerizer, status(_))
.WillOnce(Return(containerStatus1))
.WillOnce(Return(containerStatus2));
// Request information about all containers.
{
Future<Response> response = process::http::get(
slave.get()->pid,
"containers",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
Try<JSON::Value> value = JSON::parse<JSON::Value>(response->body);
ASSERT_SOME(value);
JSON::Array array = value->as<JSON::Array>();
EXPECT_TRUE(array.values.size() == 2);
Try<JSON::Value> containerJson1 = JSON::parse(
"{"
"\"executor_name\":\"\","
"\"source\":\"\","
"\"statistics\":{"
"\"mem_limit_bytes\":2048"
"},"
"\"status\":{"
"\"container_id\":{"
"\"parent\":{\"value\":\"parent\"},"
"\"value\":\"child1\""
"},"
"\"cgroup_info\":{\"net_cls\":{\"classid\":42}},"
"\"network_infos\":[{"
"\"ip_addresses\":[{\"ip_address\":\"192.168.1.20\"}]"
"}]"
"}"
"}");
Try<JSON::Value> containerJson2 = JSON::parse(
"{"
"\"executor_name\":\"\","
"\"source\":\"\","
"\"statistics\":{"
"\"mem_limit_bytes\":2048"
"},"
"\"status\":{"
"\"container_id\":{"
"\"parent\":{\"value\":\"parent\"},"
"\"value\":\"child2\""
"},"
"\"cgroup_info\":{\"net_cls\":{\"classid\":42}},"
"\"network_infos\":[{"
"\"ip_addresses\":[{\"ip_address\":\"192.168.1.21\"}]"
"}]"
"}"
"}");
// Since containers are stored in a hashmap, there is no strict guarantee of
// their ordering when listed. For this reason, we test both possibilities.
if (array.values[0].contains(containerJson1.get())) {
ASSERT_TRUE(array.values[1].contains(containerJson2.get()));
} else {
ASSERT_TRUE(array.values[0].contains(containerJson2.get()));
ASSERT_TRUE(array.values[1].contains(containerJson1.get()));
}
}
AWAIT_READY(containerId1);
AWAIT_READY(containerId2);
// Will be called once during the second request.
EXPECT_CALL(containerizer, usage(_))
.WillOnce(Return(statistics1));
// Will be called once during the second request and might be called if
// the `TASK_FAILED` update reaches the agent before the test finishes.
EXPECT_CALL(containerizer, status(_))
.WillOnce(Return(containerStatus1))
.WillRepeatedly(Return(containerStatus1));
{
Future<Response> response = process::http::get(
slave.get()->pid,
"containers?container_id=" + containerId1->value(),
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
Try<JSON::Value> value = JSON::parse<JSON::Value>(response->body);
ASSERT_SOME(value);
JSON::Array array = value->as<JSON::Array>();
EXPECT_TRUE(array.values.size() == 1);
Try<JSON::Value> expected = JSON::parse(
"[{"
"\"container_id\":\"" + containerId1->value() + "\","
"\"executor_name\":\"\","
"\"source\":\"\","
"\"statistics\":{"
"\"mem_limit_bytes\":2048"
"},"
"\"status\":{"
"\"container_id\":{"
"\"parent\":{\"value\":\"parent\"},"
"\"value\":\"child1\""
"},"
"\"cgroup_info\":{\"net_cls\":{\"classid\":42}},"
"\"network_infos\":[{"
"\"ip_addresses\":[{\"ip_address\":\"192.168.1.20\"}]"
"}]"
"}"
"}]");
ASSERT_SOME(expected);
EXPECT_TRUE(value->contains(expected.get()));
}
EXPECT_CALL(exec1, shutdown(_))
.Times(AtMost(1));
EXPECT_CALL(exec2, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// This test ensures that `/containerizer/debug` endpoint returns a non-empty
// list of pending futures when an isolator becomes unresponsive during
// container launch.
TEST_F(SlaveTest, ROOT_ContainerizerDebugEndpoint)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
slave::Flags flags = CreateSlaveFlags();
Try<Launcher*> _launcher = SubprocessLauncher::create(flags);
ASSERT_SOME(_launcher);
Owned<Launcher> launcher(_launcher.get());
MockIsolator* mockIsolator = new MockIsolator();
Future<Nothing> prepare;
Promise<Option<ContainerLaunchInfo>> promise;
EXPECT_CALL(*mockIsolator, recover(_, _))
.WillOnce(Return(Nothing()));
// Simulate a long prepare from the isolator.
EXPECT_CALL(*mockIsolator, prepare(_, _))
.WillOnce(DoAll(FutureSatisfy(&prepare),
Return(promise.future())));
EXPECT_CALL(*mockIsolator, update(_, _, _))
.WillOnce(Return(Nothing()));
// Wrap `mockIsolator` in `PendingFutureTracker`.
Try<PendingFutureTracker*> _futureTracker = PendingFutureTracker::create();
ASSERT_SOME(_futureTracker);
Owned<PendingFutureTracker> futureTracker(_futureTracker.get());
Owned<Isolator> isolator = Owned<Isolator>(new IsolatorTracker(
Owned<Isolator>(mockIsolator), "MockIsolator", futureTracker.get()));
Fetcher fetcher(flags);
Try<Owned<Provisioner>> provisioner = Provisioner::create(flags);
ASSERT_SOME(provisioner);
Try<MesosContainerizer*> create = MesosContainerizer::create(
flags,
true,
&fetcher,
nullptr,
launcher,
provisioner->share(),
{isolator});
ASSERT_SOME(create);
Owned<MesosContainerizer> containerizer(create.get());
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave =
StartSlave(SlaveOptions(detector.get())
.withContainerizer(containerizer.get())
.withFutureTracker(futureTracker.get()));
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
const Offer& offer = offers.get()[0];
// Launch a task and wait until it is in RUNNING status.
TaskInfo task = createTask(
offer.slave_id(),
Resources::parse("cpus:1;mem:32").get(),
SLEEP_COMMAND(1000));
Future<TaskStatus> statusStarting;
Future<TaskStatus> statusRunning;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusStarting))
.WillOnce(FutureArg<1>(&statusRunning));
driver.launchTasks(offer.id(), {task});
AWAIT_READY(prepare);
{
Future<Response> response = process::http::get(
slave.get()->pid,
"containerizer/debug",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
Try<JSON::Object> parse = JSON::parse<JSON::Object>(response->body);
ASSERT_SOME(parse);
ASSERT_FALSE(parse->at<JSON::Array>("pending")->values.empty());
ASSERT_TRUE(strings::contains(response->body, "MockIsolator::prepare"));
}
// Once the future returned by the `prepare` method becomes ready,
// the task should start successfully and no pending futures should be
// contained in the output returned by `/containerizer/debug` endpoint.
promise.set(Option<ContainerLaunchInfo>(ContainerLaunchInfo()));
AWAIT_READY(statusStarting);
EXPECT_EQ(task.task_id(), statusStarting->task_id());
EXPECT_EQ(TASK_STARTING, statusStarting->state());
AWAIT_READY(statusRunning);
EXPECT_EQ(task.task_id(), statusRunning->task_id());
EXPECT_EQ(TASK_RUNNING, statusRunning->state());
{
Future<Response> response = process::http::get(
slave.get()->pid,
"containerizer/debug",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
Try<JSON::Object> parse = JSON::parse<JSON::Object>(response->body);
ASSERT_SOME(parse);
ASSERT_TRUE(parse->at<JSON::Array>("pending")->values.empty());
}
driver.stop();
driver.join();
}
// This test ensures that when a slave is shutting down, it will not
// try to reregister with the master.
//
// TODO(alexr): Enable after MESOS-3509 is resolved.
TEST_F(SlaveTest, DISABLED_TerminatingSlaveDoesNotReregister)
{
// Start a master.
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
// Create a MockExecutor to enable us to catch
// ShutdownExecutorMessage later.
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
// Create a StandaloneMasterDetector to enable the slave to trigger
// re-registration later.
StandaloneMasterDetector detector(master.get()->pid);
slave::Flags flags = CreateSlaveFlags();
// Make the executor_shutdown_grace_period to be much longer than
// REGISTER_RETRY_INTERVAL, so that the slave will at least call
// call doReliableRegistration() once before the slave is actually
// terminated.
flags.executor_shutdown_grace_period = slave::REGISTER_RETRY_INTERVAL_MAX * 2;
// Start a slave.
Try<Owned<cluster::Slave>> slave =
StartSlave(&detector, &containerizer, flags);
ASSERT_SOME(slave);
// Create a task on the slave.
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
// Launch a task that uses less resource than the
// default(cpus:2, mem:1024).
EXPECT_CALL(sched, resourceOffers(_, _))
.WillOnce(LaunchTasks(DEFAULT_EXECUTOR_INFO, 1, 1, 64, "*"))
.WillRepeatedly(Return()); // Ignore subsequent offers.
EXPECT_CALL(exec, registered(_, _, _, _));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status))
.WillRepeatedly(Return()); // Ignore subsequent updates.
driver.start();
AWAIT_READY(status);
EXPECT_EQ(TASK_RUNNING, status->state());
// Pause the clock here so that after detecting a new master,
// the slave will not send multiple reregister messages
// before we change its state to TERMINATING.
Clock::pause();
Future<SlaveReregisteredMessage> slaveReregisteredMessage =
DROP_PROTOBUF(
SlaveReregisteredMessage(),
master.get()->pid,
slave.get()->pid);
// Simulate a new master detected event on the slave,
// so that the slave will do a re-registration.
detector.appoint(master.get()->pid);
// Make sure the slave has entered doReliableRegistration()
// before we change the slave's state.
AWAIT_READY(slaveReregisteredMessage);
// Setup an expectation that the master should not receive any
// ReregisterSlaveMessage in the future.
EXPECT_NO_FUTURE_PROTOBUFS(
ReregisterSlaveMessage(),
slave.get()->pid,
master.get()->pid);
// Drop the ShutdownExecutorMessage, so that the slave will
// stay in TERMINATING for a while.
DROP_PROTOBUFS(ShutdownExecutorMessage(), slave.get()->pid, _);
Future<Nothing> executorLost;
EXPECT_CALL(sched, executorLost(&driver, DEFAULT_EXECUTOR_ID, _, _))
.WillOnce(FutureSatisfy(&executorLost));
// Send a ShutdownMessage instead of calling Stop() directly
// to avoid blocking.
post(master.get()->pid, slave.get()->pid, ShutdownMessage());
// Advance the clock to trigger doReliableRegistration().
Clock::advance(slave::REGISTER_RETRY_INTERVAL_MAX * 2);
Clock::settle();
Clock::resume();
AWAIT_READY(executorLost);
driver.stop();
driver.join();
}
// This test verifies the slave will destroy a container if, when
// receiving a terminal status task update, updating the container's
// resources fails. A non-partition-aware framework should receive
// TASK_LOST in this situation.
TEST_F(SlaveTest, TerminalTaskContainerizerUpdateFailsWithLost)
{
// Start a master.
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
EXPECT_CALL(exec, registered(_, _, _, _));
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a slave.
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
// Connect a non-partition-aware scheduler.
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
Offer offer = offers.get()[0];
// Start two tasks.
vector<TaskInfo> tasks;
tasks.push_back(createTask(
offer.slave_id(),
Resources::parse("cpus:0.1;mem:32").get(),
SLEEP_COMMAND(1000),
exec.id));
tasks.push_back(createTask(
offer.slave_id(),
Resources::parse("cpus:0.1;mem:32").get(),
SLEEP_COMMAND(1000),
exec.id));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> status1, status2, status3, status4;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status1))
.WillOnce(FutureArg<1>(&status2))
.WillOnce(FutureArg<1>(&status3))
.WillOnce(FutureArg<1>(&status4));
driver.launchTasks(offer.id(), tasks);
AWAIT_READY(status1);
EXPECT_EQ(TASK_RUNNING, status1->state());
AWAIT_READY(status2);
EXPECT_EQ(TASK_RUNNING, status2->state());
// Set up the containerizer so the next update() will fail.
EXPECT_CALL(containerizer, update(_, _, _))
.WillOnce(Return(Failure("update() failed")))
.WillRepeatedly(Return(Nothing()));
EXPECT_CALL(exec, killTask(_, _))
.WillOnce(SendStatusUpdateFromTaskID(TASK_KILLED));
Future<Nothing> executorLost;
EXPECT_CALL(sched, executorLost(&driver, DEFAULT_EXECUTOR_ID, _, _))
.WillOnce(FutureSatisfy(&executorLost));
// Kill one of the tasks. The failed update should result in the
// second task going lost when the container is destroyed.
driver.killTask(tasks[0].task_id());
AWAIT_READY(status3);
EXPECT_EQ(TASK_KILLED, status3->state());
EXPECT_EQ(TaskStatus::SOURCE_EXECUTOR, status3->source());
AWAIT_READY(status4);
EXPECT_EQ(TASK_LOST, status4->state());
EXPECT_EQ(TaskStatus::SOURCE_SLAVE, status4->source());
EXPECT_EQ(TaskStatus::REASON_CONTAINER_UPDATE_FAILED, status4->reason());
AWAIT_READY(executorLost);
JSON::Object stats = Metrics();
EXPECT_EQ(0, stats.values["slave/tasks_gone"]);
EXPECT_EQ(1, stats.values["slave/tasks_lost"]);
driver.stop();
driver.join();
}
// This test verifies the slave will destroy a container if, when
// receiving a terminal status task update, updating the container's
// resources fails. A partition-aware framework should receive
// TASK_GONE in this situation.
TEST_F(SlaveTest, TerminalTaskContainerizerUpdateFailsWithGone)
{
// Start a master.
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
EXPECT_CALL(exec, registered(_, _, _, _));
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a slave.
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
// Connect a partition-aware scheduler.
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
frameworkInfo.add_capabilities()->set_type(
FrameworkInfo::Capability::PARTITION_AWARE);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, frameworkInfo, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
Offer offer = offers.get()[0];
// Start two tasks.
vector<TaskInfo> tasks;
tasks.push_back(createTask(
offer.slave_id(),
Resources::parse("cpus:0.1;mem:32").get(),
SLEEP_COMMAND(1000),
exec.id));
tasks.push_back(createTask(
offer.slave_id(),
Resources::parse("cpus:0.1;mem:32").get(),
SLEEP_COMMAND(1000),
exec.id));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> status1, status2, status3, status4;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status1))
.WillOnce(FutureArg<1>(&status2))
.WillOnce(FutureArg<1>(&status3))
.WillOnce(FutureArg<1>(&status4));
driver.launchTasks(offer.id(), tasks);
AWAIT_READY(status1);
EXPECT_EQ(TASK_RUNNING, status1->state());
AWAIT_READY(status2);
EXPECT_EQ(TASK_RUNNING, status2->state());
// Set up the containerizer so the next update() will fail.
EXPECT_CALL(containerizer, update(_, _, _))
.WillOnce(Return(Failure("update() failed")))
.WillRepeatedly(Return(Nothing()));
EXPECT_CALL(exec, killTask(_, _))
.WillOnce(SendStatusUpdateFromTaskID(TASK_KILLED));
Future<Nothing> executorLost;
EXPECT_CALL(sched, executorLost(&driver, DEFAULT_EXECUTOR_ID, _, _))
.WillOnce(FutureSatisfy(&executorLost));
// Kill one of the tasks. The failed update should result in the
// second task going lost when the container is destroyed.
driver.killTask(tasks[0].task_id());
AWAIT_READY(status3);
EXPECT_EQ(TASK_KILLED, status3->state());
EXPECT_EQ(TaskStatus::SOURCE_EXECUTOR, status3->source());
AWAIT_READY(status4);
EXPECT_EQ(TASK_GONE, status4->state());
EXPECT_EQ(TaskStatus::SOURCE_SLAVE, status4->source());
EXPECT_EQ(TaskStatus::REASON_CONTAINER_UPDATE_FAILED, status4->reason());
AWAIT_READY(executorLost);
JSON::Object stats = Metrics();
EXPECT_EQ(1, stats.values["slave/tasks_gone"]);
EXPECT_EQ(0, stats.values["slave/tasks_lost"]);
driver.stop();
driver.join();
}
// This test verifies that the resources of a container will be
// updated before tasks are sent to the executor.
TEST_F(SlaveTest, ContainerUpdatedBeforeTaskReachesExecutor)
{
// Start a master.
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
EXPECT_CALL(exec, registered(_, _, _, _));
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a slave.
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(LaunchTasks(DEFAULT_EXECUTOR_INFO, 1, 1, 128, "*"))
.WillRepeatedly(Return()); // Ignore subsequent offers.
// This is used to determine which of the following finishes first:
// `containerizer->update` or `exec->launchTask`. We want to make
// sure that containerizer update always finishes before the task is
// sent to the executor.
testing::Sequence sequence;
EXPECT_CALL(containerizer, update(_, _, _))
.InSequence(sequence)
.WillOnce(Return(Nothing()));
EXPECT_CALL(exec, launchTask(_, _))
.InSequence(sequence)
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
driver.start();
AWAIT_READY(status);
EXPECT_EQ(TASK_RUNNING, status->state());
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// This test verifies the slave will destroy a container if updating
// the container's resources fails during task launch.
TEST_F(SlaveTest, TaskLaunchContainerizerUpdateFails)
{
// Start a master.
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a slave.
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(LaunchTasks(DEFAULT_EXECUTOR_INFO, 1, 1, 128, "*"))
.WillRepeatedly(Return()); // Ignore subsequent offers.
// The executor may not receive the ExecutorRegisteredMessage if the
// container is destroyed before that.
EXPECT_CALL(exec, registered(_, _, _, _))
.Times(AtMost(1));
// Set up the containerizer so update() will fail.
EXPECT_CALL(containerizer, update(_, _, _))
.WillOnce(Return(Failure("update() failed")))
.WillRepeatedly(Return(Nothing()));
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
EXPECT_CALL(sched, executorLost(&driver, DEFAULT_EXECUTOR_ID, _, _));
driver.start();
AWAIT_READY(status);
EXPECT_EQ(TASK_LOST, status->state());
EXPECT_EQ(TaskStatus::SOURCE_SLAVE, status->source());
EXPECT_EQ(TaskStatus::REASON_CONTAINER_UPDATE_FAILED, status->reason());
driver.stop();
driver.join();
}
// This test ensures that the slave will reregister with the master
// if it does not receive any pings after registering.
TEST_F(SlaveTest, PingTimeoutNoPings)
{
// Set shorter ping timeout values.
master::Flags masterFlags = CreateMasterFlags();
masterFlags.agent_ping_timeout = Seconds(5);
masterFlags.max_agent_ping_timeouts = 2u;
Duration totalTimeout =
masterFlags.agent_ping_timeout * masterFlags.max_agent_ping_timeouts;
// Start a master.
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
// Block all pings to the slave.
DROP_PROTOBUFS(PingSlaveMessage(), _, _);
Future<SlaveRegisteredMessage> slaveRegisteredMessage =
FUTURE_PROTOBUF(SlaveRegisteredMessage(), _, _);
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a slave.
slave::Flags agentFlags = CreateSlaveFlags();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), agentFlags);
ASSERT_SOME(slave);
AWAIT_READY(slaveRegisteredMessage);
ASSERT_TRUE(slaveRegisteredMessage->has_connection());
MasterSlaveConnection connection = slaveRegisteredMessage->connection();
EXPECT_EQ(
totalTimeout,
Seconds(static_cast<int64_t>(connection.total_ping_timeout_seconds())));
// Ensure the slave processes the registration message and schedules
// the ping timeout, before we advance the clock.
Clock::pause();
Clock::settle();
// Advance to the ping timeout to trigger a re-detection and
// re-registration.
Future<Nothing> detected = FUTURE_DISPATCH(_, &Slave::detected);
Future<SlaveReregisteredMessage> slaveReregisteredMessage =
FUTURE_PROTOBUF(SlaveReregisteredMessage(), _, _);
Clock::advance(totalTimeout);
AWAIT_READY(detected);
Clock::advance(agentFlags.registration_backoff_factor);
AWAIT_READY(slaveReregisteredMessage);
}
// This test ensures that the slave will reregister with the master
// if it stops receiving pings.
TEST_F(SlaveTest, PingTimeoutSomePings)
{
// Start a master.
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
Future<SlaveRegisteredMessage> slaveRegisteredMessage =
FUTURE_PROTOBUF(SlaveRegisteredMessage(), _, _);
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a slave.
slave::Flags agentFlags = CreateSlaveFlags();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), agentFlags);
ASSERT_SOME(slave);
AWAIT_READY(slaveRegisteredMessage);
Clock::pause();
// Ensure a ping reaches the slave.
Future<Message> ping = FUTURE_MESSAGE(
Eq(PingSlaveMessage().GetTypeName()), _, _);
Clock::advance(masterFlags.agent_ping_timeout);
AWAIT_READY(ping);
// Now block further pings from the master and advance
// the clock to trigger a re-detection and re-registration on
// the slave.
DROP_PROTOBUFS(PingSlaveMessage(), _, _);
Future<Nothing> detected = FUTURE_DISPATCH(_, &Slave::detected);
Future<SlaveReregisteredMessage> slaveReregisteredMessage =
FUTURE_PROTOBUF(SlaveReregisteredMessage(), _, _);
Clock::advance(slave::DEFAULT_MASTER_PING_TIMEOUT());
AWAIT_READY(detected);
Clock::advance(agentFlags.registration_backoff_factor);
AWAIT_READY(slaveReregisteredMessage);
}
// This test ensures that when a slave removal rate limit is
// specified, the master only removes a slave that fails health checks
// when it is permitted to do so by the rate limiter.
TEST_F(SlaveTest, RateLimitSlaveRemoval)
{
// Start a master.
auto slaveRemovalLimiter = std::make_shared<MockRateLimiter>();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master =
StartMaster(slaveRemovalLimiter, masterFlags);
ASSERT_SOME(master);
// Set these expectations up before we spawn the slave so that we
// don't miss the first PING.
Future<Message> ping = FUTURE_MESSAGE(
Eq(PingSlaveMessage().GetTypeName()), _, _);
// Drop all the PONGs to simulate health check timeout.
DROP_PROTOBUFS(PongSlaveMessage(), _, _);
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a slave.
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
// Start a scheduler.
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<Nothing> resourceOffers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureSatisfy(&resourceOffers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
// Need to make sure the framework AND slave have registered with
// master. Waiting for resource offers should accomplish both.
AWAIT_READY(resourceOffers);
// Return a pending future from the rate limiter.
Future<Nothing> acquire;
Promise<Nothing> promise;
EXPECT_CALL(*slaveRemovalLimiter, acquire())
.WillOnce(DoAll(FutureSatisfy(&acquire),
Return(promise.future())));
EXPECT_CALL(sched, offerRescinded(&driver, _))
.WillOnce(Return()); // Expect a single offer to be rescinded.
Future<Nothing> slaveLost;
EXPECT_CALL(sched, slaveLost(&driver, _))
.WillOnce(FutureSatisfy(&slaveLost));
// Induce a health check failure of the slave.
Clock::pause();
size_t pings = 0;
while (true) {
AWAIT_READY(ping);
pings++;
if (pings == masterFlags.max_agent_ping_timeouts) {
break;
}
ping = FUTURE_MESSAGE(Eq(PingSlaveMessage().GetTypeName()), _, _);
Clock::advance(masterFlags.agent_ping_timeout);
}
Clock::advance(masterFlags.agent_ping_timeout);
// The master should attempt to acquire a permit.
AWAIT_READY(acquire);
// The slave should not be removed before the permit is satisfied;
// that means the scheduler shouldn't receive `slaveLost` yet.
Clock::settle();
ASSERT_TRUE(slaveLost.isPending());
// Once the permit is satisfied, the `slaveLost` scheduler callback
// should be invoked.
promise.set(Nothing());
AWAIT_READY(slaveLost);
driver.stop();
driver.join();
}
// This test verifies that when a slave responds to pings after the
// slave observer has scheduled it for removal (due to health check
// failure), the slave removal is cancelled.
TEST_F(SlaveTest, CancelSlaveRemoval)
{
// Start a master.
auto slaveRemovalLimiter = std::make_shared<MockRateLimiter>();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master =
StartMaster(slaveRemovalLimiter, masterFlags);
ASSERT_SOME(master);
// Set these expectations up before we spawn the slave so that we
// don't miss the first PING.
Future<Message> ping = FUTURE_MESSAGE(
Eq(PingSlaveMessage().GetTypeName()), _, _);
// Drop all the PONGs to simulate health check timeout.
DROP_PROTOBUFS(PongSlaveMessage(), _, _);
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a slave.
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
// Start a scheduler.
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<Nothing> resourceOffers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureSatisfy(&resourceOffers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
EXPECT_CALL(sched, slaveLost(&driver, _))
.Times(0); // The `slaveLost` callback should not be invoked.
driver.start();
// Need to make sure the framework AND slave have registered with
// master. Waiting for resource offers should accomplish both.
AWAIT_READY(resourceOffers);
// Return a pending future from the rate limiter.
Future<Nothing> acquire;
Promise<Nothing> promise;
EXPECT_CALL(*slaveRemovalLimiter, acquire())
.WillOnce(DoAll(FutureSatisfy(&acquire),
Return(promise.future())));
// Induce a health check failure of the slave.
Clock::pause();
size_t pings = 0;
while (true) {
AWAIT_READY(ping);
pings++;
if (pings == masterFlags.max_agent_ping_timeouts) {
break;
}
ping = FUTURE_MESSAGE(Eq(PingSlaveMessage().GetTypeName()), _, _);
Clock::advance(masterFlags.agent_ping_timeout);
}
Clock::advance(masterFlags.agent_ping_timeout);
// The master should attempt to acquire a permit.
AWAIT_READY(acquire);
// Settle to make sure the slave removal does not occur.
Clock::settle();
// Reset the filters to allow pongs from the slave.
filter(nullptr);
// Advance clock enough to do a ping pong.
Clock::advance(masterFlags.agent_ping_timeout);
Clock::settle();
// The master should have tried to cancel the removal.
EXPECT_TRUE(promise.future().hasDiscard());
// Allow the cancellation and settle the clock to ensure the
// `slaveLost` scheduler callback is not invoked.
promise.discard();
Clock::settle();
}
#ifndef __WINDOWS__
// This test checks that the master behaves correctly when a slave
// fails health checks, but concurrently the slave unregisters from
// the master.
TEST_F(SlaveTest, HealthCheckUnregisterRace)
{
// Start a master.
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
// Start a slave.
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
// Start a scheduler.
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
// Need to make sure the framework AND slave have registered with
// master. Waiting for resource offers should accomplish both.
AWAIT_READY(offers);
SlaveID slaveId = offers.get()[0].slave_id();
// Expect a single offer to be rescinded.
EXPECT_CALL(sched, offerRescinded(&driver, _));
Future<Nothing> slaveLost;
EXPECT_CALL(sched, slaveLost(&driver, _))
.WillOnce(FutureSatisfy(&slaveLost));
// Cause the slave to shutdown gracefully. This should result in
// the slave sending `UnregisterSlaveMessage` to the master.
Future<UnregisterSlaveMessage> unregisterSlaveMessage =
FUTURE_PROTOBUF(
UnregisterSlaveMessage(),
slave.get()->pid,
master.get()->pid);
slave.get()->shutdown();
slave->reset();
AWAIT_READY(unregisterSlaveMessage);
AWAIT_READY(slaveLost);
Clock::pause();
Clock::settle();
// We now want to arrange for the agent to fail health checks. We
// can't do that directly, because the `SlaveObserver` for this
// agent has already been removed. Instead, we dispatch to the
// master's `markUnreachable` method directly. We expect the master
// to ignore this message; in particular, the master should not
// attempt to update the registry to mark the slave unreachable.
EXPECT_CALL(*master.get()->registrar, apply(_))
.Times(0);
SlaveInfo slaveInfo;
slaveInfo.mutable_id()->CopyFrom(slaveId);
slaveInfo.set_hostname("hostname");
process::dispatch(master.get()->pid,
&Master::markUnreachable,
slaveInfo,
false,
"dummy test case dispatch");
Clock::settle();
Clock::resume();
driver.stop();
driver.join();
}
#endif // __WINDOWS__
// This test verifies that when an unreachable agent reregisters after
// master failover, the master consults and updates the registrar for
// re-admitting the agent.
//
// TODO(andschwa): Enable when Windows supports replicated log. See MESOS-5932.
TEST_F_TEMP_DISABLED_ON_WINDOWS(
SlaveTest, UnreachableAgentReregisterAfterFailover)
{
master::Flags masterFlags = CreateMasterFlags();
masterFlags.registry = "replicated_log";
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
Future<SlaveRegisteredMessage> slaveRegisteredMessage =
FUTURE_PROTOBUF(SlaveRegisteredMessage(), master.get()->pid, _);
// Reuse slaveFlags so both StartSlave() use the same work_dir.
slave::Flags slaveFlags = CreateSlaveFlags();
// Drop all the PONGs to simulate slave partition.
DROP_PROTOBUFS(PongSlaveMessage(), _, _);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), slaveFlags);
ASSERT_SOME(slave);
AWAIT_READY(slaveRegisteredMessage);
slave.get()->terminate();
slave->reset();
Clock::pause();
// Settle here to make sure the `SlaveObserver` has already started counting
// the `slavePingTimeout` before we advance the clock for the first time.
Clock::settle();
// Induce agent ping timeouts.
size_t pings = 0;
while (true) {
pings++;
if (pings == masterFlags.max_agent_ping_timeouts) {
break;
}
Clock::advance(masterFlags.agent_ping_timeout);
Clock::settle();
}
// Now set the expectation when the agent is one ping timeout away
// from being deemed unreachable.
Future<Owned<master::RegistryOperation>> markUnreachable;
EXPECT_CALL(*master.get()->registrar, apply(_))
.WillOnce(DoAll(FutureArg<0>(&markUnreachable),
Invoke(master.get()->registrar.get(),
&MockRegistrar::unmocked_apply)));
Clock::advance(masterFlags.agent_ping_timeout);
AWAIT_READY(markUnreachable);
EXPECT_NE(
nullptr,
dynamic_cast<master::MarkSlaveUnreachable*>(markUnreachable->get()));
// Make sure the registrar operation completes so the agent will be updated
// as an unreachable agent in the registry before the master terminates.
Clock::settle();
master->reset();
master = StartMaster(masterFlags);
ASSERT_SOME(master);
// Start the agent, which will cause it to reregister. Intercept the
// next registry operation, which we expect to be slave reregistration.
Future<SlaveReregisteredMessage> slaveReregisteredMessage =
FUTURE_PROTOBUF(SlaveReregisteredMessage(), master.get()->pid, _);
Future<Owned<master::RegistryOperation>> markReachable;
EXPECT_CALL(*master.get()->registrar, apply(_))
.WillOnce(DoAll(FutureArg<0>(&markReachable),
Invoke(master.get()->registrar.get(),
&MockRegistrar::unmocked_apply)));
detector = master.get()->createDetector();
slave = StartSlave(detector.get(), slaveFlags);
ASSERT_SOME(slave);
Clock::advance(slaveFlags.registration_backoff_factor);
// Verify that the reregistration involves registry update.
AWAIT_READY(markReachable);
EXPECT_NE(
nullptr,
dynamic_cast<master::MarkSlaveReachable*>(markReachable->get()));
AWAIT_READY(slaveReregisteredMessage);
}
// This test verifies that when a registered agent restarts and reregisters
// after master failover, the master does not consult the registrar in
// deciding to re-admit the agent.
//
// TODO(andschwa): Enable when Windows supports replicated log. See MESOS-5932.
TEST_F_TEMP_DISABLED_ON_WINDOWS(
SlaveTest, RegisteredAgentReregisterAfterFailover)
{
// Pause the clock to avoid registration retries and the agent
// being deemed unreachable.
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
masterFlags.registry = "replicated_log";
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
Future<SlaveRegisteredMessage> slaveRegisteredMessage =
FUTURE_PROTOBUF(SlaveRegisteredMessage(), master.get()->pid, _);
// Reuse slaveFlags so both StartSlave() use the same work_dir.
slave::Flags slaveFlags = CreateSlaveFlags();
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), slaveFlags);
ASSERT_SOME(slave);
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(slaveRegisteredMessage);
// There should be no registrar operation across both agent termination
// and reregistration.
EXPECT_CALL(*master.get()->registrar, apply(_))
.Times(0);
slave.get()->terminate();
slave->reset();
master->reset();
master = StartMaster(masterFlags);
ASSERT_SOME(master);
Future<SlaveReregisteredMessage> slaveReregisteredMessage =
FUTURE_PROTOBUF(SlaveReregisteredMessage(), master.get()->pid, _);
detector = master.get()->createDetector();
slave = StartSlave(detector.get(), slaveFlags);
ASSERT_SOME(slave);
Clock::advance(slaveFlags.registration_backoff_factor);
// No registrar operation occurs by the time the agent is fully registered.
AWAIT_READY(slaveReregisteredMessage);
}
#ifndef __WINDOWS__
// This test checks that the master behaves correctly when a slave
// fails health checks and is in the process of being marked
// unreachable in the registry, but concurrently the slave unregisters
// from the master.
TEST_F(SlaveTest, UnreachableThenUnregisterRace)
{
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
// Set these expectations up before we spawn the slave so that we
// don't miss the first PING.
Future<Message> ping = FUTURE_MESSAGE(
Eq(PingSlaveMessage().GetTypeName()), _, _);
// Drop all the PONGs to simulate slave partition.
DROP_PROTOBUFS(PongSlaveMessage(), _, _);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<Nothing> resourceOffers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureSatisfy(&resourceOffers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
// Need to make sure the framework AND slave have registered with
// master. Waiting for resource offers should accomplish both.
AWAIT_READY(resourceOffers);
Clock::pause();
EXPECT_CALL(sched, offerRescinded(&driver, _))
.Times(AtMost(1));
Future<Nothing> slaveLost;
EXPECT_CALL(sched, slaveLost(&driver, _))
.WillOnce(FutureSatisfy(&slaveLost));
// Now advance through the PINGs.
size_t pings = 0;
while (true) {
AWAIT_READY(ping);
pings++;
if (pings == masterFlags.max_agent_ping_timeouts) {
break;
}
ping = FUTURE_MESSAGE(Eq(PingSlaveMessage().GetTypeName()), _, _);
Clock::advance(masterFlags.agent_ping_timeout);
}
// Intercept the next registry operation. This operation should be
// attempting to mark the slave unreachable.
Future<Owned<master::RegistryOperation>> markUnreachable;
Promise<bool> markUnreachableContinue;
EXPECT_CALL(*master.get()->registrar, apply(_))
.WillOnce(DoAll(FutureArg<0>(&markUnreachable),
Return(markUnreachableContinue.future())));
Clock::advance(masterFlags.agent_ping_timeout);
AWAIT_READY(markUnreachable);
EXPECT_NE(
nullptr,
dynamic_cast<master::MarkSlaveUnreachable*>(
markUnreachable->get()));
// Cause the slave to shutdown gracefully. This should result in
// the slave sending `UnregisterSlaveMessage` to the master.
// Normally, the master would then remove the slave from the
// registry, but since the slave is already being marked
// unreachable, the master should ignore the unregister message.
Future<UnregisterSlaveMessage> unregisterSlaveMessage =
FUTURE_PROTOBUF(
UnregisterSlaveMessage(),
slave.get()->pid,
master.get()->pid);
EXPECT_CALL(*master.get()->registrar, apply(_))
.Times(0);
slave.get()->shutdown();
slave->reset();
AWAIT_READY(unregisterSlaveMessage);
// Apply the registry operation to mark the slave unreachable, then
// pass the result back to the master to allow it to continue.
Future<bool> applyUnreachable =
master.get()->registrar->unmocked_apply(markUnreachable.get());
AWAIT_READY(applyUnreachable);
markUnreachableContinue.set(applyUnreachable.get());
AWAIT_READY(slaveLost);
Clock::resume();
driver.stop();
driver.join();
}
#endif // __WINDOWS__
// This test checks that the master behaves correctly when a slave is
// in the process of unregistering from the master when it is marked
// unreachable.
TEST_F(SlaveTest, UnregisterThenUnreachableRace)
{
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
// Set these expectations up before we spawn the slave so that we
// don't miss the first PING.
Future<Message> ping = FUTURE_MESSAGE(
Eq(PingSlaveMessage().GetTypeName()), _, _);
// Drop all the PONGs to simulate slave partition.
DROP_PROTOBUFS(PongSlaveMessage(), _, _);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> resourceOffers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&resourceOffers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
// Need to make sure the framework AND slave have registered with
// master. Waiting for resource offers should accomplish both.
AWAIT_READY(resourceOffers);
ASSERT_EQ(1u, resourceOffers->size());
SlaveID slaveId = resourceOffers.get()[0].slave_id();
Clock::pause();
// Simulate the slave shutting down gracefully. This might happen
// normally if the slave receives SIGUSR1. However, we don't use
// that approach here, because that would also result in an `exited`
// event at the master; we want to test the case where the slave
// begins to shutdown but the socket hasn't been closed yet. Hence,
// we spoof the `UnregisterSlaveMessage`.
//
// When the master receives the `UnregisterSlaveMessage`, it should
// attempt to remove the slave from the registry.
Future<Owned<master::RegistryOperation>> removeSlave;
Promise<bool> removeSlaveContinue;
EXPECT_CALL(*master.get()->registrar, apply(_))
.WillOnce(DoAll(FutureArg<0>(&removeSlave),
Return(removeSlaveContinue.future())));
process::dispatch(master.get()->pid,
&Master::unregisterSlave,
slave.get()->pid,
slaveId);
AWAIT_READY(removeSlave);
EXPECT_NE(
nullptr,
dynamic_cast<master::RemoveSlave*>(removeSlave->get()));
// Next, cause the slave to fail health checks; master will attempt
// to mark it unreachable.
size_t pings = 0;
while (true) {
AWAIT_READY(ping);
pings++;
if (pings == masterFlags.max_agent_ping_timeouts) {
break;
}
ping = FUTURE_MESSAGE(Eq(PingSlaveMessage().GetTypeName()), _, _);
Clock::advance(masterFlags.agent_ping_timeout);
}
// We expect the `SlaveObserver` to dispatch a message to the master
// to mark the slave unreachable. The master should ignore this
// request because the slave is already being removed.
Future<Nothing> unreachableDispatch =
FUTURE_DISPATCH(master.get()->pid, &Master::markUnreachable);
EXPECT_CALL(*master.get()->registrar, apply(_))
.Times(0);
Clock::advance(masterFlags.agent_ping_timeout);
AWAIT_READY(unreachableDispatch);
EXPECT_CALL(sched, offerRescinded(&driver, _))
.Times(AtMost(1));
Future<Nothing> slaveLost;
EXPECT_CALL(sched, slaveLost(&driver, _))
.WillOnce(FutureSatisfy(&slaveLost));
// Apply the registry operation to remove the slave, then pass the
// result back to the master to allow it to continue.
Future<bool> applyRemove =
master.get()->registrar->unmocked_apply(removeSlave.get());
AWAIT_READY(applyRemove);
removeSlaveContinue.set(applyRemove.get());
AWAIT_READY(slaveLost);
Clock::resume();
driver.stop();
driver.join();
}
// This test ensures that a killTask() can happen between runTask()
// and _run() and then gets "handled properly". This means that
// the task never gets started, but also does not get lost. The end
// result is status TASK_KILLED. Essentially, killing the task is
// realized while preparing to start it. See MESOS-947. This test
// removes the framework and proves that removeFramework() is
// called. See MESOS-1945.
TEST_F(SlaveTest, KillTaskBetweenRunTaskParts)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
StandaloneMasterDetector detector(master.get()->pid);
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
None(),
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers.get()[0].slave_id());
task.mutable_resources()->MergeFrom(offers.get()[0].resources());
task.mutable_executor()->MergeFrom(DEFAULT_EXECUTOR_INFO);
EXPECT_CALL(exec, registered(_, _, _, _))
.Times(0);
EXPECT_CALL(exec, launchTask(_, _))
.Times(0);
EXPECT_CALL(exec, shutdown(_))
.Times(0);
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillRepeatedly(FutureArg<1>(&status));
EXPECT_CALL(*slave.get()->mock(), runTask(_, _, _, _, _, _, _))
.WillOnce(Invoke(slave.get()->mock(), &MockSlave::unmocked_runTask));
// Saved arguments from Slave::_run().
FrameworkInfo frameworkInfo;
ExecutorInfo executorInfo;
Option<TaskGroupInfo> taskGroup;
Option<TaskInfo> task_;
vector<ResourceVersionUUID> resourceVersionUuids;
Option<bool> launchExecutor;
// Skip what Slave::_run() normally does, save its arguments for
// later, return a pending future to pause the original continuation,
// so that we can control when the task is killed.
Promise<Nothing> promise;
Future<Nothing> _run;
EXPECT_CALL(*slave.get()->mock(), _run(_, _, _, _, _, _))
.WillOnce(DoAll(FutureSatisfy(&_run),
SaveArg<0>(&frameworkInfo),
SaveArg<1>(&executorInfo),
SaveArg<2>(&task_),
SaveArg<3>(&taskGroup),
SaveArg<4>(&resourceVersionUuids),
SaveArg<5>(&launchExecutor),
Return(promise.future())));
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(_run);
Future<Nothing> killTask;
EXPECT_CALL(*slave.get()->mock(), killTask(_, _))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_killTask),
FutureSatisfy(&killTask)));
Future<Nothing> removeFramework;
EXPECT_CALL(*slave.get()->mock(), removeFramework(_))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_removeFramework),
FutureSatisfy(&removeFramework)));
EXPECT_CALL(sched, executorLost(&driver, DEFAULT_EXECUTOR_ID, _, _))
.Times(AtMost(1));
driver.killTask(task.task_id());
AWAIT_READY(killTask);
// The agent will remove the framework when killing this task
// since there remain no more tasks.
AWAIT_READY(removeFramework);
Future<Nothing> unmocked__run = process::dispatch(slave.get()->pid, [=] {
return slave.get()->mock()->unmocked__run(
frameworkInfo,
executorInfo,
task_,
taskGroup,
resourceVersionUuids,
launchExecutor);
});
// Resume the original continuation once `unmocked__run` is complete.
promise.associate(unmocked__run);
AWAIT_READY(status);
EXPECT_EQ(TASK_KILLED, status->state());
AWAIT(unmocked__run);
driver.stop();
driver.join();
}
// This test ensures was added due to MESOS-7863, where the
// agent previously dropped TASK_KILLED in the cases outlined
// in the issue.
TEST_F(SlaveTest, KillMultiplePendingTasks)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
StandaloneMasterDetector detector(master.get()->pid);
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
None(),
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
// We only pause the clock after receiving the offer since the
// agent uses a delay to reregister.
//
// TODO(bmahler): Remove the initial agent delay within the tests.
Clock::pause();
Resources taskResources = Resources::parse("cpus:0.1;mem:32;disk:32").get();
TaskInfo task1 = createTask(
offers->at(0).slave_id(), taskResources, "echo hi");
TaskInfo task2 = createTask(
offers->at(0).slave_id(), taskResources, "echo hi");
EXPECT_CALL(containerizer, launch(_, _, _, _))
.Times(0);
Future<TaskStatus> status1, status2;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status1))
.WillOnce(FutureArg<1>(&status2));
EXPECT_CALL(*slave.get()->mock(), runTask(_, _, _, _, _, _, _))
.WillOnce(Invoke(slave.get()->mock(), &MockSlave::unmocked_runTask))
.WillOnce(Invoke(slave.get()->mock(), &MockSlave::unmocked_runTask));
// Skip what Slave::_run() normally does, save its arguments for
// later, return a pending future to pause the original continuation,
// so that we can control when the task is killed.
FrameworkInfo frameworkInfo1, frameworkInfo2;
ExecutorInfo executorInfo1, executorInfo2;
Option<TaskGroupInfo> taskGroup1, taskGroup2;
Option<TaskInfo> task_1, task_2;
vector<ResourceVersionUUID> resourceVersionUuids1, resourceVersionUuids2;
Option<bool> launchExecutor1, launchExecutor2;
Promise<Nothing> promise1, promise2;
Future<Nothing> _run1, _run2;
EXPECT_CALL(*slave.get()->mock(), _run(_, _, _, _, _, _))
.WillOnce(DoAll(FutureSatisfy(&_run1),
SaveArg<0>(&frameworkInfo1),
SaveArg<1>(&executorInfo1),
SaveArg<2>(&task_1),
SaveArg<3>(&taskGroup1),
SaveArg<4>(&resourceVersionUuids1),
SaveArg<5>(&launchExecutor1),
Return(promise1.future())))
.WillOnce(DoAll(FutureSatisfy(&_run2),
SaveArg<0>(&frameworkInfo2),
SaveArg<1>(&executorInfo2),
SaveArg<2>(&task_2),
SaveArg<3>(&taskGroup2),
SaveArg<4>(&resourceVersionUuids2),
SaveArg<5>(&launchExecutor2),
Return(promise2.future())));
driver.launchTasks(offers.get()[0].id(), {task1, task2});
AWAIT_READY(process::await(_run1, _run2));
Future<Nothing> killTask1, killTask2;
EXPECT_CALL(*slave.get()->mock(), killTask(_, _))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_killTask),
FutureSatisfy(&killTask1)))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_killTask),
FutureSatisfy(&killTask2)));
Future<Nothing> removeFramework;
EXPECT_CALL(*slave.get()->mock(), removeFramework(_))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_removeFramework),
FutureSatisfy(&removeFramework)));
driver.killTask(task1.task_id());
driver.killTask(task2.task_id());
AWAIT_READY(process::await(killTask1, killTask2));
// We expect the tasks to be killed and framework removed.
AWAIT_READY(status1);
EXPECT_EQ(TASK_KILLED, status1->state());
AWAIT_READY(status2);
EXPECT_EQ(TASK_KILLED, status2->state());
AWAIT_READY(removeFramework);
// The `__run` continuations should have no effect.
Future<Nothing> unmocked__run1 = process::dispatch(slave.get()->pid, [=] {
return slave.get()->mock()->unmocked__run(
frameworkInfo1,
executorInfo1,
task_1,
taskGroup1,
resourceVersionUuids1,
launchExecutor1);
});
Future<Nothing> unmocked__run2 = process::dispatch(slave.get()->pid, [=] {
return slave.get()->mock()->unmocked__run(
frameworkInfo2,
executorInfo2,
task_2,
taskGroup2,
resourceVersionUuids2,
launchExecutor2);
});
// Resume the original continuation once unmocked__run is complete.
promise1.associate(unmocked__run1);
promise2.associate(unmocked__run2);
Clock::settle();
driver.stop();
driver.join();
}
// This test verifies that when the agent gets a `killTask`
// message for a queued task on a registering executor, a
// the agent will generate a TASK_KILLED and will shut down
// the executor.
TEST_F(SlaveTest, KillQueuedTaskDuringExecutorRegistration)
{
Clock::pause();
// Start a master.
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a slave.
slave::Flags slaveFlags = CreateSlaveFlags();
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &containerizer, slaveFlags);
ASSERT_SOME(slave);
Clock::advance(slaveFlags.authentication_backoff_factor);
Clock::settle();
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers.get()[0].slave_id());
task.mutable_resources()->MergeFrom(offers.get()[0].resources());
task.mutable_executor()->MergeFrom(DEFAULT_EXECUTOR_INFO);
EXPECT_CALL(exec, registered(_, _, _, _))
.Times(0);
EXPECT_CALL(exec, launchTask(_, _))
.Times(0);
EXPECT_CALL(exec, shutdown(_));
// Hold on to the executor registration message so that the task stays
// queued on the agent.
Future<Message> registerExecutorMessage =
DROP_MESSAGE(Eq(RegisterExecutorMessage().GetTypeName()), _, _);
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(registerExecutorMessage);
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
Future<Nothing> executorLost;
EXPECT_CALL(sched, executorLost(&driver, DEFAULT_EXECUTOR_ID, _, _))
.WillOnce(FutureSatisfy(&executorLost));
// Kill the task enqueued on the agent.
driver.killTask(task.task_id());
AWAIT_READY(status);
EXPECT_EQ(TASK_KILLED, status->state());
EXPECT_EQ(TaskStatus::REASON_TASK_KILLED_DURING_LAUNCH, status->reason());
// Now let the executor register by spoofing the message.
RegisterExecutorMessage registerExecutor;
registerExecutor.ParseFromString(registerExecutorMessage->body);
process::post(registerExecutorMessage->from,
slave.get()->pid,
registerExecutor);
Clock::advance(slaveFlags.executor_shutdown_grace_period);
AWAIT_READY(executorLost);
driver.stop();
driver.join();
}
// This test ensures that if a `killTask()` for an HTTP based executor is
// received by the agent before the executor registers, the executor is
// properly cleaned up.
TEST_F(SlaveTest, KillTaskUnregisteredHTTPExecutor)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
auto executor = std::make_shared<v1::MockHTTPExecutor>();
Resources resources =
Resources::parse("cpus:0.1;mem:32;disk:32").get();
ExecutorInfo executorInfo;
executorInfo.set_type(ExecutorInfo::DEFAULT);
executorInfo.mutable_executor_id()->CopyFrom(DEFAULT_EXECUTOR_ID);
executorInfo.mutable_resources()->CopyFrom(resources);
const ExecutorID& executorId = executorInfo.executor_id();
TestContainerizer containerizer(executorId, executor);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers));
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(v1::DEFAULT_FRAMEWORK_INFO);
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(devolve(frameworkId));
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const SlaveID slaveId = devolve(offer.agent_id());
Future<v1::executor::Mesos*> executorLib;
EXPECT_CALL(*executor, connected(_))
.WillOnce(FutureArg<0>(&executorLib));
v1::TaskInfo task1 =
evolve(createTask(slaveId, resources, ""));
v1::TaskInfo task2 =
evolve(createTask(slaveId, resources, ""));
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(task1);
taskGroup.add_tasks()->CopyFrom(task2);
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACCEPT);
Call::Accept* accept = call.mutable_accept();
accept->add_offer_ids()->CopyFrom(offers->offers(0).id());
v1::Offer::Operation* operation = accept->add_operations();
operation->set_type(v1::Offer::Operation::LAUNCH_GROUP);
v1::Offer::Operation::LaunchGroup* launchGroup =
operation->mutable_launch_group();
launchGroup->mutable_executor()->CopyFrom(evolve(executorInfo));
launchGroup->mutable_task_group()->CopyFrom(taskGroup);
mesos.send(call);
}
// Wait for the executor to be launched and then kill the task before
// the executor subscribes with the agent.
AWAIT_READY(executorLib);
Future<v1::scheduler::Event::Update> update1;
Future<v1::scheduler::Event::Update> update2;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(FutureArg<1>(&update1))
.WillOnce(FutureArg<1>(&update2));
mesos.send(
v1::createCallKill(frameworkId, task1.task_id(), offer.agent_id()));
AWAIT_READY(update1);
AWAIT_READY(update2);
ASSERT_EQ(v1::TASK_KILLED, update1->status().state());
ASSERT_EQ(v1::TASK_KILLED, update2->status().state());
Future<Nothing> shutdown;
EXPECT_CALL(*executor, shutdown(_))
.WillOnce(FutureSatisfy(&shutdown));
// The executor should receive the shutdown event upon subscribing
// with the agent.
{
v1::executor::Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.mutable_executor_id()->CopyFrom(evolve(executorId));
call.set_type(v1::executor::Call::SUBSCRIBE);
call.mutable_subscribe();
executorLib.get()->send(call);
}
AWAIT_READY(shutdown);
}
// This test ensures that the agent sends an `ExitedExecutorMessage` when the
// executor is never launched, so that the master's executor bookkeeping entry
// is removed. See MESOS-1720.
TEST_F(SlaveTest, RemoveExecutorUponFailedLaunch)
{
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
slave::Flags slaveFlags = CreateSlaveFlags();
slaveFlags.resources = "cpus:2;mem:512;disk:512;ports:[]";
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a mock slave.
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &containerizer, slaveFlags, true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
Resources executorResources = Resources::parse("cpus:0.1;mem:32").get();
executorResources.allocate("*");
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers.get()[0].slave_id());
task.mutable_resources()->MergeFrom(
Resources(offers.get()[0].resources()) - executorResources);
task.mutable_executor()->MergeFrom(DEFAULT_EXECUTOR_INFO);
task.mutable_executor()->mutable_resources()->CopyFrom(executorResources);
EXPECT_CALL(exec, registered(_, _, _, _))
.Times(0);
EXPECT_CALL(sched, statusUpdate(&driver, _))
.Times(AtMost(1));
Future<ExitedExecutorMessage> exitedExecutorMessage =
FUTURE_PROTOBUF(ExitedExecutorMessage(), _, _);
// Saved arguments from `Slave::_run()`.
FrameworkInfo frameworkInfo;
ExecutorInfo executorInfo_;
Option<TaskGroupInfo> taskGroup_;
Option<TaskInfo> task_;
vector<ResourceVersionUUID> resourceVersionUuids;
Option<bool> launchExecutor;
// Before launching the executor in `__run`, we pause the continuation
// by returning a pending future. We then kill the task and re-dispatch
// `_run`. We use its return result to fulfill the pending future and
// resume the continuation.
Promise<Nothing> promise;
Future<Nothing> _run;
EXPECT_CALL(*slave.get()->mock(), _run(_, _, _, _, _, _))
.WillOnce(DoAll(FutureSatisfy(&_run),
SaveArg<0>(&frameworkInfo),
SaveArg<1>(&executorInfo_),
SaveArg<2>(&task_),
SaveArg<3>(&taskGroup_),
SaveArg<4>(&resourceVersionUuids),
SaveArg<5>(&launchExecutor),
Return(promise.future())));
Future<Nothing> executorLost;
EXPECT_CALL(sched, executorLost(&driver, DEFAULT_EXECUTOR_ID, _, _))
.WillOnce(FutureSatisfy(&executorLost));
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(_run);
Future<Nothing> killTask;
EXPECT_CALL(*slave.get()->mock(), killTask(_, _))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_killTask),
FutureSatisfy(&killTask)));
driver.killTask(task.task_id());
AWAIT_READY(killTask);
Future<Nothing> unmocked__run =
process::dispatch(slave.get()->pid, [=]() -> Future<Nothing> {
return slave.get()->mock()->unmocked__run(
frameworkInfo,
executorInfo_,
task_,
taskGroup_,
resourceVersionUuids,
launchExecutor);
});
promise.associate(unmocked__run);
AWAIT(unmocked__run);
// Agent needs to send `ExitedExecutorMessage` to the master because
// the executor never launched.
AWAIT_READY(exitedExecutorMessage);
AWAIT_READY(executorLost);
// Helper function to post a request to '/api/v1' master endpoint
// and return the response.
auto post = [](
const process::PID<master::Master>& pid,
const v1::master::Call& call,
const ContentType& contentType)
{
process::http::Headers headers = createBasicAuthHeaders(DEFAULT_CREDENTIAL);
headers["Accept"] = stringify(contentType);
return process::http::post(
pid,
"api/v1",
headers,
serialize(contentType, call),
stringify(contentType));
};
v1::master::Call v1Call;
v1Call.set_type(v1::master::Call::GET_EXECUTORS);
Future<process::http::Response> response =
post(master.get()->pid, v1Call, ContentType::PROTOBUF);
response.await();
ASSERT_EQ(response->status, process::http::OK().status);
Future<v1::master::Response> v1Response =
deserialize<v1::master::Response>(ContentType::PROTOBUF, response->body);
// Master has no executor entry because the executor never launched.
ASSERT_TRUE(v1Response->IsInitialized());
ASSERT_EQ(v1::master::Response::GET_EXECUTORS, v1Response->type());
ASSERT_EQ(0, v1Response->get_executors().executors_size());
driver.stop();
driver.join();
}
// This test ensures that agent sends ExitedExecutorMessage when the task group
// fails to launch due to unschedule GC failure and that master's executor
// bookkeeping entry is removed.
TEST_F(SlaveTest, RemoveExecutorUponFailedTaskGroupLaunch)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
Owned<MasterDetector> detector = master.get()->createDetector();
slave::Flags slaveFlags = CreateSlaveFlags();
MockGarbageCollector mockGarbageCollector(slaveFlags.work_dir);
// Start a mock slave.
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &mockGarbageCollector, slaveFlags, true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::ExecutorInfo executorInfo = v1::DEFAULT_EXECUTOR_INFO;
executorInfo.clear_command();
executorInfo.set_type(v1::ExecutorInfo::DEFAULT);
executorInfo.mutable_resources()->CopyFrom(resources);
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
mesos.send(v1::createCallSubscribe(v1::DEFAULT_FRAMEWORK_INFO));
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(frameworkId);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
v1::TaskInfo task1 = v1::createTask(agentId, resources, "sleep 1000");
v1::TaskInfo task2 = v1::createTask(agentId, resources, "sleep 1000");
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(task1);
taskGroup.add_tasks()->CopyFrom(task2);
v1::Offer::Operation launchGroup = v1::LAUNCH_GROUP(executorInfo, taskGroup);
// The `unschedule()` function is used to prevent premature garbage
// collection when the executor directory already exists due to a
// previously-launched task. Simulate this scenario by creating the
// executor directory manually.
string path = paths::getExecutorPath(
slaveFlags.work_dir,
devolve(agentId),
devolve(frameworkId),
devolve(executorInfo.executor_id()));
Try<Nothing> mkdir = os::mkdir(path, true);
CHECK_SOME(mkdir);
// Induce agent unschedule GC failure. This will result in
// task launch failure before the executor launch.
EXPECT_CALL(mockGarbageCollector, unschedule(_))
.WillRepeatedly(Return(Failure("")));
Future<v1::scheduler::Event::Update> update1;
Future<v1::scheduler::Event::Update> update2;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(FutureArg<1>(&update1))
.WillOnce(FutureArg<1>(&update2));
Future<ExitedExecutorMessage> exitedExecutorMessage =
FUTURE_PROTOBUF(ExitedExecutorMessage(), _, _);
EXPECT_CALL(*scheduler, failure(_, _))
.Times(AtMost(1));
mesos.send(v1::createCallAccept(frameworkId, offer, {launchGroup}));
AWAIT_READY(exitedExecutorMessage);
AWAIT_READY(update1);
AWAIT_READY(update2);
ASSERT_EQ(v1::TASK_LOST, update1->status().state());
ASSERT_EQ(v1::TASK_LOST, update2->status().state());
// Helper function to post a request to '/api/v1' master endpoint
// and return the response.
auto post = [](
const process::PID<master::Master>& pid,
const v1::master::Call& call,
const ContentType& contentType)
{
process::http::Headers headers = createBasicAuthHeaders(DEFAULT_CREDENTIAL);
headers["Accept"] = stringify(contentType);
return process::http::post(
pid,
"api/v1",
headers,
serialize(contentType, call),
stringify(contentType));
};
v1::master::Call v1Call;
v1Call.set_type(v1::master::Call::GET_EXECUTORS);
Future<process::http::Response> response =
post(master.get()->pid, v1Call, ContentType::PROTOBUF);
response.await();
ASSERT_EQ(response->status, process::http::OK().status);
Future<v1::master::Response> v1Response =
deserialize<v1::master::Response>(ContentType::PROTOBUF, response->body);
// Master has no executor entry because the executor never launched.
ASSERT_TRUE(v1Response->IsInitialized());
ASSERT_EQ(v1::master::Response::GET_EXECUTORS, v1Response->type());
ASSERT_EQ(0, v1Response->get_executors().executors_size());
}
// This test ensures that tasks using the same executor are successfully
// launched in the order in which the agent receives the RunTask(Group)Message,
// even when we manually reorder the completion of the asynchronous unschedule
// GC step. See MESOS-8624.
TEST_F(SlaveTest, LaunchTasksReorderUnscheduleGC)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
slave::Flags slaveFlags = CreateSlaveFlags();
MockGarbageCollector mockGarbageCollector(slaveFlags.work_dir);
// Start a mock slave.
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &mockGarbageCollector, slaveFlags, true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(v1::scheduler::SendSubscribe(v1::DEFAULT_FRAMEWORK_INFO));
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
v1::scheduler::TestMesos mesos(
master.get()->pid, ContentType::PROTOBUF, scheduler);
// Advance the clock to trigger both agent registration and a batch
// allocation.
Clock::advance(slaveFlags.registration_backoff_factor);
Clock::advance(masterFlags.allocation_interval);
AWAIT_READY(subscribed);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::FrameworkID frameworkId(subscribed->framework_id());
v1::ExecutorInfo executorInfo = v1::createExecutorInfo(
"default", None(), resources, v1::ExecutorInfo::DEFAULT, frameworkId);
// Create two separate task groups that use the same executor.
v1::TaskInfo taskInfo1 = v1::createTask(agentId, resources, "sleep 1000");
v1::TaskGroupInfo taskGroup1 = v1::createTaskGroupInfo({taskInfo1});
v1::TaskInfo taskInfo2 = v1::createTask(agentId, resources, "sleep 1000");
v1::TaskGroupInfo taskGroup2 = v1::createTaskGroupInfo({taskInfo2});
v1::Offer::Operation launchGroup1 =
v1::LAUNCH_GROUP(executorInfo, taskGroup1);
v1::Offer::Operation launchGroup2 =
v1::LAUNCH_GROUP(executorInfo, taskGroup2);
// The `unschedule()` function is used to prevent premature garbage
// collection when the executor directory already exists due to a
// previously-launched task. Simulate this scenario by creating the
// executor directory manually.
string path = paths::getExecutorPath(
slaveFlags.work_dir,
devolve(agentId),
devolve(frameworkId),
devolve(executorInfo.executor_id()));
Try<Nothing> mkdir = os::mkdir(path, true);
CHECK_SOME(mkdir);
Promise<bool> promise1;
// Catch the unschedule GC step and reorder the task group launches by
// pausing the processing of `taskGroup1` while allowing the processing
// of `taskGroup1` to continue.
EXPECT_CALL(mockGarbageCollector, unschedule(StrEq(path)))
.WillOnce(Return(promise1.future()))
.WillRepeatedly(Return(true));
Future<v1::scheduler::Event::Update> taskStarting1, taskStarting2;
Future<v1::scheduler::Event::Update> taskRunning1, taskRunning2;
EXPECT_CALL(
*scheduler, update(_, TaskStatusUpdateTaskIdEq(taskInfo1.task_id())))
.WillOnce(DoAll(
FutureArg<1>(&taskStarting1),
v1::scheduler::SendAcknowledge(frameworkId, agentId)))
.WillOnce(DoAll(
FutureArg<1>(&taskRunning1),
v1::scheduler::SendAcknowledge(frameworkId, agentId)));
EXPECT_CALL(
*scheduler, update(_, TaskStatusUpdateTaskIdEq(taskInfo2.task_id())))
.WillOnce(DoAll(
FutureArg<1>(&taskStarting2),
v1::scheduler::SendAcknowledge(frameworkId, agentId)))
.WillOnce(DoAll(
FutureArg<1>(&taskRunning2),
v1::scheduler::SendAcknowledge(frameworkId, agentId)));
// Launch the two task groups.
mesos.send(
v1::createCallAccept(frameworkId, offer, {launchGroup1, launchGroup2}));
// Settle the clock to finish the processing of `taskGroup2`.
Clock::settle();
ASSERT_TRUE(taskStarting2.isPending());
// Resume the processing of `taskGroup1`.
promise1.set(true);
// If taskgroup2 tries to launch the executor first (i.e. if the order is
// not corrected by the agent), taskgroup2 will be subsequently dropped. The
// successful launch of both tasks verifies that the agent enforces the task
// launch order.
AWAIT_READY(taskStarting1);
AWAIT_READY(taskStarting2);
ASSERT_EQ(v1::TASK_STARTING, taskStarting1->status().state());
ASSERT_EQ(v1::TASK_STARTING, taskStarting2->status().state());
AWAIT_READY(taskRunning1);
AWAIT_READY(taskRunning2);
ASSERT_EQ(v1::TASK_RUNNING, taskRunning1->status().state());
ASSERT_EQ(v1::TASK_RUNNING, taskRunning2->status().state());
}
// This test ensures that tasks using the same executor are successfully
// launched in the order in which the agent receives the RunTask(Group)Message,
// even when we manually reorder the completion of the asynchronous task
// authorization step. See MESOS-8624.
TEST_F(SlaveTest, LaunchTasksReorderTaskAuthorization)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
slave::Flags slaveFlags = CreateSlaveFlags();
MockAuthorizer mockAuthorizer;
// Start a mock slave.
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &mockAuthorizer, CreateSlaveFlags(), true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(v1::scheduler::SendSubscribe(v1::DEFAULT_FRAMEWORK_INFO));
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
v1::scheduler::TestMesos mesos(
master.get()->pid, ContentType::PROTOBUF, scheduler);
// Advance the clock to trigger both agent registration and a batch
// allocation.
Clock::advance(slaveFlags.registration_backoff_factor);
Clock::advance(masterFlags.allocation_interval);
AWAIT_READY(subscribed);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::FrameworkID frameworkId(subscribed->framework_id());
v1::ExecutorInfo executorInfo = v1::createExecutorInfo(
"default", None(), resources, v1::ExecutorInfo::DEFAULT, frameworkId);
// Create two separate task groups that use the same executor.
v1::TaskInfo taskInfo1 = v1::createTask(agentId, resources, "sleep 1000");
v1::TaskGroupInfo taskGroup1 = v1::createTaskGroupInfo({taskInfo1});
v1::TaskInfo taskInfo2 = v1::createTask(agentId, resources, "sleep 1000");
v1::TaskGroupInfo taskGroup2 = v1::createTaskGroupInfo({taskInfo2});
v1::Offer::Operation launchGroup1 =
v1::LAUNCH_GROUP(executorInfo, taskGroup1);
v1::Offer::Operation launchGroup2 =
v1::LAUNCH_GROUP(executorInfo, taskGroup2);
// Catch the task authorization step by returning a pending future.
Promise<bool> promise1, promise2;
EXPECT_CALL(
mockAuthorizer,
authorized(AuthorizationRequestHasTaskID(devolve(taskInfo1.task_id()))))
.WillOnce(Return(promise1.future()));
EXPECT_CALL(
mockAuthorizer,
authorized(AuthorizationRequestHasTaskID(devolve(taskInfo2.task_id()))))
.WillOnce(Return(promise2.future()));
Future<v1::scheduler::Event::Update> taskStarting1, taskStarting2;
Future<v1::scheduler::Event::Update> taskRunning1, taskRunning2;
EXPECT_CALL(
*scheduler, update(_, TaskStatusUpdateTaskIdEq(taskInfo1.task_id())))
.WillOnce(DoAll(
FutureArg<1>(&taskStarting1),
v1::scheduler::SendAcknowledge(frameworkId, agentId)))
.WillOnce(DoAll(
FutureArg<1>(&taskRunning1),
v1::scheduler::SendAcknowledge(frameworkId, agentId)));
EXPECT_CALL(
*scheduler, update(_, TaskStatusUpdateTaskIdEq(taskInfo2.task_id())))
.WillOnce(DoAll(
FutureArg<1>(&taskStarting2),
v1::scheduler::SendAcknowledge(frameworkId, agentId)))
.WillOnce(DoAll(
FutureArg<1>(&taskRunning2),
v1::scheduler::SendAcknowledge(frameworkId, agentId)));
// Launch the two task groups.
mesos.send(
v1::createCallAccept(frameworkId, offer, {launchGroup1, launchGroup2}));
// Reorder the task group launches by resuming
// the processing of `taskGroup2` first.
promise2.set(true);
// Settle the clock to finish the processing of `taskGroup2`.
Clock::settle();
ASSERT_TRUE(taskStarting2.isPending());
promise1.set(true);
// If taskgroup2 tries to launch the executor first (i.e. if the order is
// not corrected by the agent), taskgroup2 will be subsequently dropped. The
// successful launch of both tasks verifies that the agent enforces the task
// launch order.
AWAIT_READY(taskStarting1);
AWAIT_READY(taskStarting2);
ASSERT_EQ(v1::TASK_STARTING, taskStarting1->status().state());
ASSERT_EQ(v1::TASK_STARTING, taskStarting2->status().state());
AWAIT_READY(taskRunning1);
AWAIT_READY(taskRunning2);
ASSERT_EQ(v1::TASK_RUNNING, taskRunning1->status().state());
ASSERT_EQ(v1::TASK_RUNNING, taskRunning2->status().state());
}
// This test verifies the agent behavior of launching three task groups using
// the same executor. When all three task groups are launching on the agent
// (before creating any executor), if the first received task group fails to
// launch, subsequent task group launches would also fail.
TEST_F(SlaveTest, LaunchTaskGroupsUsingSameExecutorKillFirstTaskGroup)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
slave::Flags slaveFlags = CreateSlaveFlags();
// Start a mock slave.
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), slaveFlags, true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(v1::scheduler::SendSubscribe(v1::DEFAULT_FRAMEWORK_INFO));
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
EXPECT_CALL(*scheduler, failure(_, _))
.Times(AtMost(1));
v1::scheduler::TestMesos mesos(
master.get()->pid, ContentType::PROTOBUF, scheduler);
// Advance the clock to trigger both agent registration and a batch
// allocation.
Clock::advance(slaveFlags.registration_backoff_factor);
Clock::advance(masterFlags.allocation_interval);
AWAIT_READY(subscribed);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::FrameworkID frameworkId(subscribed->framework_id());
v1::ExecutorInfo executorInfo = v1::createExecutorInfo(
"default1", None(), resources, v1::ExecutorInfo::DEFAULT, frameworkId);
// Create three separate task groups that use the same executor.
v1::TaskInfo taskInfo1 = v1::createTask(agentId, resources, "sleep 1000");
v1::TaskGroupInfo taskGroup1 = v1::createTaskGroupInfo({taskInfo1});
v1::TaskInfo taskInfo2 = v1::createTask(agentId, resources, "sleep 1000");
v1::TaskGroupInfo taskGroup2 = v1::createTaskGroupInfo({taskInfo2});
v1::TaskInfo taskInfo3 = v1::createTask(agentId, resources, "sleep 1000");
v1::TaskGroupInfo taskGroup3 = v1::createTaskGroupInfo({taskInfo3});
v1::Offer::Operation launchGroup1 =
v1::LAUNCH_GROUP(executorInfo, taskGroup1);
v1::Offer::Operation launchGroup2 =
v1::LAUNCH_GROUP(executorInfo, taskGroup2);
v1::Offer::Operation launchGroup3 =
v1::LAUNCH_GROUP(executorInfo, taskGroup3);
Future<v1::scheduler::Event::Update> task1Killed;
Future<v1::scheduler::Event::Update> task2Lost;
Future<v1::scheduler::Event::Update> task3Lost;
EXPECT_CALL(
*scheduler, update(_, TaskStatusUpdateTaskIdEq(taskInfo1.task_id())))
.WillOnce(DoAll(
FutureArg<1>(&task1Killed),
v1::scheduler::SendAcknowledge(frameworkId, agentId)));
EXPECT_CALL(
*scheduler, update(_, TaskStatusUpdateTaskIdEq(taskInfo2.task_id())))
.WillOnce(DoAll(
FutureArg<1>(&task2Lost),
v1::scheduler::SendAcknowledge(frameworkId, agentId)));
EXPECT_CALL(
*scheduler, update(_, TaskStatusUpdateTaskIdEq(taskInfo3.task_id())))
.WillOnce(DoAll(
FutureArg<1>(&task3Lost),
v1::scheduler::SendAcknowledge(frameworkId, agentId)));
// Saved arguments from `Slave::_run()`.
FrameworkInfo _frameworkInfo1, _frameworkInfo2, _frameworkInfo3;
ExecutorInfo _executorInfo1, _executorInfo2, _executorInfo3;
Option<TaskGroupInfo> _taskGroup1, _taskGroup2, _taskGroup3;
Option<TaskInfo> _task1, _task2, _task3;
vector<ResourceVersionUUID>
_resourceVersionUuids1, _resourceVersionUuids2, _resourceVersionUuids3;
Option<bool> _launchExecutor1, _launchExecutor2, _launchExecutor3;
// Pause all taskgroups at `_run` by returning a pending future.
Promise<Nothing> promise1, promise2, promise3;
Future<Nothing> _run1, _run2, _run3;
EXPECT_CALL(
*slave.get()->mock(),
_run(_, _, _,
OptionTaskGroupHasTaskID(devolve(taskInfo1.task_id())),
_, _))
.WillOnce(DoAll(
FutureSatisfy(&_run1),
SaveArg<0>(&_frameworkInfo1),
SaveArg<1>(&_executorInfo1),
SaveArg<2>(&_task1),
SaveArg<3>(&_taskGroup1),
SaveArg<4>(&_resourceVersionUuids1),
SaveArg<5>(&_launchExecutor1),
Return(promise1.future())));
EXPECT_CALL(
*slave.get()->mock(),
_run(_, _, _,
OptionTaskGroupHasTaskID(devolve(taskInfo2.task_id())),
_, _))
.WillOnce(DoAll(
FutureSatisfy(&_run2),
SaveArg<0>(&_frameworkInfo2),
SaveArg<1>(&_executorInfo2),
SaveArg<2>(&_task2),
SaveArg<3>(&_taskGroup2),
SaveArg<4>(&_resourceVersionUuids2),
SaveArg<5>(&_launchExecutor2),
Return(promise2.future())));
EXPECT_CALL(
*slave.get()->mock(),
_run(_, _, _,
OptionTaskGroupHasTaskID(devolve(taskInfo3.task_id())),
_, _))
.WillOnce(DoAll(
FutureSatisfy(&_run3),
SaveArg<0>(&_frameworkInfo3),
SaveArg<1>(&_executorInfo3),
SaveArg<2>(&_task3),
SaveArg<3>(&_taskGroup3),
SaveArg<4>(&_resourceVersionUuids3),
SaveArg<5>(&_launchExecutor3),
Return(promise3.future())));
// Launch task groups.
mesos.send(
v1::createCallAccept(
frameworkId, offer, {launchGroup1, launchGroup2, launchGroup3}));
AWAIT_READY(_run1);
AWAIT_READY(_run2);
AWAIT_READY(_run3);
Future<Nothing> killTask1;
EXPECT_CALL(*slave.get()->mock(), killTask(_, _))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_killTask),
FutureSatisfy(&killTask1)));
// Kill task1.
mesos.send(v1::createCallKill(frameworkId, taskInfo1.task_id(), agentId));
AWAIT_READY(killTask1);
// Resume the continuation for `taskGroup1`.
Future<Nothing> unmocked__run1 = process::dispatch(slave.get()->pid, [=] {
return slave.get()->mock()->unmocked__run(
_frameworkInfo1,
_executorInfo1,
_task1,
_taskGroup1,
_resourceVersionUuids1,
_launchExecutor1);
});
promise1.associate(unmocked__run1);
AWAIT(unmocked__run1);
AWAIT_READY(task1Killed);
EXPECT_EQ(v1::TASK_KILLED, task1Killed->status().state());
// Resume the continuation for taskgroup2.
Future<Nothing> unmocked__run2 = process::dispatch(slave.get()->pid, [=] {
return slave.get()->mock()->unmocked__run(
_frameworkInfo2,
_executorInfo2,
_task2,
_taskGroup2,
_resourceVersionUuids2,
_launchExecutor2);
});
promise2.associate(unmocked__run2);
// Resume the continuation for taskgroup3.
Future<Nothing> unmocked__run3 = process::dispatch(slave.get()->pid, [=] {
return slave.get()->mock()->unmocked__run(
_frameworkInfo3,
_executorInfo3,
_task3,
_taskGroup3,
_resourceVersionUuids3,
_launchExecutor3);
});
promise3.associate(unmocked__run3);
AWAIT(unmocked__run2);
AWAIT_READY(task2Lost);
EXPECT_EQ(v1::TASK_LOST, task2Lost->status().state());
AWAIT(unmocked__run3);
AWAIT_READY(task3Lost);
EXPECT_EQ(v1::TASK_LOST, task3Lost->status().state());
}
// This test verifies the agent behavior of launching two task groups using
// the same executor. When both task groups are launching on the agent
// (before creating any executor), if the second received task group fails to
// launch, the first task group can continue launching successfully.
TEST_F(SlaveTest, LaunchTaskGroupsUsingSameExecutorKillLaterTaskGroup)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
slave::Flags slaveFlags = CreateSlaveFlags();
// Start a mock slave.
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), slaveFlags, true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(v1::scheduler::SendSubscribe(v1::DEFAULT_FRAMEWORK_INFO));
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers));
EXPECT_CALL(*scheduler, failure(_, _))
.Times(AtMost(1));
v1::scheduler::TestMesos mesos(
master.get()->pid, ContentType::PROTOBUF, scheduler);
// Advance the clock to trigger both agent registration and a batch
// allocation.
Clock::advance(slaveFlags.registration_backoff_factor);
Clock::advance(masterFlags.allocation_interval);
AWAIT_READY(subscribed);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::FrameworkID frameworkId(subscribed->framework_id());
v1::ExecutorInfo executorInfo = v1::createExecutorInfo(
"default1", None(), resources, v1::ExecutorInfo::DEFAULT, frameworkId);
// Create two separate task groups that use the same executor.
v1::TaskInfo taskInfo1 = v1::createTask(agentId, resources, "sleep 1000");
v1::TaskGroupInfo taskGroup1 = v1::createTaskGroupInfo({taskInfo1});
v1::TaskInfo taskInfo2 = v1::createTask(agentId, resources, "sleep 1000");
v1::TaskGroupInfo taskGroup2 = v1::createTaskGroupInfo({taskInfo2});
v1::Offer::Operation launchGroup1 =
v1::LAUNCH_GROUP(executorInfo, taskGroup1);
v1::Offer::Operation launchGroup2 =
v1::LAUNCH_GROUP(executorInfo, taskGroup2);
Future<v1::scheduler::Event::Update> task1Starting, task1Running;
Future<v1::scheduler::Event::Update> task2Killed;
EXPECT_CALL(
*scheduler, update(_, TaskStatusUpdateTaskIdEq(taskInfo1.task_id())))
.WillOnce(DoAll(
FutureArg<1>(&task1Starting),
v1::scheduler::SendAcknowledge(frameworkId, agentId)))
.WillOnce(DoAll(
FutureArg<1>(&task1Running),
v1::scheduler::SendAcknowledge(frameworkId, agentId)));
EXPECT_CALL(
*scheduler, update(_, TaskStatusUpdateTaskIdEq(taskInfo2.task_id())))
.WillOnce(DoAll(
FutureArg<1>(&task2Killed),
v1::scheduler::SendAcknowledge(frameworkId, agentId)));
// Saved arguments from `Slave::_run()`.
FrameworkInfo _frameworkInfo1, _frameworkInfo2;
ExecutorInfo _executorInfo1, _executorInfo2;
Option<TaskGroupInfo> _taskGroup1, _taskGroup2;
Option<TaskInfo> _task1, _task2;
vector<ResourceVersionUUID> _resourceVersionUuids1, _resourceVersionUuids2;
Option<bool> _launchExecutor1, _launchExecutor2;
// Pause both taskgroups at `_run` by returning a pending future.
Promise<Nothing> promise1, promise2;
Future<Nothing> _run1, _run2;
EXPECT_CALL(
*slave.get()->mock(),
_run(_, _, _,
OptionTaskGroupHasTaskID(devolve(taskInfo1.task_id())),
_, _))
.WillOnce(DoAll(
FutureSatisfy(&_run1),
SaveArg<0>(&_frameworkInfo1),
SaveArg<1>(&_executorInfo1),
SaveArg<2>(&_task1),
SaveArg<3>(&_taskGroup1),
SaveArg<4>(&_resourceVersionUuids1),
SaveArg<5>(&_launchExecutor1),
Return(promise1.future())));
EXPECT_CALL(
*slave.get()->mock(),
_run(_, _, _,
OptionTaskGroupHasTaskID(devolve(taskInfo2.task_id())),
_, _))
.WillOnce(DoAll(
FutureSatisfy(&_run2),
SaveArg<0>(&_frameworkInfo2),
SaveArg<1>(&_executorInfo2),
SaveArg<2>(&_task2),
SaveArg<3>(&_taskGroup2),
SaveArg<4>(&_resourceVersionUuids2),
SaveArg<5>(&_launchExecutor2),
Return(promise2.future())));
// Launch the two task groups.
mesos.send(
v1::createCallAccept(frameworkId, offer, {launchGroup1, launchGroup2}));
AWAIT_READY(_run1);
AWAIT_READY(_run2);
Future<Nothing> killTask2;
EXPECT_CALL(*slave.get()->mock(), killTask(_, _))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_killTask),
FutureSatisfy(&killTask2)));
// Kill task2.
mesos.send(v1::createCallKill(frameworkId, taskInfo2.task_id(), agentId));
AWAIT_READY(killTask2);
// Resume the continuation for `taskGroup2`.
Future<Nothing> unmocked__run2 = process::dispatch(slave.get()->pid, [=] {
return slave.get()->mock()->unmocked__run(
_frameworkInfo2,
_executorInfo2,
_task2,
_taskGroup2,
_resourceVersionUuids2,
_launchExecutor2);
});
promise2.associate(unmocked__run2);
AWAIT(unmocked__run2);
AWAIT_READY(task2Killed);
EXPECT_EQ(v1::TASK_KILLED, task2Killed->status().state());
// Resume the continuation for taskgroup1.
Future<Nothing> unmocked__run1 = process::dispatch(slave.get()->pid, [=] {
return slave.get()->mock()->unmocked__run(
_frameworkInfo1,
_executorInfo1,
_task1,
_taskGroup1,
_resourceVersionUuids1,
_launchExecutor1);
});
promise1.associate(unmocked__run1);
AWAIT(unmocked__run1);
AWAIT_READY(task1Starting);
EXPECT_EQ(v1::TASK_STARTING, task1Starting->status().state());
AWAIT_READY(task1Running);
EXPECT_EQ(v1::TASK_RUNNING, task1Running->status().state());
}
// This test verifies that when agent shuts down a running executor, launching
// tasks on the agent that use the same executor will be dropped properly.
TEST_F(SlaveTest, ShutdownExecutorWhileTaskLaunching)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
slave::Flags slaveFlags = CreateSlaveFlags();
// Start a mock slave.
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), slaveFlags, true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(v1::scheduler::SendSubscribe(v1::DEFAULT_FRAMEWORK_INFO));
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers));
EXPECT_CALL(*scheduler, failure(_, _))
.Times(AtMost(1));
v1::scheduler::TestMesos mesos(
master.get()->pid, ContentType::PROTOBUF, scheduler);
// Advance the clock to trigger both agent registration and a batch
// allocation.
Clock::advance(slaveFlags.registration_backoff_factor);
Clock::advance(masterFlags.allocation_interval);
AWAIT_READY(subscribed);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::FrameworkID frameworkId(subscribed->framework_id());
v1::ExecutorInfo executorInfo = v1::createExecutorInfo(
"default1", None(), resources, v1::ExecutorInfo::DEFAULT, frameworkId);
// Create two separate task groups that use the same executor.
v1::TaskInfo taskInfo1 = v1::createTask(agentId, resources, "sleep 1000");
v1::TaskGroupInfo taskGroup1 = v1::createTaskGroupInfo({taskInfo1});
v1::TaskInfo taskInfo2 = v1::createTask(agentId, resources, "sleep 1000");
v1::TaskGroupInfo taskGroup2 = v1::createTaskGroupInfo({taskInfo2});
v1::Offer::Operation launchGroup1 =
v1::LAUNCH_GROUP(executorInfo, taskGroup1);
v1::Offer::Operation launchGroup2 =
v1::LAUNCH_GROUP(executorInfo, taskGroup2);
Future<v1::scheduler::Event::Update> task1Starting, task1Running;
Future<v1::scheduler::Event::Update> task2Lost;
EXPECT_CALL(
*scheduler, update(_, TaskStatusUpdateTaskIdEq(taskInfo1.task_id())))
.WillOnce(DoAll(
FutureArg<1>(&task1Starting),
v1::scheduler::SendAcknowledge(frameworkId, agentId)))
.WillOnce(DoAll(
FutureArg<1>(&task1Running),
v1::scheduler::SendAcknowledge(frameworkId, agentId)));
EXPECT_CALL(
*scheduler, update(_, TaskStatusUpdateTaskIdEq(taskInfo2.task_id())))
.WillOnce(DoAll(
FutureArg<1>(&task2Lost),
v1::scheduler::SendAcknowledge(frameworkId, agentId)));
// Saved arguments from `Slave::_run()` for `taskGroup2`.
FrameworkInfo _frameworkInfo;
ExecutorInfo _executorInfo;
Option<TaskGroupInfo> _taskGroup;
Option<TaskInfo> _task;
vector<ResourceVersionUUID> _resourceVersionUuids;
Option<bool> _launchExecutor;
// Pause the launch of `taskGroup2` at `_run` by returning a pending future.
Promise<Nothing> promiseTask2;
Future<Nothing> runTask2;
EXPECT_CALL(
*slave.get()->mock(),
_run(_, _, _,
OptionTaskGroupHasTaskID(devolve(taskInfo2.task_id())),
_, _))
.WillOnce(
DoAll(FutureSatisfy(&runTask2),
SaveArg<0>(&_frameworkInfo),
SaveArg<1>(&_executorInfo),
SaveArg<2>(&_task),
SaveArg<3>(&_taskGroup),
SaveArg<4>(&_resourceVersionUuids),
SaveArg<5>(&_launchExecutor),
Return(promiseTask2.future())));
// Launch the two task groups.
mesos.send(
v1::createCallAccept(frameworkId, offer, {launchGroup1, launchGroup2}));
AWAIT_READY(runTask2);
// `taskGroup1` launches successfully.
AWAIT_READY(task1Starting);
EXPECT_EQ(v1::TASK_STARTING, task1Starting->status().state());
AWAIT_READY(task1Running);
EXPECT_EQ(v1::TASK_RUNNING, task1Running->status().state());
// Shutdown the executor while `taskGroup2` is still launching.
Future<Nothing> shutdownExecutor;
EXPECT_CALL(*slave.get()->mock(), shutdownExecutor(_, _, _))
.WillOnce(DoAll(
Invoke(slave.get()->mock(), &MockSlave::unmocked_shutdownExecutor),
FutureSatisfy(&shutdownExecutor)));
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::SHUTDOWN);
Call::Shutdown* shutdown = call.mutable_shutdown();
shutdown->mutable_executor_id()->CopyFrom(executorInfo.executor_id());
shutdown->mutable_agent_id()->CopyFrom(agentId);
mesos.send(call);
}
AWAIT_READY(shutdownExecutor);
// Resume launching `taskGroup2`.
Future<Nothing> unmocked__run = process::dispatch(slave.get()->pid, [=] {
return slave.get()->mock()->unmocked__run(
_frameworkInfo,
_executorInfo,
_task,
_taskGroup,
_resourceVersionUuids,
_launchExecutor);
});
promiseTask2.associate(unmocked__run);
// `taskGroup2` is dropped because the executor is terminated.
AWAIT_READY(task2Lost);
EXPECT_EQ(v1::TASK_LOST, task2Lost->status().state());
}
// This test ensures that agent sends ExitedExecutorMessage when the task
// fails to launch due to task authorization failure and that master's executor
// bookkeeping entry is removed.
TEST_F(SlaveTest, RemoveExecutorUponFailedTaskAuthorization)
{
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
MockAuthorizer mockAuthorizer;
slave::Flags slaveFlags = CreateSlaveFlags();
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a mock slave.
Try<Owned<cluster::Slave>> slave = StartSlave(
detector.get(),
&containerizer,
&mockAuthorizer,
slaveFlags,
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
Resources executorResources = Resources::parse("cpus:0.1;mem:32").get();
executorResources.allocate("*");
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers.get()[0].slave_id());
task.mutable_resources()->MergeFrom(
Resources(offers.get()[0].resources()) - executorResources);
task.mutable_executor()->MergeFrom(DEFAULT_EXECUTOR_INFO);
task.mutable_executor()->mutable_resources()->CopyFrom(executorResources);
EXPECT_CALL(exec, registered(_, _, _, _))
.Times(0);
Future<TaskStatus> statusError;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusError));
EXPECT_CALL(sched, executorLost(&driver, DEFAULT_EXECUTOR_ID, _, _))
.Times(AtMost(1));
Future<ExitedExecutorMessage> exitedExecutorMessage =
FUTURE_PROTOBUF(ExitedExecutorMessage(), _, _);
// Induce agent task authorization failure. This will result in
// task launch failure before the executor launch.
EXPECT_CALL(mockAuthorizer, authorized(_))
.WillRepeatedly(Return(false));
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(exitedExecutorMessage);
AWAIT_READY(statusError);
ASSERT_EQ(TASK_ERROR, statusError->state());
// Helper function to post a request to '/api/v1' master endpoint
// and return the response.
auto post = [](
const process::PID<master::Master>& pid,
const v1::master::Call& call,
const ContentType& contentType)
{
process::http::Headers headers = createBasicAuthHeaders(DEFAULT_CREDENTIAL);
headers["Accept"] = stringify(contentType);
return process::http::post(
pid,
"api/v1",
headers,
serialize(contentType, call),
stringify(contentType));
};
v1::master::Call v1Call;
v1Call.set_type(v1::master::Call::GET_EXECUTORS);
Future<process::http::Response> response =
post(master.get()->pid, v1Call, ContentType::PROTOBUF);
response.await();
ASSERT_EQ(response->status, process::http::OK().status);
Future<v1::master::Response> v1Response =
deserialize<v1::master::Response>(ContentType::PROTOBUF, response->body);
// Master has no executor entry because the executor never launched.
ASSERT_TRUE(v1Response->IsInitialized());
ASSERT_EQ(v1::master::Response::GET_EXECUTORS, v1Response->type());
ASSERT_EQ(0, v1Response->get_executors().executors_size());
driver.stop();
driver.join();
}
// This test verifies that the executor is shutdown if all of its initial
// tasks could not be delivered, even after the executor has been registered.
// See MESOS-8411.
TEST_F(SlaveTest, KillAllInitialTasksTerminatesExecutor)
{
// Start a master.
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
// Start a mock slave.
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &containerizer, CreateSlaveFlags(), true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
Resources executorResources = Resources::parse("cpus:0.1;mem:32").get();
executorResources.allocate("*");
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers->at(0).slave_id());
task.mutable_resources()->MergeFrom(
Resources(offers->at(0).resources()) - executorResources);
task.mutable_executor()->MergeFrom(DEFAULT_EXECUTOR_INFO);
task.mutable_executor()->mutable_resources()->CopyFrom(executorResources);
Future<TaskStatus> killTaskStatus;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&killTaskStatus));
// Saved arguments from `Slave::___run()`.
Future<Nothing> future;
FrameworkID frameworkId;
ExecutorID executorId;
ContainerID containerId;
vector<TaskInfo> tasks;
vector<TaskGroupInfo> taskGroups;
// Kill the task after executor registration but before
// task launch in `___run()`.
Future<Nothing> ___run;
EXPECT_CALL(*slave.get()->mock(), ___run(_, _, _, _, _, _))
.WillOnce(DoAll(
SaveArg<0>(&future),
SaveArg<1>(&frameworkId),
SaveArg<2>(&executorId),
SaveArg<3>(&containerId),
SaveArg<4>(&tasks),
SaveArg<5>(&taskGroups),
FutureSatisfy(&___run)
));
EXPECT_CALL(exec, registered(_, _, _, _));
EXPECT_CALL(exec, launchTask(_, _))
.Times(0);
// The exeuctor is killed because its initial task is killed
// and cannot be delivered.
Future<Nothing> executorShutdown;
EXPECT_CALL(exec, shutdown(_))
.WillOnce(FutureSatisfy(&executorShutdown));
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(___run);
driver.killTask(task.task_id());
// Task is killed before the actual `___run()` is triggered.
AWAIT_READY(killTaskStatus);
EXPECT_EQ(TASK_KILLED, killTaskStatus->state());
EXPECT_EQ(
TaskStatus::REASON_TASK_KILLED_DURING_LAUNCH,
killTaskStatus->reason());
// We continue dispatching `___run()` to make sure the `CHECK()` in the
// continuation passes.
Future<Nothing> unmocked___run = process::dispatch(slave.get()->pid, [=] {
slave.get()->mock()->unmocked____run(
future, frameworkId, executorId, containerId, tasks, taskGroups);
return Nothing();
});
AWAIT_READY(unmocked___run);
AWAIT_READY(executorShutdown);
driver.stop();
driver.join();
}
// This test verifies that the executor is shutdown during re-registration if
// all of its initial tasks could not be delivered.
TEST_F(SlaveTest, AgentFailoverTerminatesExecutorWithNoTask)
{
// Start a master.
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
slave::Flags slaveFlags = CreateSlaveFlags();
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a mock slave.
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), slaveFlags, true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
// Enable checkpointing for the framework.
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_checkpoint(true);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, frameworkInfo, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
TaskInfo task = createTask(offers->front(), "sleep 1000");
// Before sending the the task to the executor, restart the agent.
Future<Nothing> ___run;
EXPECT_CALL(*slave.get()->mock(), ___run(_, _, _, _, _, _))
.WillOnce(FutureSatisfy(&___run));
driver.launchTasks(offers->at(0).id(), {task});
AWAIT_READY(___run);
slave.get()->terminate();
slave = StartSlave(detector.get(), slaveFlags, true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
Future<Nothing> shutdownExecutor;
EXPECT_CALL(*slave.get()->mock(), _shutdownExecutor(_, _))
.WillOnce(FutureSatisfy(&shutdownExecutor));
slave.get()->start();
// The executor is killed during reregistration because its initial task is
// killed and cannot be delivered.
AWAIT_READY(shutdownExecutor);
driver.stop();
driver.join();
}
// This test verifies that the v1 executor is shutdown if all of its initial
// task group could not be delivered, even after the executor has been
// registered. See MESOS-8411. This test only uses task group.
//
// TODO(mzhu): This test could be simplified if we had a test scheduler that
// provides some basic task launching functionality (see MESOS-8511).
TEST_F(SlaveTest, KillAllInitialTasksTerminatesHTTPExecutor)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
auto executor = std::make_shared<v1::MockHTTPExecutor>();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::ExecutorInfo executorInfo = v1::DEFAULT_EXECUTOR_INFO;
executorInfo.set_type(v1::ExecutorInfo::CUSTOM);
executorInfo.mutable_resources()->CopyFrom(resources);
const v1::ExecutorID& executorId = executorInfo.executor_id();
TestContainerizer containerizer(devolve(executorId), executor);
// Start a mock slave.
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &containerizer, CreateSlaveFlags(), true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
mesos.send(
v1::createCallSubscribe(v1::DEFAULT_FRAMEWORK_INFO));
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(frameworkId);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
Future<v1::executor::Mesos*> executorLib;
EXPECT_CALL(*executor, connected(_))
.WillOnce(FutureArg<0>(&executorLib));
EXPECT_CALL(*executor, subscribed(_, _));
// Saved arguments from `Slave::___run()`.
Future<Nothing> _future;
FrameworkID _frameworkId;
ExecutorID _executorId;
ContainerID _containerId;
vector<TaskInfo> _tasks;
vector<TaskGroupInfo> _taskGroups;
// Kill the task after executor subscription but before
// task launch in `___run()`.
Future<Nothing> taskRun;
EXPECT_CALL(*slave.get()->mock(), ___run(_, _, _, _, _, _))
.WillOnce(DoAll(
SaveArg<0>(&_future),
SaveArg<1>(&_frameworkId),
SaveArg<2>(&_executorId),
SaveArg<3>(&_containerId),
SaveArg<4>(&_tasks),
SaveArg<5>(&_taskGroups),
FutureSatisfy(&taskRun)
));
v1::TaskInfo task1 =
v1::createTask(agentId, resources, "");
v1::TaskInfo task2 =
v1::createTask(agentId, resources, "");
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(task1);
taskGroup.add_tasks()->CopyFrom(task2);
v1::Offer::Operation launchGroup = v1::LAUNCH_GROUP(executorInfo, taskGroup);
mesos.send(
v1::createCallAccept(frameworkId, offer, {launchGroup}));
AWAIT_READY(executorLib);
{
v1::executor::Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.mutable_executor_id()->CopyFrom(executorId);
call.set_type(v1::executor::Call::SUBSCRIBE);
call.mutable_subscribe();
executorLib.get()->send(call);
}
// Kill the task right before the task launch. By now, the executor has
// already subscribed.
AWAIT_READY(taskRun);
Future<Nothing> shutdown;
EXPECT_CALL(*executor, shutdown(_))
.WillOnce(FutureSatisfy(&shutdown));
Future<v1::scheduler::Event::Update> update1;
Future<v1::scheduler::Event::Update> update2;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(FutureArg<1>(&update1))
.WillOnce(FutureArg<1>(&update2));
EXPECT_CALL(*scheduler, failure(_, _))
.Times(AtMost(1));
// We kill only one of the tasks and expect the entire task group to be
// killed.
mesos.send(
v1::createCallKill(frameworkId, task1.task_id(), offer.agent_id()));
AWAIT_READY(update1);
AWAIT_READY(update2);
ASSERT_EQ(v1::TASK_KILLED, update1->status().state());
ASSERT_EQ(v1::TASK_KILLED, update2->status().state());
// We continue dispatching `___run()` to make sure the `CHECK()` in the
// continuation passes.
Future<Nothing> unmocked____run = process::dispatch(slave.get()->pid, [=] {
slave.get()->mock()->unmocked____run(
_future, _frameworkId, _executorId, _containerId, _tasks, _taskGroups);
return Nothing();
});
// The executor is killed because all of its initial tasks are killed
// and cannot be delivered.
AWAIT_READY(shutdown);
// It is necessary to wait for the `unmocked____run` call to finish before we
// get out of scope. Otherwise, the objects we captured by reference and
// passed to `unmocked___run` may be destroyed in the middle of the function.
AWAIT_READY(unmocked____run);
}
// This test verifies that if an agent fails over after registering
// a v1 executor but before delivering its initial task groups, the
// executor will be shut down since all of its initial task groups
// were dropped. See MESOS-8411.
//
// TODO(mzhu): This test could be simplified if we had a test scheduler that
// provides some basic task launching functionality (see MESOS-8511).
TEST_F(SlaveTest, AgentFailoverTerminatesHTTPExecutorWithNoTask)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
Owned<MasterDetector> detector = master.get()->createDetector();
slave::Flags slaveFlags = CreateSlaveFlags();
string processId = process::ID::generate("slave");
// Start a mock slave.
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), processId, slaveFlags, true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
// Enable checkpointing for the framework.
v1::FrameworkInfo frameworkInfo = v1::DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_checkpoint(true);
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::ExecutorInfo executorInfo = v1::DEFAULT_EXECUTOR_INFO;
executorInfo.clear_command();
executorInfo.set_type(v1::ExecutorInfo::DEFAULT);
executorInfo.mutable_resources()->CopyFrom(resources);
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
mesos.send(
v1::createCallSubscribe(frameworkInfo));
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(frameworkId);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
Future<Nothing> ___run;
EXPECT_CALL(*slave.get()->mock(), ___run(_, _, _, _, _, _))
.WillOnce(FutureSatisfy(&___run));
v1::TaskInfo task1 =
v1::createTask(agentId, resources, "sleep 1000");
v1::TaskInfo task2 =
v1::createTask(agentId, resources, "sleep 1000");
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(task1);
taskGroup.add_tasks()->CopyFrom(task2);
v1::Offer::Operation launchGroup = v1::LAUNCH_GROUP(executorInfo, taskGroup);
mesos.send(
v1::createCallAccept(frameworkId, offer, {launchGroup}));
// Before sending the the task to the executor, restart the agent.
AWAIT_READY(___run);
slave.get()->terminate();
slave->reset();
slave = StartSlave(detector.get(), processId, slaveFlags, true);
slave.get()->start();
Future<Nothing> _shutdownExecutor;
EXPECT_CALL(*slave.get()->mock(), _shutdownExecutor(_, _))
.WillOnce(FutureSatisfy(&_shutdownExecutor));
// The executor is killed because all of its initial tasks are killed
// and cannot be delivered.
AWAIT_READY(_shutdownExecutor);
}
// This test verifies that when a slave reregisters with the master
// it correctly includes the latest and status update task states.
TEST_F(SlaveTest, ReregisterWithStatusUpdateTaskState)
{
Clock::pause();
// Start a master.
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
// Create a StandaloneMasterDetector to enable the slave to trigger
// re-registration later.
StandaloneMasterDetector detector(master.get()->pid);
// Start a slave.
slave::Flags agentFlags = CreateSlaveFlags();
Try<Owned<cluster::Slave>> slave =
StartSlave(&detector, &containerizer, agentFlags);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
EXPECT_CALL(sched, resourceOffers(_, _))
.WillOnce(LaunchTasks(DEFAULT_EXECUTOR_INFO, 1, 2, 1024, "*"))
.WillRepeatedly(Return()); // Ignore subsequent offers.
ExecutorDriver* execDriver;
EXPECT_CALL(exec, registered(_, _, _, _))
.WillOnce(SaveArg<0>(&execDriver));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
// Signal when the first update is dropped.
Future<StatusUpdateMessage> statusUpdateMessage =
DROP_PROTOBUF(StatusUpdateMessage(), _, master.get()->pid);
Future<Nothing> ___statusUpdate = FUTURE_DISPATCH(_, &Slave::___statusUpdate);
driver.start();
Clock::advance(masterFlags.allocation_interval);
// Wait until TASK_RUNNING is sent to the master.
AWAIT_READY(statusUpdateMessage);
// Ensure task status update manager handles TASK_RUNNING update.
AWAIT_READY(___statusUpdate);
Future<Nothing> ___statusUpdate2 =
FUTURE_DISPATCH(_, &Slave::___statusUpdate);
// Now send TASK_FINISHED update.
TaskStatus finishedStatus;
finishedStatus = statusUpdateMessage->update().status();
finishedStatus.set_state(TASK_FINISHED);
execDriver->sendStatusUpdate(finishedStatus);
// Ensure task status update manager handles TASK_FINISHED update.
AWAIT_READY(___statusUpdate2);
Future<ReregisterSlaveMessage> reregisterSlaveMessage =
FUTURE_PROTOBUF(ReregisterSlaveMessage(), _, _);
// Drop any updates to the failed over master.
DROP_PROTOBUFS(StatusUpdateMessage(), _, master.get()->pid);
// Simulate a new master detected event on the slave,
// so that the slave will do a re-registration.
detector.appoint(master.get()->pid);
// Force evaluation of master detection before we advance clock to trigger
// agent registration.
Clock::settle();
// Capture and inspect the slave reregistration message.
Clock::advance(agentFlags.registration_backoff_factor);
AWAIT_READY(reregisterSlaveMessage);
ASSERT_EQ(1, reregisterSlaveMessage->tasks_size());
// The latest state of the task should be TASK_FINISHED.
ASSERT_EQ(TASK_FINISHED, reregisterSlaveMessage->tasks(0).state());
// The status update state of the task should be TASK_RUNNING.
ASSERT_EQ(TASK_RUNNING,
reregisterSlaveMessage->tasks(0).status_update_state());
// The status update uuid should match the TASK_RUNNING's uuid.
ASSERT_EQ(statusUpdateMessage->update().uuid(),
reregisterSlaveMessage->tasks(0).status_update_uuid());
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// The agent's operation status update manager should retry updates for
// operations on agent default resources. Here we drop the first such update and
// verify that the update is sent again after the retry interval elapses.
TEST_F(SlaveTest, UpdateOperationStatusRetry)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
slave::Flags slaveFlags = CreateSlaveFlags();
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), slaveFlags);
ASSERT_SOME(slave);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
v1::FrameworkInfo frameworkInfo = v1::DEFAULT_FRAMEWORK_INFO;
frameworkInfo.clear_roles();
frameworkInfo.add_roles("test-role");
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(v1::scheduler::SendSubscribe(frameworkInfo));
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
ContentType contentType = ContentType::PROTOBUF;
v1::scheduler::TestMesos mesos(
master.get()->pid,
contentType,
scheduler);
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
Clock::advance(masterFlags.allocation_interval);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
v1::Resources unreserved = offer.resources();
v1::Resources reserved = unreserved.pushReservation(
v1::createDynamicReservationInfo(
frameworkInfo.roles(0),
frameworkInfo.principal()));
v1::Offer::Operation reserve = v1::RESERVE(reserved);
Future<ApplyOperationMessage> applyOperationMessage =
FUTURE_PROTOBUF(ApplyOperationMessage(), _, slave.get()->pid);
// Drop the first operation status update.
Future<UpdateOperationStatusMessage> droppedOperation =
DROP_PROTOBUF(UpdateOperationStatusMessage(), _, master.get()->pid);
mesos.send(
v1::createCallAccept(
frameworkId,
offer,
{reserve}));
AWAIT_READY(applyOperationMessage);
UUID operationUuid = applyOperationMessage->operation_uuid();
AWAIT_READY(droppedOperation);
ASSERT_EQ(droppedOperation->operation_uuid(), operationUuid);
// Confirm that the agent retries the update.
Future<UpdateOperationStatusMessage> updateOperationStatusMessage =
FUTURE_PROTOBUF(UpdateOperationStatusMessage(), _, master.get()->pid);
Clock::advance(slave::STATUS_UPDATE_RETRY_INTERVAL_MIN);
AWAIT_READY(updateOperationStatusMessage);
ASSERT_EQ(updateOperationStatusMessage->operation_uuid(), operationUuid);
EXPECT_TRUE(metricEquals("master/operations/finished", 1));
Clock::resume();
}
// This test verifies that the slave should properly handle the case
// where the containerizer usage call fails when getting the usage
// information.
TEST_F(SlaveTest, ContainerizerUsageFailure)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
StandaloneMasterDetector detector(master.get()->pid);
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
None(),
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
EXPECT_CALL(exec, registered(_, _, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
const Offer& offer = offers.get()[0];
TaskInfo task = createTask(
offer.slave_id(),
Resources::parse("cpus:0.1;mem:32").get(),
SLEEP_COMMAND(1000),
exec.id);
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
driver.launchTasks(offer.id(), {task});
AWAIT_READY(status);
EXPECT_EQ(TASK_RUNNING, status->state());
// Set up the containerizer so the next usage() will fail.
EXPECT_CALL(containerizer, usage(_))
.WillOnce(Return(Failure("Injected failure")));
// We expect that the slave will still returns ResourceUsage but no
// statistics will be found.
Future<ResourceUsage> usage = slave.get()->mock()->usage();
AWAIT_READY(usage);
ASSERT_EQ(1, usage->executors_size());
EXPECT_FALSE(usage->executors(0).has_statistics());
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// This test verifies that DiscoveryInfo and Port messages, set in TaskInfo,
// are exposed over the slave state endpoint. The test launches a task with
// the DiscoveryInfo and Port message fields populated. It then makes an HTTP
// request to the state endpoint of the slave and retrieves the JSON data from
// the endpoint. The test passes if the DiscoveryInfo and Port message data in
// JSON matches the corresponding data set in the TaskInfo used to launch the
// task.
TEST_F(SlaveTest, DiscoveryInfoAndPorts)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
TaskInfo task = createTask(
offers.get()[0],
SLEEP_COMMAND(100),
DEFAULT_EXECUTOR_ID);
Labels labels1;
labels1.add_labels()->CopyFrom(createLabel("ACTION", "port:7987 DENY"));
Labels labels2;
labels2.add_labels()->CopyFrom(createLabel("ACTION", "port:7789 PERMIT"));
Ports ports;
Port* port1 = ports.add_ports();
port1->set_number(80);
port1->mutable_labels()->CopyFrom(labels1);
Port* port2 = ports.add_ports();
port2->set_number(8081);
port2->mutable_labels()->CopyFrom(labels2);
DiscoveryInfo discovery;
discovery.set_name("test_discovery");
discovery.set_visibility(DiscoveryInfo::CLUSTER);
discovery.mutable_ports()->CopyFrom(ports);
task.mutable_discovery()->CopyFrom(discovery);
EXPECT_CALL(exec, registered(_, _, _, _));
Future<Nothing> launchTask;
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(FutureSatisfy(&launchTask));
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(launchTask);
// Verify label key and value in slave state endpoint.
Future<Response> response = process::http::get(
slave.get()->pid,
"state",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
Try<JSON::Object> parse = JSON::parse<JSON::Object>(response->body);
ASSERT_SOME(parse);
Result<JSON::Object> discoveryResult = parse->find<JSON::Object>(
"frameworks[0].executors[0].tasks[0].discovery");
EXPECT_SOME(discoveryResult);
JSON::Object discoveryObject = discoveryResult.get();
EXPECT_EQ(JSON::Object(JSON::protobuf(discovery)), discoveryObject);
// Check the ports are set in the `DiscoveryInfo` object.
Result<JSON::Object> portResult1 = discoveryObject.find<JSON::Object>(
"ports.ports[0]");
Result<JSON::Object> portResult2 = discoveryObject.find<JSON::Object>(
"ports.ports[1]");
EXPECT_SOME(portResult1);
EXPECT_SOME(portResult2);
// Verify that the ports retrieved from state endpoint are the ones
// that were set.
EXPECT_EQ(JSON::Object(JSON::protobuf(*port1)), portResult1.get());
EXPECT_EQ(JSON::Object(JSON::protobuf(*port2)), portResult2.get());
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// This test verifies that executor labels are
// exposed in the slave's state endpoint.
TEST_F(SlaveTest, ExecutorLabels)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers.get()[0].slave_id());
task.mutable_resources()->MergeFrom(offers.get()[0].resources());
task.mutable_executor()->MergeFrom(DEFAULT_EXECUTOR_INFO);
// Add three labels to the executor (two of which share the same key).
Labels* labels = task.mutable_executor()->mutable_labels();
labels->add_labels()->CopyFrom(createLabel("key1", "value1"));
labels->add_labels()->CopyFrom(createLabel("key2", "value2"));
labels->add_labels()->CopyFrom(createLabel("key1", "value3"));
EXPECT_CALL(exec, registered(_, _, _, _));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(status);
EXPECT_EQ(TASK_RUNNING, status->state());
// Verify label key and value in slave state endpoint.
Future<Response> response = process::http::get(
slave.get()->pid,
"state",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
Try<JSON::Object> parse = JSON::parse<JSON::Object>(response->body);
ASSERT_SOME(parse);
Result<JSON::Array> labels_ = parse->find<JSON::Array>(
"frameworks[0].executors[0].labels");
ASSERT_SOME(labels_);
// Verify the contents of labels.
EXPECT_EQ(3u, labels_->values.size());
EXPECT_EQ(JSON::Value(JSON::protobuf(createLabel("key1", "value1"))),
labels_->values[0]);
EXPECT_EQ(JSON::Value(JSON::protobuf(createLabel("key2", "value2"))),
labels_->values[1]);
EXPECT_EQ(JSON::Value(JSON::protobuf(createLabel("key1", "value3"))),
labels_->values[2]);
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// This test verifies that label values can be set for tasks and that
// they are exposed over the slave state endpoint.
TEST_F(SlaveTest, TaskLabels)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers.get()[0].slave_id());
task.mutable_resources()->MergeFrom(offers.get()[0].resources());
task.mutable_executor()->MergeFrom(DEFAULT_EXECUTOR_INFO);
// Add three labels to the task (two of which share the same key).
Labels* labels = task.mutable_labels();
labels->add_labels()->CopyFrom(createLabel("foo", "bar"));
labels->add_labels()->CopyFrom(createLabel("bar", "baz"));
labels->add_labels()->CopyFrom(createLabel("bar", "qux"));
EXPECT_CALL(exec, registered(_, _, _, _));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<Nothing> update;
EXPECT_CALL(containerizer,
update(_, Resources(offers.get()[0].resources()), _))
.WillOnce(DoAll(FutureSatisfy(&update),
Return(Nothing())));
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(status);
EXPECT_EQ(TASK_RUNNING, status->state());
AWAIT_READY(update);
// Verify label key and value in slave state endpoint.
Future<Response> response = process::http::get(
slave.get()->pid,
"state",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
Try<JSON::Object> parse = JSON::parse<JSON::Object>(response->body);
ASSERT_SOME(parse);
Result<JSON::Array> find = parse->find<JSON::Array>(
"frameworks[0].executors[0].tasks[0].labels");
EXPECT_SOME(find);
JSON::Array labelsObject = find.get();
// Verify the contents of 'foo:bar', 'bar:baz', and 'bar:qux' pairs.
EXPECT_EQ(
JSON::Value(JSON::protobuf(createLabel("foo", "bar"))),
labelsObject.values[0]);
EXPECT_EQ(
JSON::Value(JSON::protobuf(createLabel("bar", "baz"))),
labelsObject.values[1]);
EXPECT_EQ(
JSON::Value(JSON::protobuf(createLabel("bar", "qux"))),
labelsObject.values[2]);
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// This test verifies that TaskStatus label values are exposed over
// the slave state endpoint.
TEST_F(SlaveTest, TaskStatusLabels)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
TaskInfo task = createTask(
offers.get()[0],
SLEEP_COMMAND(100),
DEFAULT_EXECUTOR_ID);
ExecutorDriver* execDriver;
EXPECT_CALL(exec, registered(_, _, _, _))
.WillOnce(SaveArg<0>(&execDriver));
Future<TaskInfo> execTask;
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(FutureArg<1>(&execTask));
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(execTask);
// Now send TASK_RUNNING update.
TaskStatus runningStatus;
runningStatus.mutable_task_id()->MergeFrom(execTask->task_id());
runningStatus.set_state(TASK_RUNNING);
// Add three labels to the task (two of which share the same key).
Labels* labels = runningStatus.mutable_labels();
labels->add_labels()->CopyFrom(createLabel("foo", "bar"));
labels->add_labels()->CopyFrom(createLabel("bar", "baz"));
labels->add_labels()->CopyFrom(createLabel("bar", "qux"));
execDriver->sendStatusUpdate(runningStatus);
AWAIT_READY(status);
// Verify label key and value in master state endpoint.
Future<Response> response = process::http::get(
slave.get()->pid,
"state",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
Try<JSON::Object> parse = JSON::parse<JSON::Object>(response->body);
ASSERT_SOME(parse);
Result<JSON::Array> find = parse->find<JSON::Array>(
"frameworks[0].executors[0].tasks[0].statuses[0].labels");
EXPECT_SOME(find);
JSON::Array labelsObject = find.get();
// Verify the contents of 'foo:bar', 'bar:baz', and 'bar:qux' pairs.
EXPECT_EQ(
JSON::Value(JSON::protobuf(createLabel("foo", "bar"))),
labelsObject.values[0]);
EXPECT_EQ(
JSON::Value(JSON::protobuf(createLabel("bar", "baz"))),
labelsObject.values[1]);
EXPECT_EQ(
JSON::Value(JSON::protobuf(createLabel("bar", "qux"))),
labelsObject.values[2]);
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// This test verifies that TaskStatus::container_status an is exposed over
// the slave state endpoint.
TEST_F(SlaveTest, TaskStatusContainerStatus)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
TaskInfo task = createTask(
offers.get()[0],
SLEEP_COMMAND(100),
DEFAULT_EXECUTOR_ID);
EXPECT_CALL(exec, registered(_, _, _, _));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(status);
const string slaveIPAddress = stringify(slave.get()->pid.address.ip);
// Validate that the Slave has passed in its IP address in
// TaskStatus.container_status.network_infos[0].ip_address.
EXPECT_TRUE(status->has_container_status());
EXPECT_EQ(1, status->container_status().network_infos().size());
EXPECT_EQ(1, status->container_status().network_infos(0).ip_addresses().size()); // NOLINT(whitespace/line_length)
NetworkInfo::IPAddress ipAddress =
status->container_status().network_infos(0).ip_addresses(0);
ASSERT_TRUE(ipAddress.has_ip_address());
EXPECT_EQ(slaveIPAddress, ipAddress.ip_address());
// Now do the same validation with state endpoint.
Future<Response> response = process::http::get(
slave.get()->pid,
"state",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
Try<JSON::Object> parse = JSON::parse<JSON::Object>(response->body);
ASSERT_SOME(parse);
// Validate that the IP address passed in by the Slave is available at the
// state endpoint.
ASSERT_SOME_EQ(
slaveIPAddress,
parse->find<JSON::String>(
"frameworks[0].executors[0].tasks[0].statuses[0]"
".container_status.network_infos[0]"
".ip_addresses[0].ip_address"));
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// Test that we can set the executors environment variables and it
// won't inherit the slaves.
TEST_F(SlaveTest, ExecutorEnvironmentVariables)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
// Need flags for 'executor_environment_variables'.
slave::Flags flags = CreateSlaveFlags();
const std::string path = os::host_default_path();
flags.executor_environment_variables = JSON::Object{{"PATH", path}};
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), flags);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
// Launch a task with the command executor.
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers.get()[0].slave_id());
task.mutable_resources()->MergeFrom(offers.get()[0].resources());
// Command executor will run as user running test.
CommandInfo command;
#ifdef __WINDOWS__
command.set_shell(false);
command.set_value("powershell.exe");
command.add_arguments("powershell.exe");
command.add_arguments("-NoProfile");
command.add_arguments("-Command");
command.add_arguments(
"if ($env:PATH -eq '" + path + "') { exit 0 } else { exit 1 }");
#else
command.set_shell(true);
command.set_value("test $PATH = " + path);
#endif // __WINDOWS__
task.mutable_command()->MergeFrom(command);
Future<TaskStatus> statusStarting;
Future<TaskStatus> statusRunning;
Future<TaskStatus> statusFinished;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusStarting))
.WillOnce(FutureArg<1>(&statusRunning))
.WillOnce(FutureArg<1>(&statusFinished));
driver.launchTasks(offers.get()[0].id(), {task});
// Scheduler should first receive TASK_STARTING, followed by
// TASK_STARTING and TASK_FINISHED from the executor.
AWAIT_READY(statusStarting);
EXPECT_EQ(TASK_STARTING, statusStarting->state());
AWAIT_READY(statusRunning);
EXPECT_EQ(TASK_RUNNING, statusRunning->state());
AWAIT_READY(statusFinished);
EXPECT_EQ(TASK_FINISHED, statusFinished->state());
driver.stop();
driver.join();
}
// This test verifies that the slave should properly show total slave
// resources.
TEST_F(SlaveTest, TotalSlaveResourcesIncludedInUsage)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
TestContainerizer containerizer;
StandaloneMasterDetector detector(master.get()->pid);
slave::Flags flags = CreateSlaveFlags();
flags.resources = "cpus:2;gpus:0;mem:1024;disk:1024;ports:[31000-32000]";
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
flags,
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
Clock::pause();
// Wait for slave to be initialized.
Clock::settle();
// We expect that the slave will return ResourceUsage with
// total resources reported.
Future<ResourceUsage> usage = slave.get()->mock()->usage();
AWAIT_READY(usage);
// Total resources should match the resources from flag.resources.
EXPECT_EQ(Resources(usage->total()),
Resources::parse(flags.resources.get()).get());
}
// This test verifies that the slave should properly show total slave
// resources with checkpointed resources applied.
TEST_F(SlaveTest, CheckpointedResourcesIncludedInUsage)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
TestContainerizer containerizer;
StandaloneMasterDetector detector(master.get()->pid);
slave::Flags flags = CreateSlaveFlags();
flags.resources = "cpus:2;cpus(role1):3;mem:1024;disk:1024;disk(role1):64;"
"ports:[31000-32000]";
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
flags,
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
Clock::pause();
// Wait for slave to be initialized.
Clock::settle();
Resource dynamicReservation = createReservedResource(
"cpus", "1", createDynamicReservationInfo("role1", "principal"));
Resource persistentVolume = createPersistentVolume(
Megabytes(64),
"role1",
"id1",
"path1");
vector<Resource> checkpointedResources =
{dynamicReservation, persistentVolume};
// Add checkpointed resources.
slave.get()->mock()->checkpointResourceState(checkpointedResources, true);
// We expect that the slave will return ResourceUsage with
// total and checkpointed slave resources reported.
Future<ResourceUsage> usage = slave.get()->mock()->usage();
AWAIT_READY(usage);
Resources usageTotalResources(usage->total());
// Reported total field should contain persistent volumes and dynamic
// reservations.
EXPECT_EQ(usageTotalResources.persistentVolumes(), persistentVolume);
EXPECT_TRUE(usageTotalResources.contains(dynamicReservation));
}
// Ensures that the slave correctly handles a framework without
// a pid, which will be the case for HTTP schedulers. In
// particular, executor messages should be routed through the
// master.
TEST_F(SlaveTest, HTTPScheduler)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
EXPECT_CALL(sched, resourceOffers(_, _))
.WillOnce(LaunchTasks(DEFAULT_EXECUTOR_INFO, 1, 2, 1024, "*"))
.WillRepeatedly(Return()); // Ignore subsequent offers.
// Capture the run task message to unset the framework pid.
Future<RunTaskMessage> runTaskMessage =
DROP_PROTOBUF(RunTaskMessage(), master.get()->pid, slave.get()->pid);
driver.start();
AWAIT_READY(runTaskMessage);
EXPECT_CALL(exec, registered(_, _, _, _));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendFrameworkMessage("message"));
// The slave should forward the message through the master.
Future<ExecutorToFrameworkMessage> executorToFrameworkMessage1 =
FUTURE_PROTOBUF(
ExecutorToFrameworkMessage(),
slave.get()->pid,
master.get()->pid);
// The master should then forward the message to the framework.
Future<ExecutorToFrameworkMessage> executorToFrameworkMessage2 =
FUTURE_PROTOBUF(ExecutorToFrameworkMessage(), master.get()->pid, _);
Future<Nothing> frameworkMessage;
EXPECT_CALL(sched, frameworkMessage(&driver, _, _, "message"))
.WillOnce(FutureSatisfy(&frameworkMessage));
// Clear the pid in the run task message so that the slave
// thinks this is an HTTP scheduler.
RunTaskMessage spoofed = runTaskMessage.get();
spoofed.set_pid("");
process::post(master.get()->pid, slave.get()->pid, spoofed);
AWAIT_READY(executorToFrameworkMessage1);
AWAIT_READY(executorToFrameworkMessage2);
AWAIT_READY(frameworkMessage);
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// Ensures that the slave correctly handles a framework upgrading
// to HTTP (going from having a pid, to not having a pid). In
// particular, executor messages should be routed through the
// master.
TEST_F(SlaveTest, HTTPSchedulerLiveUpgrade)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, frameworkInfo, master.get()->pid, DEFAULT_CREDENTIAL);
Future<FrameworkID> frameworkId;
EXPECT_CALL(sched, registered(&driver, _, _))
.WillOnce(FutureArg<1>(&frameworkId));
EXPECT_CALL(sched, resourceOffers(_, _))
.WillOnce(LaunchTasks(DEFAULT_EXECUTOR_INFO, 1, 2, 1024, "*"))
.WillRepeatedly(Return()); // Ignore subsequent offers.
ExecutorDriver* execDriver;
EXPECT_CALL(exec, registered(_, _, _, _))
.WillOnce(SaveArg<0>(&execDriver));
Future<Nothing> launchTask;
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(FutureSatisfy(&launchTask));
driver.start();
AWAIT_READY(frameworkId);
AWAIT_READY(launchTask);
// Set the `FrameworkID` in `FrameworkInfo`.
frameworkInfo.mutable_id()->CopyFrom(frameworkId.get());
// Now spoof a live upgrade of the framework by updating
// the framework information to have an empty pid.
UpdateFrameworkMessage updateFrameworkMessage;
updateFrameworkMessage.mutable_framework_id()->CopyFrom(frameworkId.get());
updateFrameworkMessage.set_pid("");
updateFrameworkMessage.mutable_framework_info()->CopyFrom(frameworkInfo);
process::post(master.get()->pid, slave.get()->pid, updateFrameworkMessage);
// Send a message from the executor; the slave should forward
// the message through the master.
Future<ExecutorToFrameworkMessage> executorToFrameworkMessage1 =
FUTURE_PROTOBUF(
ExecutorToFrameworkMessage(),
slave.get()->pid,
master.get()->pid);
Future<ExecutorToFrameworkMessage> executorToFrameworkMessage2 =
FUTURE_PROTOBUF(ExecutorToFrameworkMessage(), master.get()->pid, _);
Future<Nothing> frameworkMessage;
EXPECT_CALL(sched, frameworkMessage(&driver, _, _, "message"))
.WillOnce(FutureSatisfy(&frameworkMessage));
execDriver->sendFrameworkMessage("message");
AWAIT_READY(executorToFrameworkMessage1);
AWAIT_READY(executorToFrameworkMessage2);
AWAIT_READY(frameworkMessage);
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// Ensures that the slave can restart when there is an empty
// framework pid. Executor messages should go through the
// master (instead of directly to the scheduler!).
TEST_F(SlaveTest, HTTPSchedulerSlaveRestart)
{
Try<Owned<cluster::Master>> master = this->StartMaster();
ASSERT_SOME(master);
slave::Flags flags = this->CreateSlaveFlags();
Fetcher fetcher(flags);
Try<MesosContainerizer*> _containerizer =
MesosContainerizer::create(flags, true, &fetcher);
ASSERT_SOME(_containerizer);
Owned<MesosContainerizer> containerizer(_containerizer.get());
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave =
this->StartSlave(detector.get(), containerizer.get(), flags);
ASSERT_SOME(slave);
// Enable checkpointing for the framework.
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_checkpoint(true);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, frameworkInfo, master.get()->pid, DEFAULT_CREDENTIAL);
FrameworkID frameworkId;
EXPECT_CALL(sched, registered(_, _, _))
.WillOnce(SaveArg<1>(&frameworkId));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
// Capture the executor information.
Future<Message> registerExecutorMessage =
FUTURE_MESSAGE(Eq(RegisterExecutorMessage().GetTypeName()), _, _);
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
SlaveID slaveId = offers.get()[0].slave_id();
// Capture the run task so that we can unset the framework pid.
Future<RunTaskMessage> runTaskMessage =
DROP_PROTOBUF(RunTaskMessage(), master.get()->pid, slave.get()->pid);
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(_, _))
.WillOnce(FutureArg<1>(&status))
.WillRepeatedly(Return()); // Ignore subsequent updates.
TaskInfo task = createTask(offers.get()[0], SLEEP_COMMAND(1000));
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(runTaskMessage);
// Clear the pid in the run task message so that the slave
// thinks this is an HTTP scheduler.
RunTaskMessage spoofedRunTaskMessage = runTaskMessage.get();
spoofedRunTaskMessage.set_pid("");
process::post(master.get()->pid, slave.get()->pid, spoofedRunTaskMessage);
AWAIT_READY(registerExecutorMessage);
RegisterExecutorMessage registerExecutor;
registerExecutor.ParseFromString(registerExecutorMessage->body);
ExecutorID executorId = registerExecutor.executor_id();
UPID executorPid = registerExecutorMessage->from;
AWAIT_READY(status);
EXPECT_EQ(TASK_STARTING, status->state());
// Restart the slave.
slave.get()->terminate();
_containerizer = MesosContainerizer::create(flags, true, &fetcher);
ASSERT_SOME(_containerizer);
containerizer.reset(_containerizer.get());
Future<ReregisterExecutorMessage> reregisterExecutorMessage =
FUTURE_PROTOBUF(ReregisterExecutorMessage(), _, _);
Future<SlaveReregisteredMessage> slaveReregisteredMessage =
FUTURE_PROTOBUF(SlaveReregisteredMessage(), _, _);
// Capture this so that we can unset the framework pid.
Future<UpdateFrameworkMessage> updateFrameworkMessage =
DROP_PROTOBUF(UpdateFrameworkMessage(), _, _);
// Ensure that there will be no reregistration retries from the
// slave resulting in another UpdateFrameworkMessage from master.
Clock::pause();
slave = StartSlave(detector.get(), containerizer.get(), flags);
ASSERT_SOME(slave);
// Let the executor reregister.
AWAIT_READY(reregisterExecutorMessage);
// Ensure the slave considers itself recovered and reregisters.
Clock::settle();
Clock::advance(flags.executor_reregistration_timeout);
Clock::settle();
Clock::advance(flags.registration_backoff_factor);
AWAIT_READY(slaveReregisteredMessage);
AWAIT_READY(updateFrameworkMessage);
// Make sure the slave sees an empty framework pid after recovery.
UpdateFrameworkMessage spoofedUpdateFrameworkMessage =
updateFrameworkMessage.get();
spoofedUpdateFrameworkMessage.set_pid("");
process::post(
master.get()->pid,
slave.get()->pid,
spoofedUpdateFrameworkMessage);
// Spoof a message from the executor, to ensure the slave
// sends it through the master (instead of directly to the
// scheduler driver!).
Future<ExecutorToFrameworkMessage> executorToFrameworkMessage1 =
FUTURE_PROTOBUF(
ExecutorToFrameworkMessage(),
slave.get()->pid,
master.get()->pid);
Future<ExecutorToFrameworkMessage> executorToFrameworkMessage2 =
FUTURE_PROTOBUF(ExecutorToFrameworkMessage(), master.get()->pid, _);
Future<Nothing> frameworkMessage;
EXPECT_CALL(sched, frameworkMessage(&driver, _, _, "message"))
.WillOnce(FutureSatisfy(&frameworkMessage));
ExecutorToFrameworkMessage executorToFrameworkMessage;
executorToFrameworkMessage.mutable_slave_id()->CopyFrom(slaveId);
executorToFrameworkMessage.mutable_framework_id()->CopyFrom(frameworkId);
executorToFrameworkMessage.mutable_executor_id()->CopyFrom(executorId);
executorToFrameworkMessage.set_data("message");
process::post(executorPid, slave.get()->pid, executorToFrameworkMessage);
AWAIT_READY(executorToFrameworkMessage1);
AWAIT_READY(executorToFrameworkMessage2);
AWAIT_READY(frameworkMessage);
driver.stop();
driver.join();
// We must resume the clock to ensure the agent can reap the
// executor after we destroy it.
Clock::resume();
}
// Ensures that if `ExecutorInfo.shutdown_grace_period` is set, it
// overrides the default value from the agent flag, is observed by
// executor, and is enforced by the agent.
TEST_F(SlaveTest, ExecutorShutdownGracePeriod)
{
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
slave::Flags agentFlags = CreateSlaveFlags();
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &containerizer, agentFlags);
ASSERT_SOME(slave);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
// We need framework's ID to shutdown the executor later on.
Future<FrameworkID> frameworkId;
EXPECT_CALL(sched, registered(&driver, _, _))
.WillOnce(FutureArg<1>(&frameworkId));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return());
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
Offer offer = offers.get()[0];
// Customize executor shutdown grace period to be larger than the
// default agent flag value, so that we can check it is respected.
Duration customGracePeriod = agentFlags.executor_shutdown_grace_period * 2;
ExecutorInfo executorInfo(DEFAULT_EXECUTOR_INFO);
executorInfo.mutable_shutdown_grace_period()->set_nanoseconds(
customGracePeriod.ns());
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value("2");
task.mutable_slave_id()->MergeFrom(offer.slave_id());
task.mutable_resources()->MergeFrom(offer.resources());
task.mutable_executor()->MergeFrom(executorInfo);
Future<TaskStatus> statusRunning;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusRunning));
EXPECT_CALL(exec, registered(_, _, _, _));
Future<TaskInfo> receivedTask;
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(DoAll(SendStatusUpdateFromTask(TASK_RUNNING),
FutureArg<1>(&receivedTask)));
driver.launchTasks(offer.id(), {task});
AWAIT_READY(statusRunning);
EXPECT_EQ(TASK_RUNNING, statusRunning->state());
EXPECT_EQ(customGracePeriod.ns(),
receivedTask->executor().shutdown_grace_period().nanoseconds());
// If executor is asked to shutdown but fails to do so within the grace
// shutdown period, the shutdown is enforced by the agent. The agent
// adjusts its timeout according to `ExecutorInfo.shutdown_grace_period`.
//
// NOTE: Executors relying on the executor driver have a built-in suicide
// mechanism (`ShutdownProcess`), that kills the OS process where the
// executor is running after the grace period ends. This mechanism is
// disabled in tests, hence we do not observe crashes induced by this test.
// The test containerizer only accepts "local" executors and it considers
// them "terminated" only once destroy is called.
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1))
.WillOnce(Return());
// Once the grace period ends, the agent forcibly shuts down the executor.
Future<Nothing> executorShutdownTimeout =
FUTURE_DISPATCH(slave.get()->pid, &Slave::shutdownExecutorTimeout);
Future<TaskStatus> statusFailed;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusFailed));
Future<ExecutorID> lostExecutorId;
EXPECT_CALL(sched, executorLost(&driver, DEFAULT_EXECUTOR_ID, _, _))
.WillOnce(FutureArg<1>(&lostExecutorId));
// Ask executor to shutdown. There is no support in the scheduler
// driver for shutting down executors, hence we have to spoof it.
AWAIT_READY(frameworkId);
ShutdownExecutorMessage shutdownMessage;
shutdownMessage.mutable_executor_id()->CopyFrom(DEFAULT_EXECUTOR_ID);
shutdownMessage.mutable_framework_id()->CopyFrom(frameworkId.get());
post(master.get()->pid, slave.get()->pid, shutdownMessage);
// Ensure the `ShutdownExecutorMessage` message is
// received by the agent before we start the timer.
Clock::pause();
Clock::settle();
Clock::advance(agentFlags.executor_shutdown_grace_period);
Clock::settle();
// The executor shutdown timeout should not have fired, since the
// `ExecutorInfo` contains a grace period larger than the agent flag.
EXPECT_TRUE(executorShutdownTimeout.isPending());
// Trigger the shutdown grace period from the `ExecutorInfo`
// (note that is is 2x the agent flag).
Clock::advance(agentFlags.executor_shutdown_grace_period);
AWAIT_READY(executorShutdownTimeout);
AWAIT_READY(statusFailed);
EXPECT_EQ(TASK_FAILED, statusFailed->state());
EXPECT_EQ(TaskStatus::REASON_EXECUTOR_TERMINATED,
statusFailed->reason());
AWAIT_EXPECT_EQ(DEFAULT_EXECUTOR_ID, lostExecutorId);
Clock::resume();
driver.stop();
driver.join();
}
// This test verifies that the agent can forward a task group to an
// executor atomically via the `LAUNCH_GROUP` event.
TEST_F(SlaveTest, RunTaskGroup)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
auto executor = std::make_shared<v1::MockHTTPExecutor>();
Resources resources =
Resources::parse("cpus:0.1;mem:32;disk:32").get();
ExecutorInfo executorInfo = DEFAULT_EXECUTOR_INFO;
executorInfo.set_type(ExecutorInfo::CUSTOM);
executorInfo.mutable_resources()->CopyFrom(resources);
const ExecutorID& executorId = executorInfo.executor_id();
TestContainerizer containerizer(executorId, executor);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(v1::DEFAULT_FRAMEWORK_INFO);
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(devolve(frameworkId));
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
EXPECT_CALL(*executor, connected(_))
.WillOnce(v1::executor::SendSubscribe(frameworkId, evolve(executorId)));
EXPECT_CALL(*executor, subscribed(_, _));
EXPECT_CALL(*executor, launch(_, _))
.Times(0);
Future<v1::executor::Event::LaunchGroup> launchGroupEvent;
EXPECT_CALL(*executor, launchGroup(_, _))
.WillOnce(FutureArg<1>(&launchGroupEvent));
const v1::Offer& offer = offers->offers(0);
const SlaveID slaveId = devolve(offer.agent_id());
v1::TaskInfo taskInfo1 =
evolve(createTask(slaveId, resources, ""));
v1::TaskInfo taskInfo2 =
evolve(createTask(slaveId, resources, ""));
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(taskInfo1);
taskGroup.add_tasks()->CopyFrom(taskInfo2);
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACCEPT);
Call::Accept* accept = call.mutable_accept();
accept->add_offer_ids()->CopyFrom(offer.id());
v1::Offer::Operation* operation = accept->add_operations();
operation->set_type(v1::Offer::Operation::LAUNCH_GROUP);
v1::Offer::Operation::LaunchGroup* launchGroup =
operation->mutable_launch_group();
launchGroup->mutable_executor()->CopyFrom(evolve(executorInfo));
launchGroup->mutable_task_group()->CopyFrom(taskGroup);
mesos.send(call);
}
AWAIT_READY(launchGroupEvent);
ASSERT_EQ(2, launchGroupEvent->task_group().tasks().size());
const hashset<v1::TaskID> tasks{taskInfo1.task_id(), taskInfo2.task_id()};
const hashset<v1::TaskID> launchedTasks{
launchGroupEvent->task_group().tasks(0).task_id(),
launchGroupEvent->task_group().tasks(1).task_id()};
EXPECT_EQ(tasks, launchedTasks);
EXPECT_CALL(*executor, shutdown(_))
.Times(AtMost(1));
}
// This test verifies that TASK_FAILED updates are sent correctly for all the
// tasks in a task group when secret generation fails.
TEST_F(SlaveTest, RunTaskGroupFailedSecretGeneration)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
auto executor = std::make_shared<v1::MockHTTPExecutor>();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::ExecutorInfo executorInfo = v1::DEFAULT_EXECUTOR_INFO;
executorInfo.set_type(v1::ExecutorInfo::CUSTOM);
executorInfo.mutable_resources()->CopyFrom(resources);
const v1::ExecutorID& executorId = executorInfo.executor_id();
TestContainerizer containerizer(devolve(executorId), executor);
StandaloneMasterDetector detector(master.get()->pid);
Owned<MockSecretGenerator> secretGenerator(new MockSecretGenerator());
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
secretGenerator.get(),
None(),
None(),
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(v1::DEFAULT_FRAMEWORK_INFO);
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(frameworkId);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
v1::TaskInfo taskInfo1 = v1::createTask(agentId, resources, "");
v1::TaskInfo taskInfo2 = v1::createTask(agentId, resources, "");
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(taskInfo1);
taskGroup.add_tasks()->CopyFrom(taskInfo2);
const hashset<v1::TaskID> tasks{taskInfo1.task_id(), taskInfo2.task_id()};
// The tasks will fail to launch because the executor secret generation fails.
const string failureMessage = "Mock secret generator failed";
EXPECT_CALL(*secretGenerator, generate(_))
.WillOnce(Return(Failure(failureMessage)));
EXPECT_CALL(*executor, connected(_))
.Times(0);
EXPECT_CALL(*executor, subscribed(_, _))
.Times(0);
EXPECT_CALL(*executor, shutdown(_))
.Times(0);
EXPECT_CALL(*executor, launchGroup(_, _))
.Times(0);
EXPECT_CALL(*executor, launch(_, _))
.Times(0);
EXPECT_CALL(*slave.get()->mock(), executorTerminated(_, _, _))
.WillOnce(Invoke(slave.get()->mock(),
&MockSlave::unmocked_executorTerminated));
Future<v1::scheduler::Event::Update> update1;
Future<v1::scheduler::Event::Update> update2;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(FutureArg<1>(&update1))
.WillOnce(FutureArg<1>(&update2));
Future<Nothing> failure;
EXPECT_CALL(*scheduler, failure(_, _))
.WillOnce(FutureSatisfy(&failure));
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACCEPT);
Call::Accept* accept = call.mutable_accept();
accept->add_offer_ids()->CopyFrom(offer.id());
v1::Offer::Operation* operation = accept->add_operations();
operation->set_type(v1::Offer::Operation::LAUNCH_GROUP);
v1::Offer::Operation::LaunchGroup* launchGroup =
operation->mutable_launch_group();
launchGroup->mutable_executor()->CopyFrom(executorInfo);
launchGroup->mutable_task_group()->CopyFrom(taskGroup);
mesos.send(call);
}
AWAIT_READY(update1);
AWAIT_READY(update2);
AWAIT_READY(failure);
const hashset<v1::TaskID> failedTasks{
update1->status().task_id(), update2->status().task_id()};
ASSERT_EQ(v1::TASK_FAILED, update1->status().state());
ASSERT_EQ(v1::TASK_FAILED, update2->status().state());
EXPECT_TRUE(strings::contains(update1->status().message(), failureMessage));
EXPECT_TRUE(strings::contains(update2->status().message(), failureMessage));
ASSERT_EQ(tasks, failedTasks);
// Since this is the only task group for this framework, the
// framework should be removed after secret generation fails.
Future<Nothing> removeFramework;
EXPECT_CALL(*slave.get()->mock(), removeFramework(_))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_removeFramework),
FutureSatisfy(&removeFramework)));
// Acknowledge the status updates so that the agent will remove the framework.
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACKNOWLEDGE);
Call::Acknowledge* acknowledge = call.mutable_acknowledge();
acknowledge->mutable_task_id()->CopyFrom(update1->status().task_id());
acknowledge->mutable_agent_id()->CopyFrom(offer.agent_id());
acknowledge->set_uuid(update1->status().uuid());
mesos.send(call);
}
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACKNOWLEDGE);
Call::Acknowledge* acknowledge = call.mutable_acknowledge();
acknowledge->mutable_task_id()->CopyFrom(update2->status().task_id());
acknowledge->mutable_agent_id()->CopyFrom(offer.agent_id());
acknowledge->set_uuid(update2->status().uuid());
mesos.send(call);
}
AWAIT_READY(removeFramework);
}
// This test verifies that TASK_FAILED updates are sent correctly for all the
// tasks in a task group when the secret generator returns an invalid secret.
TEST_F(SlaveTest, RunTaskGroupInvalidExecutorSecret)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
auto executor = std::make_shared<v1::MockHTTPExecutor>();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::ExecutorInfo executorInfo = v1::DEFAULT_EXECUTOR_INFO;
executorInfo.set_type(v1::ExecutorInfo::CUSTOM);
executorInfo.mutable_resources()->CopyFrom(resources);
const v1::ExecutorID& executorId = executorInfo.executor_id();
TestContainerizer containerizer(devolve(executorId), executor);
StandaloneMasterDetector detector(master.get()->pid);
Owned<MockSecretGenerator> secretGenerator(new MockSecretGenerator());
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
secretGenerator.get(),
None(),
None(),
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(v1::DEFAULT_FRAMEWORK_INFO);
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(frameworkId);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
v1::TaskInfo taskInfo1 = v1::createTask(agentId, resources, "");
v1::TaskInfo taskInfo2 = v1::createTask(agentId, resources, "");
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(taskInfo1);
taskGroup.add_tasks()->CopyFrom(taskInfo2);
const hashset<v1::TaskID> tasks{taskInfo1.task_id(), taskInfo2.task_id()};
// The tasks will fail to launch because the executor secret is invalid
// (VALUE type secrets must not have the `reference` member set).
Secret authenticationToken;
authenticationToken.set_type(Secret::VALUE);
authenticationToken.mutable_reference()->set_name("secret_name");
authenticationToken.mutable_reference()->set_key("secret_key");
EXPECT_CALL(*secretGenerator, generate(_))
.WillOnce(Return(authenticationToken));
EXPECT_CALL(*executor, connected(_))
.Times(0);
EXPECT_CALL(*executor, subscribed(_, _))
.Times(0);
EXPECT_CALL(*executor, shutdown(_))
.Times(0);
EXPECT_CALL(*executor, launchGroup(_, _))
.Times(0);
EXPECT_CALL(*executor, launch(_, _))
.Times(0);
EXPECT_CALL(*slave.get()->mock(), executorTerminated(_, _, _))
.WillOnce(Invoke(slave.get()->mock(),
&MockSlave::unmocked_executorTerminated));
Future<v1::scheduler::Event::Update> update1;
Future<v1::scheduler::Event::Update> update2;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(FutureArg<1>(&update1))
.WillOnce(FutureArg<1>(&update2));
Future<Nothing> failure;
EXPECT_CALL(*scheduler, failure(_, _))
.WillOnce(FutureSatisfy(&failure));
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACCEPT);
Call::Accept* accept = call.mutable_accept();
accept->add_offer_ids()->CopyFrom(offer.id());
v1::Offer::Operation* operation = accept->add_operations();
operation->set_type(v1::Offer::Operation::LAUNCH_GROUP);
v1::Offer::Operation::LaunchGroup* launchGroup =
operation->mutable_launch_group();
launchGroup->mutable_executor()->CopyFrom(executorInfo);
launchGroup->mutable_task_group()->CopyFrom(taskGroup);
mesos.send(call);
}
AWAIT_READY(update1);
AWAIT_READY(update2);
AWAIT_READY(failure);
const hashset<v1::TaskID> failedTasks{
update1->status().task_id(), update2->status().task_id()};
ASSERT_EQ(v1::TASK_FAILED, update1->status().state());
ASSERT_EQ(v1::TASK_FAILED, update2->status().state());
const string failureMessage =
"Secret of type VALUE must have the 'value' field set";
EXPECT_TRUE(strings::contains(update1->status().message(), failureMessage));
EXPECT_TRUE(strings::contains(update2->status().message(), failureMessage));
ASSERT_EQ(tasks, failedTasks);
// Since this is the only task group for this framework, the
// framework should be removed after secret generation fails.
Future<Nothing> removeFramework;
EXPECT_CALL(*slave.get()->mock(), removeFramework(_))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_removeFramework),
FutureSatisfy(&removeFramework)));
// Acknowledge the status updates so that the agent will remove the framework.
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACKNOWLEDGE);
Call::Acknowledge* acknowledge = call.mutable_acknowledge();
acknowledge->mutable_task_id()->CopyFrom(update1->status().task_id());
acknowledge->mutable_agent_id()->CopyFrom(offer.agent_id());
acknowledge->set_uuid(update1->status().uuid());
mesos.send(call);
}
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACKNOWLEDGE);
Call::Acknowledge* acknowledge = call.mutable_acknowledge();
acknowledge->mutable_task_id()->CopyFrom(update2->status().task_id());
acknowledge->mutable_agent_id()->CopyFrom(offer.agent_id());
acknowledge->set_uuid(update2->status().uuid());
mesos.send(call);
}
AWAIT_READY(removeFramework);
}
// This test verifies that TASK_FAILED updates are sent correctly for all the
// tasks in a task group when the secret generator returns a REFERENCE type
// secret. Only VALUE type secrets are supported at this time.
TEST_F(SlaveTest, RunTaskGroupReferenceTypeSecret)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
auto executor = std::make_shared<v1::MockHTTPExecutor>();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::ExecutorInfo executorInfo = v1::DEFAULT_EXECUTOR_INFO;
executorInfo.set_type(v1::ExecutorInfo::CUSTOM);
executorInfo.mutable_resources()->CopyFrom(resources);
const v1::ExecutorID& executorId = executorInfo.executor_id();
TestContainerizer containerizer(devolve(executorId), executor);
StandaloneMasterDetector detector(master.get()->pid);
Owned<MockSecretGenerator> secretGenerator(new MockSecretGenerator());
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
secretGenerator.get(),
None(),
None(),
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(v1::DEFAULT_FRAMEWORK_INFO);
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(frameworkId);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
v1::TaskInfo taskInfo1 = v1::createTask(agentId, resources, "");
v1::TaskInfo taskInfo2 = v1::createTask(agentId, resources, "");
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(taskInfo1);
taskGroup.add_tasks()->CopyFrom(taskInfo2);
const hashset<v1::TaskID> tasks{taskInfo1.task_id(), taskInfo2.task_id()};
// The tasks will fail to launch because the executor secret is invalid
// (only VALUE type secrets are supported at this time).
Secret authenticationToken;
authenticationToken.set_type(Secret::REFERENCE);
authenticationToken.mutable_reference()->set_name("secret_name");
authenticationToken.mutable_reference()->set_key("secret_key");
EXPECT_CALL(*secretGenerator, generate(_))
.WillOnce(Return(authenticationToken));
EXPECT_CALL(*executor, connected(_))
.Times(0);
EXPECT_CALL(*executor, subscribed(_, _))
.Times(0);
EXPECT_CALL(*executor, shutdown(_))
.Times(0);
EXPECT_CALL(*executor, launchGroup(_, _))
.Times(0);
EXPECT_CALL(*executor, launch(_, _))
.Times(0);
EXPECT_CALL(*slave.get()->mock(), executorTerminated(_, _, _))
.WillOnce(Invoke(slave.get()->mock(),
&MockSlave::unmocked_executorTerminated));
Future<v1::scheduler::Event::Update> update1;
Future<v1::scheduler::Event::Update> update2;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(FutureArg<1>(&update1))
.WillOnce(FutureArg<1>(&update2));
Future<Nothing> failure;
EXPECT_CALL(*scheduler, failure(_, _))
.WillOnce(FutureSatisfy(&failure));
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACCEPT);
Call::Accept* accept = call.mutable_accept();
accept->add_offer_ids()->CopyFrom(offer.id());
v1::Offer::Operation* operation = accept->add_operations();
operation->set_type(v1::Offer::Operation::LAUNCH_GROUP);
v1::Offer::Operation::LaunchGroup* launchGroup =
operation->mutable_launch_group();
launchGroup->mutable_executor()->CopyFrom(executorInfo);
launchGroup->mutable_task_group()->CopyFrom(taskGroup);
mesos.send(call);
}
AWAIT_READY(update1);
AWAIT_READY(update2);
AWAIT_READY(failure);
const hashset<v1::TaskID> failedTasks{
update1->status().task_id(), update2->status().task_id()};
ASSERT_EQ(v1::TASK_FAILED, update1->status().state());
ASSERT_EQ(v1::TASK_FAILED, update2->status().state());
const string failureMessage =
"Expecting generated secret to be of VALUE type instead of REFERENCE type";
EXPECT_TRUE(strings::contains(update1->status().message(), failureMessage));
EXPECT_TRUE(strings::contains(update2->status().message(), failureMessage));
ASSERT_EQ(tasks, failedTasks);
// Since this is the only task group for this framework, the
// framework should be removed after secret generation fails.
Future<Nothing> removeFramework;
EXPECT_CALL(*slave.get()->mock(), removeFramework(_))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_removeFramework),
FutureSatisfy(&removeFramework)));
// Acknowledge the status updates so that the agent will remove the framework.
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACKNOWLEDGE);
Call::Acknowledge* acknowledge = call.mutable_acknowledge();
acknowledge->mutable_task_id()->CopyFrom(update1->status().task_id());
acknowledge->mutable_agent_id()->CopyFrom(offer.agent_id());
acknowledge->set_uuid(update1->status().uuid());
mesos.send(call);
}
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACKNOWLEDGE);
Call::Acknowledge* acknowledge = call.mutable_acknowledge();
acknowledge->mutable_task_id()->CopyFrom(update2->status().task_id());
acknowledge->mutable_agent_id()->CopyFrom(offer.agent_id());
acknowledge->set_uuid(update2->status().uuid());
mesos.send(call);
}
AWAIT_READY(removeFramework);
}
// This test verifies that TASK_FAILED updates and an executor FAILURE message
// are sent correctly when the secret generator returns the executor secret
// after the scheduler has shutdown the executor.
TEST_F(SlaveTest, RunTaskGroupGenerateSecretAfterShutdown)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
auto executor = std::make_shared<v1::MockHTTPExecutor>();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::ExecutorInfo executorInfo = v1::DEFAULT_EXECUTOR_INFO;
executorInfo.set_type(v1::ExecutorInfo::CUSTOM);
executorInfo.mutable_resources()->CopyFrom(resources);
const v1::ExecutorID& executorId = executorInfo.executor_id();
TestContainerizer containerizer(devolve(executorId), executor);
StandaloneMasterDetector detector(master.get()->pid);
Owned<MockSecretGenerator> secretGenerator(new MockSecretGenerator());
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
secretGenerator.get(),
None(),
None(),
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(v1::DEFAULT_FRAMEWORK_INFO);
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(frameworkId);
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
v1::TaskInfo taskInfo1 = v1::createTask(agentId, resources, "");
v1::TaskInfo taskInfo2 = v1::createTask(agentId, resources, "");
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(taskInfo1);
taskGroup.add_tasks()->CopyFrom(taskInfo2);
const hashset<v1::TaskID> tasks{taskInfo1.task_id(), taskInfo2.task_id()};
// We return this promise's future so that we can delay its fulfillment
// until after the scheduler has shutdown the executor.
Promise<Secret> secret;
Future<Nothing> generate;
EXPECT_CALL(*secretGenerator, generate(_))
.WillOnce(DoAll(FutureSatisfy(&generate),
Return(secret.future())));
EXPECT_CALL(*executor, connected(_))
.Times(0);
EXPECT_CALL(*executor, subscribed(_, _))
.Times(0);
EXPECT_CALL(*executor, shutdown(_))
.Times(0);
EXPECT_CALL(*executor, launchGroup(_, _))
.Times(0);
EXPECT_CALL(*executor, launch(_, _))
.Times(0);
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACCEPT);
Call::Accept* accept = call.mutable_accept();
accept->add_offer_ids()->CopyFrom(offer.id());
v1::Offer::Operation* operation = accept->add_operations();
operation->set_type(v1::Offer::Operation::LAUNCH_GROUP);
v1::Offer::Operation::LaunchGroup* launchGroup =
operation->mutable_launch_group();
launchGroup->mutable_executor()->CopyFrom(executorInfo);
launchGroup->mutable_task_group()->CopyFrom(taskGroup);
mesos.send(call);
}
AWAIT_READY(generate);
Future<Nothing> shutdownExecutor;
EXPECT_CALL(*slave.get()->mock(), shutdownExecutor(_, _, _))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_shutdownExecutor),
FutureSatisfy(&shutdownExecutor)));
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::SHUTDOWN);
Call::Shutdown* shutdown = call.mutable_shutdown();
shutdown->mutable_executor_id()->CopyFrom(executorId);
shutdown->mutable_agent_id()->CopyFrom(offer.agent_id());
mesos.send(call);
}
AWAIT_READY(shutdownExecutor);
Future<v1::scheduler::Event::Update> update1;
Future<v1::scheduler::Event::Update> update2;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(FutureArg<1>(&update1))
.WillOnce(FutureArg<1>(&update2));
Future<Nothing> failure;
EXPECT_CALL(*scheduler, failure(_, _))
.WillOnce(FutureSatisfy(&failure));
EXPECT_CALL(*slave.get()->mock(), executorTerminated(_, _, _))
.WillOnce(Invoke(slave.get()->mock(),
&MockSlave::unmocked_executorTerminated));
// The tasks will fail to launch because the executor has been shutdown.
Secret authenticationToken;
authenticationToken.set_type(Secret::VALUE);
authenticationToken.mutable_value()->set_data("secret_data");
secret.set(authenticationToken);
AWAIT_READY(update1);
AWAIT_READY(update2);
AWAIT_READY(failure);
const hashset<v1::TaskID> failedTasks{
update1->status().task_id(), update2->status().task_id()};
ASSERT_EQ(v1::TASK_FAILED, update1->status().state());
ASSERT_EQ(v1::TASK_FAILED, update2->status().state());
const string failureMessage = "Executor terminating";
EXPECT_TRUE(strings::contains(update1->status().message(), failureMessage));
EXPECT_TRUE(strings::contains(update2->status().message(), failureMessage));
ASSERT_EQ(tasks, failedTasks);
// Since this is the only task group for this framework, the
// framework should be removed after secret generation fails.
Future<Nothing> removeFramework;
EXPECT_CALL(*slave.get()->mock(), removeFramework(_))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_removeFramework),
FutureSatisfy(&removeFramework)));
// Acknowledge the status updates so that the agent will remove the framework.
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACKNOWLEDGE);
Call::Acknowledge* acknowledge = call.mutable_acknowledge();
acknowledge->mutable_task_id()->CopyFrom(update1->status().task_id());
acknowledge->mutable_agent_id()->CopyFrom(offer.agent_id());
acknowledge->set_uuid(update1->status().uuid());
mesos.send(call);
}
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACKNOWLEDGE);
Call::Acknowledge* acknowledge = call.mutable_acknowledge();
acknowledge->mutable_task_id()->CopyFrom(update2->status().task_id());
acknowledge->mutable_agent_id()->CopyFrom(offer.agent_id());
acknowledge->set_uuid(update2->status().uuid());
mesos.send(call);
}
AWAIT_READY(removeFramework);
}
#ifdef USE_SSL_SOCKET
// This test verifies that a default executor which is launched when secret
// generation is enabled and HTTP executor authentication is not required will
// be able to re-subscribe successfully when the agent is restarted with
// required HTTP executor authentication.
//
TEST_F(SlaveTest, RestartSlaveRequireExecutorAuthentication)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
slave::Flags flags = CreateSlaveFlags();
flags.authenticate_http_executors = false;
flags.authenticate_http_readwrite = false;
Owned<MasterDetector> detector = master.get()->createDetector();
// Start the agent with a static process ID. This allows the executor to
// reconnect with the agent upon a process restart.
const string id("agent");
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), id, flags);
ASSERT_SOME(slave);
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
v1::FrameworkInfo frameworkInfo = v1::DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_checkpoint(true);
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(frameworkInfo);
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
Future<v1::scheduler::Event::Update> updateStarting;
Future<v1::scheduler::Event::Update> updateRunning;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(
DoAll(
FutureArg<1>(&updateStarting),
v1::scheduler::SendAcknowledge(frameworkId, agentId)))
.WillOnce(
DoAll(
FutureArg<1>(&updateRunning),
v1::scheduler::SendAcknowledge(frameworkId, agentId)))
.WillRepeatedly(Return()); // Ignore subsequent updates.
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
// Create a task which should run indefinitely.
const string command =
#ifdef __WINDOWS__
"more";
#else
"cat";
#endif // __WINDOWS__
v1::TaskInfo taskInfo = v1::createTask(agentId, resources, command);
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(taskInfo);
v1::ExecutorInfo executorInfo = v1::DEFAULT_EXECUTOR_INFO;
executorInfo.clear_command();
executorInfo.mutable_framework_id()->CopyFrom(subscribed->framework_id());
executorInfo.set_type(v1::ExecutorInfo::DEFAULT);
executorInfo.mutable_resources()->CopyFrom(resources);
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACCEPT);
Call::Accept* accept = call.mutable_accept();
accept->add_offer_ids()->CopyFrom(offer.id());
v1::Offer::Operation* operation = accept->add_operations();
operation->set_type(v1::Offer::Operation::LAUNCH_GROUP);
v1::Offer::Operation::LaunchGroup* launchGroup =
operation->mutable_launch_group();
launchGroup->mutable_executor()->CopyFrom(executorInfo);
launchGroup->mutable_task_group()->CopyFrom(taskGroup);
mesos.send(call);
}
AWAIT_READY(updateStarting);
ASSERT_EQ(v1::TASK_STARTING, updateStarting->status().state());
ASSERT_EQ(taskInfo.task_id(), updateStarting->status().task_id());
AWAIT_READY(updateRunning);
ASSERT_EQ(v1::TASK_RUNNING, updateRunning->status().state());
ASSERT_EQ(taskInfo.task_id(), updateRunning->status().task_id());
// Restart the agent.
slave.get()->terminate();
// Enable authentication.
flags.authenticate_http_executors = true;
flags.authenticate_http_readwrite = true;
// Confirm that the executor does not fail.
EXPECT_CALL(*scheduler, failure(_, _))
.Times(0);
Future<Nothing> __recover =
FUTURE_DISPATCH(slave.get()->pid, &Slave::__recover);
slave = StartSlave(detector.get(), id, flags);
ASSERT_SOME(slave);
AWAIT_READY(__recover);
Future<Response> response = process::http::get(
slave.get()->pid,
"containers",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_READY(response);
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
Try<JSON::Value> value = JSON::parse(response->body);
ASSERT_SOME(value);
Try<JSON::Value> expected = JSON::parse(
"[{"
"\"executor_id\":\"" + stringify(executorInfo.executor_id()) + "\""
"}]");
ASSERT_SOME(expected);
EXPECT_TRUE(value->contains(expected.get()));
// Settle the clock to ensure that an executor failure would be detected.
Clock::pause();
Clock::settle();
Clock::resume();
}
#endif // USE_SSL_SOCKET
// This test ensures that a `killTask()` can happen between `runTask()`
// and `_run()` and then gets "handled properly" for a task group.
// This should result in TASK_KILLED updates for all the tasks in the
// task group.
TEST_F(SlaveTest, KillTaskGroupBetweenRunTaskParts)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
auto executor = std::make_shared<v1::MockHTTPExecutor>();
Resources resources =
Resources::parse("cpus:0.1;mem:32;disk:32").get();
ExecutorInfo executorInfo = DEFAULT_EXECUTOR_INFO;
executorInfo.set_type(ExecutorInfo::CUSTOM);
executorInfo.mutable_resources()->CopyFrom(resources);
const ExecutorID& executorId = executorInfo.executor_id();
TestContainerizer containerizer(executorId, executor);
StandaloneMasterDetector detector(master.get()->pid);
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&containerizer,
None(),
true);
ASSERT_SOME(slave);
ASSERT_NE(nullptr, slave.get()->mock());
slave.get()->start();
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return());
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
EXPECT_CALL(*scheduler, failure(_, _))
.Times(AtMost(1));
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(v1::DEFAULT_FRAMEWORK_INFO);
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(devolve(frameworkId));
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
EXPECT_CALL(*executor, connected(_))
.Times(0);
EXPECT_CALL(*executor, subscribed(_, _))
.Times(0);
EXPECT_CALL(*executor, shutdown(_))
.Times(0);
EXPECT_CALL(*executor, launchGroup(_, _))
.Times(0);
EXPECT_CALL(*executor, launch(_, _))
.Times(0);
Future<v1::scheduler::Event::Update> update1;
Future<v1::scheduler::Event::Update> update2;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(FutureArg<1>(&update1))
.WillOnce(FutureArg<1>(&update2))
.WillRepeatedly(Return());
EXPECT_CALL(*slave.get()->mock(), runTaskGroup(_, _, _, _, _, _))
.WillOnce(Invoke(slave.get()->mock(),
&MockSlave::unmocked_runTaskGroup));
// Saved arguments from `Slave::_run()`.
FrameworkInfo frameworkInfo;
ExecutorInfo executorInfo_;
Option<TaskGroupInfo> taskGroup_;
Option<TaskInfo> task_;
vector<ResourceVersionUUID> resourceVersionUuids;
Option<bool> launchExecutor;
// Skip what `Slave::_run()` normally does, save its arguments for
// later, return a pending future to pause the original continuation,
// till reaching the critical moment when to kill the task in the future.
Promise<Nothing> promise;
Future<Nothing> _run;
EXPECT_CALL(*slave.get()->mock(), _run(_, _, _, _, _, _))
.WillOnce(DoAll(FutureSatisfy(&_run),
SaveArg<0>(&frameworkInfo),
SaveArg<1>(&executorInfo_),
SaveArg<2>(&task_),
SaveArg<3>(&taskGroup_),
SaveArg<4>(&resourceVersionUuids),
SaveArg<5>(&launchExecutor),
Return(promise.future())));
const v1::Offer& offer = offers->offers(0);
const SlaveID slaveId = devolve(offer.agent_id());
v1::TaskInfo taskInfo1 =
evolve(createTask(slaveId, resources, ""));
v1::TaskInfo taskInfo2 =
evolve(createTask(slaveId, resources, ""));
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(taskInfo1);
taskGroup.add_tasks()->CopyFrom(taskInfo2);
const hashset<v1::TaskID> tasks{taskInfo1.task_id(), taskInfo2.task_id()};
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACCEPT);
Call::Accept* accept = call.mutable_accept();
accept->add_offer_ids()->CopyFrom(offer.id());
v1::Offer::Operation* operation = accept->add_operations();
operation->set_type(v1::Offer::Operation::LAUNCH_GROUP);
v1::Offer::Operation::LaunchGroup* launchGroup =
operation->mutable_launch_group();
launchGroup->mutable_executor()->CopyFrom(evolve(executorInfo));
launchGroup->mutable_task_group()->CopyFrom(taskGroup);
mesos.send(call);
}
AWAIT_READY(_run);
Future<Nothing> killTask;
EXPECT_CALL(*slave.get()->mock(), killTask(_, _))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_killTask),
FutureSatisfy(&killTask)));
// Since this is the only task group for this framework, the
// framework should get removed when the task is killed.
Future<Nothing> removeFramework;
EXPECT_CALL(*slave.get()->mock(), removeFramework(_))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_removeFramework),
FutureSatisfy(&removeFramework)));
mesos.send(
v1::createCallKill(frameworkId, taskInfo1.task_id(), offer.agent_id()));
AWAIT_READY(killTask);
AWAIT_READY(removeFramework);
Future<Nothing> unmocked__run = process::dispatch(slave.get()->pid, [=] {
return slave.get()->mock()->unmocked__run(
frameworkInfo,
executorInfo_,
task_,
taskGroup_,
resourceVersionUuids,
launchExecutor);
});
// Resume the original continuation once `unmocked__run` is complete.
promise.associate(unmocked__run);
AWAIT_READY(update1);
AWAIT_READY(update2);
AWAIT(unmocked__run);
const hashset<v1::TaskID> killedTasks{
update1->status().task_id(), update2->status().task_id()};
EXPECT_EQ(v1::TASK_KILLED, update1->status().state());
EXPECT_EQ(v1::TASK_KILLED, update2->status().state());
EXPECT_EQ(tasks, killedTasks);
}
// This test verifies that the agent correctly populates the
// command info for default executor.
//
// TODO(andschwa): Enable when user impersonation works on Windows.
TEST_F_TEMP_DISABLED_ON_WINDOWS(SlaveTest, DefaultExecutorCommandInfo)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
auto executor = std::make_shared<v1::MockHTTPExecutor>();
Resources resources =
Resources::parse("cpus:0.1;mem:32;disk:32").get();
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
ExecutorInfo executorInfo;
executorInfo.set_type(ExecutorInfo::DEFAULT);
executorInfo.mutable_executor_id()->CopyFrom(DEFAULT_EXECUTOR_ID);
executorInfo.mutable_resources()->CopyFrom(resources);
const ExecutorID& executorId = executorInfo.executor_id();
TestContainerizer containerizer(executorId, executor);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return());
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(evolve(frameworkInfo));
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(devolve(frameworkId));
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
Future<ContainerConfig> containerConfig;
EXPECT_CALL(containerizer, launch(_, _, _, _))
.WillOnce(DoAll(FutureArg<1>(&containerConfig),
Return(Future<Containerizer::LaunchResult>())));
const v1::Offer& offer = offers->offers(0);
const SlaveID slaveId = devolve(offer.agent_id());
v1::TaskInfo taskInfo =
evolve(createTask(slaveId, resources, ""));
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(taskInfo);
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACCEPT);
Call::Accept* accept = call.mutable_accept();
accept->add_offer_ids()->CopyFrom(offer.id());
v1::Offer::Operation* operation = accept->add_operations();
operation->set_type(v1::Offer::Operation::LAUNCH_GROUP);
v1::Offer::Operation::LaunchGroup* launchGroup =
operation->mutable_launch_group();
launchGroup->mutable_executor()->CopyFrom(evolve(executorInfo));
launchGroup->mutable_task_group()->CopyFrom(taskGroup);
mesos.send(call);
}
AWAIT_READY(containerConfig);
// TODO(anand): Add a `strings::contains()` check to ensure
// `MESOS_DEFAULT_EXECUTOR` is present in the command when
// we add the executable for default executor.
ASSERT_TRUE(containerConfig->has_executor_info());
ASSERT_TRUE(containerConfig->executor_info().has_command());
EXPECT_EQ(
frameworkInfo.user(),
containerConfig->executor_info().command().user());
}
// This test verifies that the agent correctly populates the resources
// for default executor. This is a regression test for MESOS-9925.
TEST_F(SlaveTest, DefaultExecutorResources)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
auto executor = std::make_shared<v1::MockHTTPExecutor>();
Resources resources =
Resources::parse("cpus:0.1;mem:32;disk:32").get();
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
ExecutorInfo executorInfo;
executorInfo.set_type(ExecutorInfo::DEFAULT);
executorInfo.mutable_executor_id()->CopyFrom(DEFAULT_EXECUTOR_ID);
executorInfo.mutable_resources()->CopyFrom(resources);
const ExecutorID& executorId = executorInfo.executor_id();
TestContainerizer containerizer(executorId, executor);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return());
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(evolve(frameworkInfo));
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(devolve(frameworkId));
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
Future<ContainerConfig> containerConfig;
EXPECT_CALL(containerizer, launch(_, _, _, _))
.WillOnce(DoAll(FutureArg<1>(&containerConfig),
Return(Future<Containerizer::LaunchResult>())));
const v1::Offer& offer = offers->offers(0);
const SlaveID slaveId = devolve(offer.agent_id());
v1::TaskInfo taskInfo =
evolve(createTask(slaveId, resources, ""));
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(taskInfo);
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACCEPT);
Call::Accept* accept = call.mutable_accept();
accept->add_offer_ids()->CopyFrom(offer.id());
v1::Offer::Operation* operation = accept->add_operations();
operation->set_type(v1::Offer::Operation::LAUNCH_GROUP);
v1::Offer::Operation::LaunchGroup* launchGroup =
operation->mutable_launch_group();
launchGroup->mutable_executor()->CopyFrom(evolve(executorInfo));
launchGroup->mutable_task_group()->CopyFrom(taskGroup);
mesos.send(call);
}
AWAIT_READY(containerConfig);
Resources containerResources = Resources(containerConfig->resources());
containerResources.unallocate();
// The resources used to launch executor container should include
// both executorInfo's resources and all task's resources.
EXPECT_EQ(containerResources, resources + resources);
}
// This test ensures that we do not send a queued task group to
// the executor if any of its tasks are killed before the executor
// subscribes with the agent.
TEST_F(SlaveTest, KillQueuedTaskGroup)
{
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
auto executor = std::make_shared<v1::MockHTTPExecutor>();
Resources resources =
Resources::parse("cpus:0.1;mem:32;disk:32").get();
ExecutorInfo executorInfo = DEFAULT_EXECUTOR_INFO;
executorInfo.set_type(ExecutorInfo::CUSTOM);
executorInfo.mutable_resources()->CopyFrom(resources);
const ExecutorID& executorId = executorInfo.executor_id();
TestContainerizer containerizer(executorId, executor);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), &containerizer);
ASSERT_SOME(slave);
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return());
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(v1::DEFAULT_FRAMEWORK_INFO);
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
// Update `executorInfo` with the subscribed `frameworkId`.
executorInfo.mutable_framework_id()->CopyFrom(devolve(frameworkId));
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
Future<v1::executor::Mesos*> executorLibrary;
EXPECT_CALL(*executor, connected(_))
.WillOnce(FutureArg<0>(&executorLibrary));
const v1::Offer& offer = offers->offers(0);
const SlaveID slaveId = devolve(offer.agent_id());
// Launch a task and task group.
v1::TaskInfo taskInfo1 =
evolve(createTask(slaveId, resources, "", executorId));
taskInfo1.mutable_executor()->CopyFrom(evolve(executorInfo));
v1::TaskInfo taskInfo2 =
evolve(createTask(slaveId, resources, ""));
v1::TaskInfo taskInfo3 =
evolve(createTask(slaveId, resources, ""));
v1::TaskGroupInfo taskGroup;
taskGroup.add_tasks()->CopyFrom(taskInfo2);
taskGroup.add_tasks()->CopyFrom(taskInfo3);
const hashset<v1::TaskID> tasks{taskInfo2.task_id(), taskInfo3.task_id()};
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACCEPT);
Call::Accept* accept = call.mutable_accept();
accept->add_offer_ids()->CopyFrom(offer.id());
v1::Offer::Operation* operation1 = accept->add_operations();
operation1->set_type(v1::Offer::Operation::LAUNCH);
operation1->mutable_launch()->add_task_infos()->CopyFrom(taskInfo1);
v1::Offer::Operation* operation2 = accept->add_operations();
operation2->set_type(v1::Offer::Operation::LAUNCH_GROUP);
v1::Offer::Operation::LaunchGroup* launchGroup =
operation2->mutable_launch_group();
launchGroup->mutable_executor()->CopyFrom(evolve(executorInfo));
launchGroup->mutable_task_group()->CopyFrom(taskGroup);
mesos.send(call);
}
AWAIT_READY(executorLibrary);
Future<v1::scheduler::Event::Update> update1;
Future<v1::scheduler::Event::Update> update2;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(FutureArg<1>(&update1))
.WillOnce(FutureArg<1>(&update2))
.WillRepeatedly(Return());
// Kill a task in the task group before the executor
// subscribes with the agent.
mesos.send(
v1::createCallKill(frameworkId, taskInfo2.task_id(), offer.agent_id()));
AWAIT_READY(update1);
AWAIT_READY(update2);
const hashset<v1::TaskID> killedTasks{
update1->status().task_id(), update2->status().task_id()};
EXPECT_EQ(v1::TASK_KILLED, update1->status().state());
EXPECT_EQ(v1::TASK_KILLED, update2->status().state());
EXPECT_EQ(tasks, killedTasks);
EXPECT_CALL(*executor, subscribed(_, _));
// The executor should only receive the queued task upon subscribing
// with the agent since the task group has been killed in the meantime.
Future<Nothing> launch;
EXPECT_CALL(*executor, launch(_, _))
.WillOnce(FutureSatisfy(&launch));
EXPECT_CALL(*executor, launchGroup(_, _))
.Times(0);
{
v1::executor::Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.mutable_executor_id()->CopyFrom(evolve(executorId));
call.set_type(v1::executor::Call::SUBSCRIBE);
call.mutable_subscribe();
executorLibrary.get()->send(call);
}
AWAIT_READY(launch);
EXPECT_CALL(*executor, shutdown(_))
.Times(AtMost(1));
}
// Test the max_completed_executors_per_framework flag.
TEST_F(SlaveTest, MaxCompletedExecutorsPerFrameworkFlag)
{
Clock::pause();
// We verify that the proper amount of history is maintained
// by launching a single framework with exactly 2 executors. We
// do this when setting `max_completed_executors_per_framework`
// to 0, 1, and 2. This covers the cases of maintaining no
// history, some history less than the total number of executors
// launched, and history equal to the total number of executors
// launched.
const size_t totalExecutorsPerFramework = 2;
const size_t maxExecutorsPerFrameworkArray[] = {0, 1, 2};
foreach (const size_t maxExecutorsPerFramework,
maxExecutorsPerFrameworkArray) {
master::Flags masterFlags = MesosTest::CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
hashmap<ExecutorID, Executor*> executorMap;
vector<Owned<MockExecutor>> executors;
vector<ExecutorInfo> executorInfos;
for (size_t i = 0; i < totalExecutorsPerFramework; i++) {
ExecutorInfo executorInfo = createExecutorInfo(stringify(i), "exit 1");
executorInfos.push_back(executorInfo);
Owned<MockExecutor> executor =
Owned<MockExecutor>(new MockExecutor(executorInfo.executor_id()));
executorMap.put(executorInfo.executor_id(), executor.get());
executors.push_back(executor);
}
TestContainerizer containerizer(executorMap);
slave::Flags agentFlags = CreateSlaveFlags();
agentFlags.max_completed_executors_per_framework = maxExecutorsPerFramework;
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> agent =
StartSlave(detector.get(), &containerizer, agentFlags);
ASSERT_SOME(agent);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
Future<Nothing> schedRegistered;
EXPECT_CALL(sched, registered(_, _, _))
.WillOnce(FutureSatisfy(&schedRegistered));
process::Queue<Offer> offers;
EXPECT_CALL(sched, resourceOffers(_, _))
.WillRepeatedly(EnqueueOffers(&offers));
driver.start();
AWAIT_READY(schedRegistered);
for (size_t i = 0; i < totalExecutorsPerFramework; i++) {
// Advance the clock to trigger both agent registration and a
// batch allocation.
Clock::advance(agentFlags.registration_backoff_factor);
Clock::advance(masterFlags.allocation_interval);
Future<Offer> offer = offers.get();
AWAIT_READY(offer);
TaskInfo task;
task.set_name("");
task.mutable_task_id()->set_value(stringify(i));
task.mutable_slave_id()->MergeFrom(offer->slave_id());
task.mutable_resources()->MergeFrom(offer->resources());
task.mutable_executor()->MergeFrom(executorInfos[i]);
EXPECT_CALL(*executors[i], registered(_, _, _, _));
// Make sure the task passes through its `TASK_FINISHED`
// state properly. We force this state change through
// the launchTask() callback on our MockExecutor.
Future<TaskStatus> statusFinished;
EXPECT_CALL(*executors[i], launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_FINISHED));
EXPECT_CALL(sched, statusUpdate(_, _))
.WillOnce(FutureArg<1>(&statusFinished));
driver.launchTasks(offer->id(), {task});
AWAIT_READY(statusFinished);
EXPECT_EQ(TASK_FINISHED, statusFinished->state());
EXPECT_CALL(*executors[i], shutdown(_))
.Times(AtMost(1));
}
// Destroy all of the containers to complete the executors.
Future<hashset<ContainerID>> containerIds = containerizer.containers();
AWAIT_READY(containerIds);
foreach (const ContainerID& containerId, containerIds.get()) {
Future<Nothing> executorLost;
EXPECT_CALL(sched, executorLost(_, _, _, _))
.WillOnce(FutureSatisfy(&executorLost));
AWAIT_READY(containerizer.destroy(containerId));
AWAIT_READY(executorLost);
}
// Ensure the agent processes the executor terminations.
Clock::settle();
// At this point the agent would have considered the framework
// completed since it no longer has active executors.
Future<Response> response = process::http::get(
agent.get()->pid,
"state",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
Try<JSON::Object> parse = JSON::parse<JSON::Object>(response->body);
ASSERT_SOME(parse);
JSON::Object state = parse.get();
Result<JSON::Array> completedFrameworks =
state.values["completed_frameworks"].as<JSON::Array>();
// There should be only 1 framework.
ASSERT_EQ(1u, completedFrameworks->values.size());
JSON::Object completedFramework =
completedFrameworks->values[0].as<JSON::Object>();
Result<JSON::Array> completedExecutorsPerFramework =
completedFramework.values["completed_executors"].as<JSON::Array>();
// The number of completed executors in the completed framework
// should match the limit.
EXPECT_EQ(maxExecutorsPerFramework,
completedExecutorsPerFramework->values.size());
driver.stop();
driver.join();
}
}
// This ensures that if the executor reconnect retry is disabled,
// PID-based V0 executors are disallowed from reregistering in
// the steady state.
//
// TODO(bmahler): It should be simpler to write a test that
// follows a standard recipe (e.g. bring up a mock executor).
TEST_F(SlaveTest, ShutdownV0ExecutorIfItReregistersWithoutReconnect)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
slave::Flags agentFlags = CreateSlaveFlags();
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &containerizer, agentFlags);
ASSERT_SOME(slave);
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_checkpoint(true); // Enable checkpointing.
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, frameworkInfo, master.get()->pid, DEFAULT_CREDENTIAL);
FrameworkID frameworkId;
EXPECT_CALL(sched, registered(_, _, _))
.WillOnce(SaveArg<1>(&frameworkId));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
// Advance the clock to trigger both agent registration and a batch
// allocation.
Clock::advance(agentFlags.registration_backoff_factor);
Clock::advance(masterFlags.allocation_interval);
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
EXPECT_CALL(exec, registered(_, _, _, _));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
// Capture the agent and executor PIDs.
Future<Message> registerExecutorMessage =
FUTURE_MESSAGE(Eq(RegisterExecutorMessage().GetTypeName()), _, _);
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(_, _))
.WillOnce(FutureArg<1>(&status));
TaskInfo task;
task.set_name("test-task");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers->at(0).slave_id());
task.mutable_resources()->MergeFrom(offers->at(0).resources());
task.mutable_executor()->MergeFrom(DEFAULT_EXECUTOR_INFO);
driver.launchTasks(offers->at(0).id(), {task});
AWAIT_READY(registerExecutorMessage);
AWAIT_READY(status);
EXPECT_EQ(TASK_RUNNING, status->state());
// Now spoof an executor re-registration, the executor
// should be shut down.
Future<Nothing> executorShutdown;
EXPECT_CALL(exec, shutdown(_))
.WillOnce(FutureSatisfy(&executorShutdown));
UPID executorPid = registerExecutorMessage->from;
UPID agentPid = registerExecutorMessage->to;
ReregisterExecutorMessage reregisterExecutorMessage;
reregisterExecutorMessage.mutable_executor_id()->CopyFrom(
task.executor().executor_id());
reregisterExecutorMessage.mutable_framework_id()->CopyFrom(
frameworkId);
process::post(executorPid, agentPid, reregisterExecutorMessage);
AWAIT_READY(executorShutdown);
driver.stop();
driver.join();
}
// This ensures that if the executor reconnect retry is enabled,
// re-registrations from PID-based V0 executors are ignored when
// already (re-)registered.
//
// TODO(bmahler): It should be simpler to write a test that
// follows a standard recipe (e.g. bring up a mock executor).
TEST_F(SlaveTest, IgnoreV0ExecutorIfItReregistersWithoutReconnect)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
slave::Flags agentFlags = CreateSlaveFlags();
agentFlags.executor_reregistration_timeout = Seconds(2);
agentFlags.executor_reregistration_retry_interval = Seconds(1);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &containerizer, agentFlags);
ASSERT_SOME(slave);
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_checkpoint(true); // Enable checkpointing.
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, frameworkInfo, master.get()->pid, DEFAULT_CREDENTIAL);
FrameworkID frameworkId;
EXPECT_CALL(sched, registered(_, _, _))
.WillOnce(SaveArg<1>(&frameworkId));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
// Advance the clock to trigger both agent registration and a batch
// allocation.
Clock::advance(agentFlags.registration_backoff_factor);
Clock::advance(masterFlags.allocation_interval);
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
EXPECT_CALL(exec, registered(_, _, _, _));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
// Capture the agent and executor PIDs.
Future<Message> registerExecutorMessage =
FUTURE_MESSAGE(Eq(RegisterExecutorMessage().GetTypeName()), _, _);
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(_, _))
.WillOnce(FutureArg<1>(&status));
TaskInfo task;
task.set_name("test-task");
task.mutable_task_id()->set_value("1");
task.mutable_slave_id()->MergeFrom(offers->at(0).slave_id());
task.mutable_resources()->MergeFrom(offers->at(0).resources());
task.mutable_executor()->MergeFrom(DEFAULT_EXECUTOR_INFO);
driver.launchTasks(offers->at(0).id(), {task});
AWAIT_READY(registerExecutorMessage);
AWAIT_READY(status);
EXPECT_EQ(TASK_RUNNING, status->state());
// Now spoof an executor re-registration, it should be ignored
// and the agent should not respond.
EXPECT_NO_FUTURE_PROTOBUFS(ExecutorReregisteredMessage(), _, _);
Future<Nothing> executorShutdown;
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1))
.WillOnce(FutureSatisfy(&executorShutdown));
UPID executorPid = registerExecutorMessage->from;
UPID agentPid = registerExecutorMessage->to;
ReregisterExecutorMessage reregisterExecutorMessage;
reregisterExecutorMessage.mutable_executor_id()->CopyFrom(
task.executor().executor_id());
reregisterExecutorMessage.mutable_framework_id()->CopyFrom(
frameworkId);
process::post(executorPid, agentPid, reregisterExecutorMessage);
Clock::settle();
EXPECT_TRUE(executorShutdown.isPending());
driver.stop();
driver.join();
}
// This test verifies that an executor's latest run directory can
// be browsed via the `/files` endpoint both while the executor is
// still running and after the executor terminates.
//
// Note that we only test the recommended virtual path format:
// `/framework/FID/executor/EID/latest`.
TEST_F(SlaveTest, BrowseExecutorSandboxByVirtualPath)
{
master::Flags masterFlags = this->CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
slave::Flags agentFlags = this->CreateSlaveFlags();
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Future<Nothing> __recover = FUTURE_DISPATCH(_, &Slave::__recover);
Owned<MasterDetector> detector = master.get()->createDetector();
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &containerizer, agentFlags);
ASSERT_SOME(slave);
// Ensure slave has finished recovery.
AWAIT_READY(__recover);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
// Advance the clock to trigger both agent registration and a batch
// allocation.
Clock::advance(agentFlags.registration_backoff_factor);
Clock::advance(masterFlags.allocation_interval);
AWAIT_READY(offers);
EXPECT_FALSE(offers->empty());
Resources executorResources = Resources::parse("cpus:0.1;mem:32").get();
executorResources.allocate("*");
TaskID taskId;
taskId.set_value("1");
TaskInfo task;
task.set_name("");
task.mutable_task_id()->MergeFrom(taskId);
task.mutable_slave_id()->MergeFrom(offers.get()[0].slave_id());
task.mutable_resources()->MergeFrom(
Resources(offers.get()[0].resources()) - executorResources);
task.mutable_executor()->MergeFrom(DEFAULT_EXECUTOR_INFO);
task.mutable_executor()->mutable_resources()->CopyFrom(executorResources);
EXPECT_CALL(exec, registered(_, _, _, _));
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> status;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status));
driver.launchTasks(offers.get()[0].id(), {task});
AWAIT_READY(status);
EXPECT_EQ(TASK_RUNNING, status->state());
// Manually inject a file into the sandbox.
FrameworkID frameworkId = offers->front().framework_id();
SlaveID slaveId = offers->front().slave_id();
const string latestRunPath = paths::getExecutorLatestRunPath(
agentFlags.work_dir, slaveId, frameworkId, DEFAULT_EXECUTOR_ID);
EXPECT_TRUE(os::exists(latestRunPath));
ASSERT_SOME(os::write(path::join(latestRunPath, "foo.bar"), "testing"));
const string virtualPath =
paths::getExecutorVirtualPath(frameworkId, DEFAULT_EXECUTOR_ID);
const process::UPID files("files", slave.get()->pid.address);
{
const string query = string("path=") + virtualPath;
Future<Response> response = process::http::get(
files,
"browse",
query,
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_ASSERT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_ASSERT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
Try<JSON::Array> parse = JSON::parse<JSON::Array>(response->body);
ASSERT_SOME(parse);
EXPECT_NE(0u, parse->values.size());
}
{
const string query =
string("path=") + path::join(virtualPath, "foo.bar") + "&offset=0";
process::UPID files("files", slave.get()->pid.address);
Future<Response> response = process::http::get(
files,
"read",
query,
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_ASSERT_RESPONSE_STATUS_EQ(OK().status, response);
JSON::Object expected;
expected.values["offset"] = 0;
expected.values["data"] = "testing";
AWAIT_EXPECT_RESPONSE_BODY_EQ(stringify(expected), response);
}
// Now destroy the executor and make sure that the sandbox is
// still available. We're sure that the GC won't prune the
// sandbox since the clock is paused.
Future<TaskStatus> status2;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&status2));
EXPECT_CALL(sched, executorLost(_, _, _, _))
.Times(AtMost(1));
AWAIT_READY(containerizer.destroy(frameworkId, DEFAULT_EXECUTOR_ID));
AWAIT_READY(status2);
EXPECT_EQ(TASK_FAILED, status2->state());
{
const string query = string("path=") + virtualPath;
Future<Response> response = process::http::get(
files,
"browse",
query,
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_ASSERT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_ASSERT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
Try<JSON::Array> parse = JSON::parse<JSON::Array>(response->body);
ASSERT_SOME(parse);
EXPECT_NE(0u, parse->values.size());
}
{
const string query =
string("path=") + path::join(virtualPath, "foo.bar") + "&offset=0";
process::UPID files("files", slave.get()->pid.address);
Future<Response> response = process::http::get(
files,
"read",
query,
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_ASSERT_RESPONSE_STATUS_EQ(OK().status, response);
JSON::Object expected;
expected.values["offset"] = 0;
expected.values["data"] = "testing";
AWAIT_EXPECT_RESPONSE_BODY_EQ(stringify(expected), response);
}
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
}
// This test verifies that a disconnected PID-based executor will drop
// RunTaskMessages.
TEST_F(SlaveTest, DisconnectedExecutorDropsMessages)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
slave::Flags slaveFlags = CreateSlaveFlags();
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
Future<SlaveRegisteredMessage> slaveRegisteredMessage =
FUTURE_PROTOBUF(SlaveRegisteredMessage(), _, _);
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &containerizer, slaveFlags);
ASSERT_SOME(slave);
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(slaveRegisteredMessage);
// Enable checkpointing for the framework so that the executor continues
// running after agent termination.
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_checkpoint(true);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, frameworkInfo, master.get()->pid, false, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(_, _))
.WillOnce(FutureArg<1>(&offers));
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
FrameworkID frameworkId = offers->front().framework_id();
TaskInfo runningTask =
createTask(offers->front(), "sleep 1000", DEFAULT_EXECUTOR_ID);
// Capture the executor registration message to get the executor's pid.
Future<Message> registerExecutor =
FUTURE_MESSAGE(Eq(RegisterExecutorMessage().GetTypeName()), _, _);
EXPECT_CALL(exec, registered(_, _, _, _));
// Capture the `RunTaskMessage` so that we can use the framework pid to spoof
// another `RunTaskMessage` later.
Future<RunTaskMessage> capturedRunTaskMessage =
FUTURE_PROTOBUF(RunTaskMessage(), master.get()->pid, slave.get()->pid);
// In addition to returning the expected task status here, this expectation
// will also ensure that the spoofed `RunTaskMessage` we send later does not
// trigger a call to `launchTask`.
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> statusUpdate;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusUpdate));
driver.launchTasks(offers->front().id(), {runningTask});
AWAIT_READY(registerExecutor);
UPID executorPid = registerExecutor->from;
AWAIT_READY(capturedRunTaskMessage);
AWAIT_READY(statusUpdate);
ASSERT_EQ(TASK_RUNNING, statusUpdate->state());
Future<Nothing> _statusUpdateAcknowledgement =
FUTURE_DISPATCH(_, &Slave::_statusUpdateAcknowledgement);
driver.acknowledgeStatusUpdate(statusUpdate.get());
AWAIT_READY(_statusUpdateAcknowledgement);
// Ensure that the executor continues running after agent termination.
EXPECT_CALL(exec, shutdown(_))
.Times(0);
// Terminate the agent so that the executor becomes disconnected.
slave.get()->terminate();
Clock::settle();
TaskInfo droppedTask =
createTask(offers->front(), "sleep 1000", DEFAULT_EXECUTOR_ID);
RunTaskMessage runTaskMessage;
runTaskMessage.mutable_framework_id()->CopyFrom(frameworkId);
runTaskMessage.mutable_framework()->CopyFrom(frameworkInfo);
runTaskMessage.mutable_task()->CopyFrom(droppedTask);
runTaskMessage.set_pid(capturedRunTaskMessage->pid());
// Send the executor a `RunTaskMessage` while it's disconnected.
// This message should be dropped.
process::post(executorPid, runTaskMessage);
// Settle the clock to ensure that the `RunTaskMessage` is processed. If it is
// not ignored, the test would fail due to a violation of the expectation we
// previously registered on `Executor::launchTask`.
Clock::settle();
// Executor may call shutdown during test teardown.
EXPECT_CALL(exec, shutdown(_))
.Times(AtMost(1));
driver.stop();
driver.join();
Clock::resume();
}
// This test verifies that the 'executor_reregistration_timeout' agent flag
// successfully extends the timeout within which an executor can reregister.
TEST_F(SlaveTest, ExecutorReregistrationTimeoutFlag)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
// Set the executor reregister timeout to a value greater than the default.
slave::Flags slaveFlags = CreateSlaveFlags();
slaveFlags.executor_reregistration_timeout = process::TEST_AWAIT_TIMEOUT;
Fetcher fetcher(slaveFlags);
Try<MesosContainerizer*> _containerizer =
MesosContainerizer::create(slaveFlags, true, &fetcher);
ASSERT_SOME(_containerizer);
Owned<slave::Containerizer> containerizer(_containerizer.get());
Owned<MasterDetector> detector = master.get()->createDetector();
Future<SlaveRegisteredMessage> slaveRegisteredMessage =
FUTURE_PROTOBUF(SlaveRegisteredMessage(), _, _);
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), containerizer.get(), slaveFlags);
ASSERT_SOME(slave);
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(slaveRegisteredMessage);
// Enable checkpointing for the framework.
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_checkpoint(true);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, frameworkInfo, master.get()->pid, false, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(_, _))
.WillOnce(FutureArg<1>(&offers));
driver.start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
TaskInfo task = createTask(offers->front(), SLEEP_COMMAND(1000));
Future<TaskStatus> statusUpdate0;
Future<TaskStatus> statusUpdate1;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusUpdate0))
.WillOnce(FutureArg<1>(&statusUpdate1));
driver.launchTasks(offers->front().id(), {task});
AWAIT_READY(statusUpdate0);
ASSERT_EQ(TASK_STARTING, statusUpdate0->state());
driver.acknowledgeStatusUpdate(statusUpdate0.get());
AWAIT_READY(statusUpdate1);
ASSERT_EQ(TASK_RUNNING, statusUpdate1->state());
Future<Nothing> _statusUpdateAcknowledgement =
FUTURE_DISPATCH(_, &Slave::_statusUpdateAcknowledgement);
driver.acknowledgeStatusUpdate(statusUpdate1.get());
AWAIT_READY(_statusUpdateAcknowledgement);
slave.get()->terminate();
Future<ReregisterExecutorMessage> reregisterExecutor =
DROP_PROTOBUF(ReregisterExecutorMessage(), _, _);
Future<SlaveReregisteredMessage> slaveReregistered =
FUTURE_PROTOBUF(SlaveReregisteredMessage(), _, _);
// Restart the slave (use same flags) with a new containerizer.
_containerizer = MesosContainerizer::create(slaveFlags, true, &fetcher);
ASSERT_SOME(_containerizer);
containerizer.reset(_containerizer.get());
slave = StartSlave(detector.get(), containerizer.get(), slaveFlags);
ASSERT_SOME(slave);
// Ensure that the executor attempts to reregister, so that we can capture
// its re-registration message.
AWAIT_READY(reregisterExecutor);
// Make sure that we're advancing the clock more than the default timeout.
ASSERT_TRUE(
slaveFlags.executor_reregistration_timeout * 0.9 >
slave::EXECUTOR_REREGISTRATION_TIMEOUT);
Clock::advance(slaveFlags.executor_reregistration_timeout * 0.9);
// Send the executor's delayed re-registration message.
process::post(slave.get()->pid, reregisterExecutor.get());
// Advance the clock to prompt the agent to reregister, and ensure that the
// executor's task would have been marked unreachable if the executor had not
// reregistered successfully.
Clock::advance(slaveFlags.executor_reregistration_timeout * 0.2);
Clock::resume();
AWAIT_READY(slaveReregistered);
// Perform reconciliation to verify that the task has not been transitioned to
// TASK_LOST, as would occur if the agent had been deemed unreachable.
vector<TaskStatus> statuses;
TaskStatus status;
status.mutable_task_id()->CopyFrom(task.task_id());
status.set_state(TASK_STAGING); // Dummy value.
statuses.push_back(status);
Future<TaskStatus> statusUpdate2;
EXPECT_CALL(sched, statusUpdate(_, _))
.WillOnce(FutureArg<1>(&statusUpdate2));
driver.reconcileTasks(statuses);
AWAIT_READY(statusUpdate2);
EXPECT_EQ(TASK_RUNNING, statusUpdate2->state());
EXPECT_EQ(TaskStatus::SOURCE_MASTER, statusUpdate2->source());
EXPECT_EQ(TaskStatus::REASON_RECONCILIATION, statusUpdate2->reason());
driver.stop();
driver.join();
}
class DefaultContainerDNSFlagTest
: public MesosTest,
public WithParamInterface<string> {};
INSTANTIATE_TEST_CASE_P(
ContainerizerType,
DefaultContainerDNSFlagTest,
::testing::Values("mesos", "docker"));
// This test verifies the validation for the
// agent flag `--default_container_dns`.
TEST_P(DefaultContainerDNSFlagTest, ValidateFlag)
{
const int argc = 4;
const char* argv[argc] = {
"/path/to/program",
"--master=127.0.0.1:5050",
"--work_dir=/tmp"
};
string containerizer = GetParam();
// Verifies the unknown network mode is not supported.
//
// TODO(qianzhang): Change the value of the `network_mode`
// to an non-existent enum value once MESOS-7828 is resolved.
string defaultContainerDNSInfo =
"--default_container_dns={"
" \"" + containerizer + "\": [\n"
" {\n"
" \"network_mode\": \"UNKNOWN\",\n"
" \"dns\": {\n"
" \"nameservers\": [ \"8.8.8.8\" ]\n"
" }\n"
" }\n"
" ]\n"
"}";
argv[3] = defaultContainerDNSInfo.c_str();
{
slave::Flags flags;
Try<flags::Warnings> load = flags.load(None(), argc, argv);
EXPECT_ERROR(load);
}
// Verifies the host network mode is not supported.
defaultContainerDNSInfo =
"--default_container_dns={"
" \"" + containerizer + "\": [\n"
" {\n"
" \"network_mode\": \"HOST\",\n"
" \"dns\": {\n"
" \"nameservers\": [ \"8.8.8.8\" ]\n"
" }\n"
" }\n"
" ]\n"
"}";
argv[3] = defaultContainerDNSInfo.c_str();
{
slave::Flags flags;
Try<flags::Warnings> load = flags.load(None(), argc, argv);
EXPECT_ERROR(load);
}
string network_mode = (containerizer == "mesos" ? "CNI" : "USER");
// Verifies multiple DNS configuration without network name for
// user-defined CNM network or CNI network is not supported.
defaultContainerDNSInfo =
"--default_container_dns={"
" \"" + containerizer + "\": [\n"
" {\n"
" \"network_mode\": \"" + network_mode + "\",\n"
" \"dns\": {\n"
" \"nameservers\": [ \"8.8.8.8\" ]\n"
" }\n"
" },\n"
" {\n"
" \"network_mode\": \"" + network_mode + "\",\n"
" \"dns\": {\n"
" \"nameservers\": [ \"8.8.8.8\" ]\n"
" }\n"
" }\n"
" ]\n"
"}";
argv[3] = defaultContainerDNSInfo.c_str();
{
slave::Flags flags;
Try<flags::Warnings> load = flags.load(None(), argc, argv);
EXPECT_ERROR(load);
}
// Verifies multiple DNS configuration with the same network name for CNI
// network or user-defined CNM network or CNI network is not supported.
defaultContainerDNSInfo =
"--default_container_dns={"
" \"" + containerizer + "\": [\n"
" {\n"
" \"network_mode\": \"" + network_mode + "\",\n"
" \"network_name\": \"net1\",\n"
" \"dns\": {\n"
" \"nameservers\": [ \"8.8.8.8\" ]\n"
" }\n"
" },\n"
" {\n"
" \"network_mode\": \"" + network_mode + "\",\n"
" \"network_name\": \"net1\",\n"
" \"dns\": {\n"
" \"nameservers\": [ \"8.8.8.8\" ]\n"
" }\n"
" }\n"
" ]\n"
"}";
argv[3] = defaultContainerDNSInfo.c_str();
{
slave::Flags flags;
Try<flags::Warnings> load = flags.load(None(), argc, argv);
EXPECT_ERROR(load);
}
// Verifies multiple DNS configuration for Docker
// default bridge network is not supported.
if (containerizer == "docker") {
// Verifies the host network mode is not supported.
defaultContainerDNSInfo =
"--default_container_dns={"
" \"" + containerizer + "\": [\n"
" {\n"
" \"network_mode\": \"BRIDGE\",\n"
" \"dns\": {\n"
" \"nameservers\": [ \"8.8.8.8\" ]\n"
" }\n"
" },\n"
" {\n"
" \"network_mode\": \"BRIDGE\",\n"
" \"dns\": {\n"
" \"nameservers\": [ \"8.8.8.8\" ]\n"
" }\n"
" }\n"
" ]\n"
"}";
argv[3] = defaultContainerDNSInfo.c_str();
{
slave::Flags flags;
Try<flags::Warnings> load = flags.load(None(), argc, argv);
EXPECT_ERROR(load);
}
}
}
// This test checks that when a resource provider subscribes with the
// agent's resource provider manager, the agent send an
// `UpdateSlaveMessage` reflecting the updated capacity.
//
// TODO(bbannier): We should also add tests for the agent behavior
// with resource providers where the agent ultimately resends the
// previous total when the master fails over, or for the interaction
// with the usual oversubscription protocol (oversubscribed resources
// vs. updates of total).
TEST_F(SlaveTest, ResourceProviderSubscribe)
{
Clock::pause();
// Start an agent and a master.
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Future<SlaveRegisteredMessage> slaveRegisteredMessage =
FUTURE_PROTOBUF(SlaveRegisteredMessage(), _, _);
// Specify the agent resources so we can check the reported total later.
slave::Flags slaveFlags = CreateSlaveFlags();
slaveFlags.resources = "cpus:2;mem:512;disk:512;ports:[]";
StandaloneMasterDetector detector(master.get()->pid);
Try<Owned<cluster::Slave>> slave = StartSlave(&detector, slaveFlags);
ASSERT_SOME(slave);
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(slaveRegisteredMessage);
mesos::v1::ResourceProviderInfo resourceProviderInfo;
resourceProviderInfo.set_type("org.apache.mesos.resource_provider.test");
resourceProviderInfo.set_name("test");
// Register a local resource provider with the agent.
v1::TestResourceProvider resourceProvider(resourceProviderInfo);
Future<Nothing> connected;
EXPECT_CALL(*resourceProvider.process, connected())
.WillOnce(FutureSatisfy(&connected));
Owned<EndpointDetector> endpointDetector(
resource_provider::createEndpointDetector(slave.get()->pid));
resourceProvider.start(std::move(endpointDetector), ContentType::PROTOBUF);
AWAIT_READY(connected);
Future<mesos::v1::resource_provider::Event::Subscribed> subscribed;
EXPECT_CALL(*resourceProvider.process, subscribed(_))
.WillOnce(FutureArg<0>(&subscribed));
Future<UpdateSlaveMessage> updateSlaveMessage =
FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
{
mesos::v1::resource_provider::Call call;
call.set_type(mesos::v1::resource_provider::Call::SUBSCRIBE);
call.mutable_subscribe()->mutable_resource_provider_info()->CopyFrom(
resourceProviderInfo);
resourceProvider.send(call);
}
// The subscription event contains the assigned resource provider id.
AWAIT_READY(subscribed);
const mesos::v1::ResourceProviderID& resourceProviderId =
subscribed->provider_id();
v1::Resource resourceProviderResources =
v1::Resources::parse("disk", "8096", "*").get();
resourceProviderResources.mutable_provider_id()->CopyFrom(resourceProviderId);
const string resourceVersionUuid = id::UUID::random().toBytes();
{
mesos::v1::resource_provider::Call call;
call.set_type(mesos::v1::resource_provider::Call::UPDATE_STATE);
call.mutable_resource_provider_id()->CopyFrom(resourceProviderId);
mesos::v1::resource_provider::Call::UpdateState* updateState =
call.mutable_update_state();
updateState->mutable_resources()->CopyFrom(
v1::Resources(resourceProviderResources));
updateState->mutable_resource_version_uuid()->set_value(
resourceVersionUuid);
resourceProvider.send(call);
}
AWAIT_READY(updateSlaveMessage);
ASSERT_TRUE(updateSlaveMessage->has_resource_providers());
ASSERT_EQ(1, updateSlaveMessage->resource_providers().providers_size());
const UpdateSlaveMessage::ResourceProvider& receivedResourceProvider =
updateSlaveMessage->resource_providers().providers(0);
EXPECT_EQ(
Resources(devolve(resourceProviderResources)),
Resources(receivedResourceProvider.total_resources()));
EXPECT_EQ(
resourceVersionUuid,
receivedResourceProvider.resource_version_uuid().value());
}
// This test checks that before a workload (executor or task) is
// launched, all resources from resoruce providers nended to run the
// current set of workloads are properly published.
TEST_F(SlaveTest, ResourceProviderPublishAll)
{
// Start an agent and a master.
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
Future<UpdateSlaveMessage> updateSlaveMessage =
FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get());
ASSERT_SOME(slave);
AWAIT_READY(updateSlaveMessage);
// Register a mock local resource provider with the agent.
v1::ResourceProviderInfo resourceProviderInfo;
resourceProviderInfo.set_type("org.apache.mesos.rp.local.mock");
resourceProviderInfo.set_name("test");
vector<v1::Resource> resources = {
v1::Resources::parse("disk", "4096", "role1").get(),
v1::Resources::parse("disk", "4096", "role2").get()
};
v1::TestResourceProvider resourceProvider(resourceProviderInfo, resources);
Owned<EndpointDetector> endpointDetector(
resource_provider::createEndpointDetector(slave.get()->pid));
updateSlaveMessage = FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
resourceProvider.start(std::move(endpointDetector), ContentType::PROTOBUF);
AWAIT_READY(updateSlaveMessage);
// We want to register two frameworks to launch two concurrent tasks
// that use the provider resources, and verify that when the second
// task is launched, all provider resources are published.
// NOTE: The mock schedulers and drivers are stored outside the loop
// to avoid implicit destruction before the test ends.
vector<Owned<MockScheduler>> scheds;
vector<Owned<MesosSchedulerDriver>> drivers;
// We use the filter explicitly here so that the resources will not
// be filtered for 5 seconds (the default).
Filters filters;
filters.set_refuse_seconds(0);
for (size_t i = 0; i < resources.size(); i++) {
FrameworkInfo framework = DEFAULT_FRAMEWORK_INFO;
framework.set_roles(0, resources.at(i).reservations(0).role());
Owned<MockScheduler> sched(new MockScheduler());
Owned<MesosSchedulerDriver> driver(new MesosSchedulerDriver(
sched.get(), framework, master.get()->pid, DEFAULT_CREDENTIAL));
EXPECT_CALL(*sched, registered(driver.get(), _, _));
Future<vector<Offer>> offers;
// Decline unmatched offers.
// NOTE: This ensures that this framework do not hold the agent's
// default resources. Otherwise, the other one will get no offer.
EXPECT_CALL(*sched, resourceOffers(driver.get(), _))
.WillRepeatedly(DeclineOffers());
EXPECT_CALL(*sched, resourceOffers(driver.get(), OffersHaveAnyResource(
std::bind(&Resources::isReserved, lambda::_1, framework.roles(0)))))
.WillOnce(FutureArg<1>(&offers));
driver->start();
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
Future<mesos::v1::resource_provider::Event::PublishResources> publish;
// Two PUBLISH_RESOURCES events will be received: one for launching the
// executor, and the other for launching the task.
auto resourceProviderProcess = resourceProvider.process->self();
EXPECT_CALL(*resourceProvider.process, publishResources(_))
.WillOnce(
Invoke([resourceProviderProcess](
const typename Event::PublishResources& publish) {
dispatch(
resourceProviderProcess,
&v1::TestResourceProviderProcess::publishDefault,
publish);
}))
.WillOnce(DoAll(
Invoke([resourceProviderProcess](
const typename Event::PublishResources& publish) {
dispatch(
resourceProviderProcess,
&v1::TestResourceProviderProcess::publishDefault,
publish);
}),
FutureArg<0>(&publish)));
Future<TaskStatus> taskStarting;
Future<TaskStatus> taskRunning;
EXPECT_CALL(*sched, statusUpdate(driver.get(), _))
.WillOnce(FutureArg<1>(&taskStarting))
.WillOnce(FutureArg<1>(&taskRunning));
// Launch a task using a provider resource.
driver->acceptOffers(
{offers->at(0).id()},
{LAUNCH({createTask(
offers->at(0).slave_id(),
Resources(offers->at(0).resources()).reserved(framework.roles(0)),
createCommandInfo(SLEEP_COMMAND(1000)))})},
filters);
AWAIT_READY(publish);
// Test if the resources of all running executors are published.
// This is checked through counting how many reservatinos there are
// in the published resources: one (role1) when launching the first
// task, two (role1, role2) when the second task is launched.
EXPECT_EQ(i + 1, v1::Resources(publish->resources()).reservations().size());
AWAIT_READY(taskStarting);
EXPECT_EQ(TASK_STARTING, taskStarting->state());
AWAIT_READY(taskRunning);
EXPECT_EQ(TASK_RUNNING, taskRunning->state());
// Store the mock scheduler and driver to prevent destruction.
scheds.emplace_back(std::move(sched));
drivers.emplace_back(std::move(driver));
}
}
// This test checks that a resource provider gets properly disconnected when
// being marked gone, and is not able to reconnect. We expect pending operations
// to be transition to the correct terminal status.
TEST_F(SlaveTest, RemoveResourceProvider)
{
Clock::pause();
// Start an agent and a master.
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Future<UpdateSlaveMessage> updateSlaveMessage =
FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
slave::Flags slaveFlags = CreateSlaveFlags();
StandaloneMasterDetector detector(master.get()->pid);
Try<Owned<cluster::Slave>> slave = StartSlave(&detector, slaveFlags);
ASSERT_SOME(slave);
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(updateSlaveMessage);
v1::ResourceProviderInfo resourceProviderInfo;
resourceProviderInfo.set_type("org.apache.mesos.rp.test");
resourceProviderInfo.set_name("test");
// Register a local resource provider with the agent.
v1::Resource disk = v1::createDiskResource(
"200",
"storage",
None(),
None(),
v1::createDiskSourceRaw(None(), "profile"));
v1::TestResourceProvider resourceProvider(resourceProviderInfo, disk);
Owned<EndpointDetector> endpointDetector(
resource_provider::createEndpointDetector(slave.get()->pid));
updateSlaveMessage = FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
resourceProvider.start(std::move(endpointDetector), ContentType::PROTOBUF);
AWAIT_READY(updateSlaveMessage);
// Register a framework and perform a pending operations on the
// resource provider resources.
v1::FrameworkInfo framework = v1::DEFAULT_FRAMEWORK_INFO;
framework.set_roles(0, disk.reservations(0).role());
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(v1::scheduler::SendSubscribe(framework));
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
// Since the resources from the resource provider have already reached the
// master at this point, the framework will be offered resource provider
// resources.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
Option<v1::Resource> rawDisk;
foreach (const v1::Resource& resource, offer.resources()) {
if (resource.has_provider_id() && resource.has_disk() &&
resource.disk().has_source() &&
resource.disk().source().type() ==
v1::Resource::DiskInfo::Source::RAW) {
rawDisk = resource;
break;
}
}
ASSERT_SOME(rawDisk);
// Create a pending operation.
Future<v1::resource_provider::Event::ApplyOperation> applyOperation;
EXPECT_CALL(*resourceProvider.process, applyOperation(_))
.WillOnce(FutureArg<0>(&applyOperation));
v1::OperationID operationId;
operationId.set_value("operation");
mesos.send(v1::createCallAccept(
frameworkId,
offer,
{v1::CREATE_DISK(
rawDisk.get(),
v1::Resource::DiskInfo::Source::MOUNT,
None(),
operationId)}));
// Wait for the operation to reach the resource provider.
AWAIT_READY(applyOperation);
// A resource provider cannot be removed while it still has resources.
Future<v1::ResourceProviderID> resourceProviderId =
resourceProvider.process->id();
AWAIT_READY(resourceProviderId);
v1::agent::Call v1Call;
v1Call.set_type(v1::agent::Call::MARK_RESOURCE_PROVIDER_GONE);
v1Call.mutable_mark_resource_provider_gone()
->mutable_resource_provider_id()
->CopyFrom(resourceProviderId.get());
constexpr ContentType contentType = ContentType::PROTOBUF;
process::http::Headers headers = createBasicAuthHeaders(DEFAULT_CREDENTIAL);
headers["Accept"] = stringify(contentType);
Future<process::http::Response> response = process::http::post(
slave.get()->pid,
"api/v1",
headers,
serialize(contentType, v1Call),
stringify(contentType));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(InternalServerError().status, response);
// Remove all resources on the resource provider.
updateSlaveMessage = FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
{
using mesos::v1::resource_provider::Call;
using mesos::v1::resource_provider::Event;
Call call;
call.set_type(Call::UPDATE_STATE);
call.mutable_resource_provider_id()->CopyFrom(resourceProviderId.get());
Call::UpdateState* update = call.mutable_update_state();
update->mutable_resources()->Clear();
update->mutable_resource_version_uuid()->set_value(
id::UUID::random().toBytes());
// Still report the operation. This allows us to send an operation
// status update later on.
mesos::v1::Operation* operation = update->add_operations();
ASSERT_TRUE(applyOperation->has_framework_id());
operation->mutable_framework_id()->CopyFrom(applyOperation->framework_id());
operation->mutable_info()->CopyFrom(applyOperation->info());
operation->mutable_uuid()->CopyFrom(applyOperation->operation_uuid());
operation->mutable_latest_status()->CopyFrom(
evolve(protobuf::createOperationStatus(OPERATION_PENDING)));
operation->add_statuses()->CopyFrom(operation->latest_status());
AWAIT_READY(resourceProvider.send(call));
}
// Once the master has seen that there is no resource managed
// by the resource provider it can be removed successfully.
AWAIT_READY(updateSlaveMessage);
Future<v1::scheduler::Event::UpdateOperationStatus> updateOperationStatus;
EXPECT_CALL(*scheduler, updateOperationStatus(_, _))
.WillOnce(FutureArg<1>(&updateOperationStatus));
// The agent will eventually update the master on its resources
// after the resource provider disconnected.
updateSlaveMessage = FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
// The resource provider will receive a TEARDOWN event on being marked gone.
Future<Nothing> teardown;
EXPECT_CALL(*resourceProvider.process, teardown())
.WillOnce(FutureSatisfy(&teardown));
// We expect at least two disconnection events, one initially when the
// connected resource provider gets removed, and when the automatic attempt
// to resubscribe fails and leads the remote to close the connection.
Future<Nothing> disconnected;
EXPECT_CALL(*resourceProvider.process, disconnected())
.WillOnce(DoDefault())
.WillOnce(FutureSatisfy(&disconnected))
.WillRepeatedly(Return()); // Ignore additional ddisconnection events.
// The resource provider will automatically attempt to reconnect.
Future<Nothing> connected;
EXPECT_CALL(*resourceProvider.process, connected())
.WillOnce(DoDefault())
.WillRepeatedly(Return());
// The resource provider should never successfully resubscribe.
EXPECT_CALL(*resourceProvider.process, subscribed(_))
.Times(Exactly(0));
response = process::http::post(
slave.get()->pid,
"api/v1",
headers,
serialize(contentType, v1Call),
stringify(contentType));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_READY(teardown);
// On resource provider removal the framework should have received
// an operation status update.
AWAIT_READY(updateOperationStatus);
EXPECT_EQ(
v1::OperationState::OPERATION_GONE_BY_OPERATOR,
updateOperationStatus->status().state());
// The status update should be generated by the agent and have no
// associated resource provider ID.
EXPECT_TRUE(updateOperationStatus->status().has_agent_id());
EXPECT_FALSE(updateOperationStatus->status().has_resource_provider_id());
// The associated metrics should have also been increased.
EXPECT_TRUE(metricEquals("master/operations/gone_by_operator", 1));
// The agent should also report a change to its resources.
AWAIT_READY(updateSlaveMessage);
// Once we have seen the second disconnection event we know that the
// attempt to resubscribe was unsuccessful.
AWAIT_READY(disconnected);
// Settle the clock to ensure no more `subscribed` calls to the
// resource provider are enqueued.
Clock::settle();
}
// This test checks that the agent correctly updates and sends
// resource version values when it registers or reregisters.
TEST_F(SlaveTest, ResourceVersions)
{
Clock::pause();
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
// Check that the agent sends its resource version uuid with
// `RegisterSlaveMessage`.
Future<RegisterSlaveMessage> registerSlaveMessage =
FUTURE_PROTOBUF(RegisterSlaveMessage(), _, _);
StandaloneMasterDetector detector(master.get()->pid);
slave::Flags slaveFlags = CreateSlaveFlags();
Try<Owned<cluster::Slave>> slave = StartSlave(&detector, slaveFlags);
ASSERT_SOME(slave);
Clock::settle();
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(registerSlaveMessage);
// Since no resource providers registered, the agent only sends its
// own resource version uuid. The agent has no resource provider id.
ASSERT_TRUE(registerSlaveMessage->has_resource_version_uuid());
// Check that the agent sends its resource version uuid in
// `ReregisterSlaveMessage`.
Future<ReregisterSlaveMessage> reregisterSlaveMessage =
FUTURE_PROTOBUF(ReregisterSlaveMessage(), _, _);
// Simulate a new master detected event on the slave,
// so that the slave will attempt to reregister.
detector.appoint(master.get()->pid);
Clock::settle();
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(reregisterSlaveMessage);
// No resource changes occurred on the agent and we expect the
// resource version uuid to be unchanged to the one sent in the
// original registration.
ASSERT_TRUE(reregisterSlaveMessage->has_resource_version_uuid());
EXPECT_EQ(
registerSlaveMessage->resource_version_uuid(),
reregisterSlaveMessage->resource_version_uuid());
}
// Test that it is possible to add additional resources, attributes,
// and a domain when the reconfiguration policy is set to
// `additive`.
TEST_F(SlaveTest, ReconfigurationPolicy)
{
DomainInfo domain = flags::parse<DomainInfo>(
"{"
" \"fault_domain\": {"
" \"region\": {\"name\": \"europe\"},"
" \"zone\": {\"name\": \"europe-b2\"}"
" }"
"}").get();
master::Flags masterFlags = CreateMasterFlags();
// Need to set a master domain, otherwise it will reject a slave with
// a configured domain.
masterFlags.domain = domain;
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
slave::Flags slaveFlags = CreateSlaveFlags();
slaveFlags.attributes = "distro:debian";
slaveFlags.resources = "cpus:4;mem:32;disk:512";
MockExecutor exec(DEFAULT_EXECUTOR_ID);
TestContainerizer containerizer(&exec);
Owned<MasterDetector> detector = master.get()->createDetector();
// Start a slave.
Future<SlaveRegisteredMessage> slaveRegisteredMessage =
FUTURE_PROTOBUF(SlaveRegisteredMessage(), master.get()->pid, _);
Try<Owned<cluster::Slave>> slave =
StartSlave(detector.get(), &containerizer, slaveFlags);
ASSERT_SOME(slave);
// Wait until the slave registers to ensure that it has successfully
// checkpointed its state.
AWAIT_READY(slaveRegisteredMessage);
slave.get()->terminate();
slave->reset();
// Do a valid reconfiguration.
slaveFlags.reconfiguration_policy = "additive";
slaveFlags.resources = "cpus:8;mem:128;disk:512";
slaveFlags.attributes = "distro:debian;version:8";
slaveFlags.domain = domain;
// Restart slave.
Future<SlaveReregisteredMessage> slaveReregisteredMessage =
FUTURE_PROTOBUF(SlaveReregisteredMessage(), master.get()->pid, _);
slave = StartSlave(detector.get(), &containerizer, slaveFlags);
ASSERT_SOME(slave);
// If we get here without the slave exiting, things are working as expected.
AWAIT_READY(slaveReregisteredMessage);
// Start scheduler and check that it gets offered the updated resources
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, DEFAULT_FRAMEWORK_INFO, master.get()->pid, DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(&driver, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(&driver, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(offers);
EXPECT_EQ(1u, offers->size());
// Verify that the offer contains the new domain, attributes and resources.
EXPECT_TRUE(offers.get()[0].has_domain());
EXPECT_EQ(
Attributes(offers.get()[0].attributes()),
Attributes::parse(slaveFlags.attributes.get()));
// The resources are slightly transformed by both master and slave
// before they end up in an offer (in particular, ports are implicitly
// added and they're assigned to role '*'), so we cannot simply compare
// for equality.
Resources offeredResources = Resources(offers.get()[0].resources());
Resources reconfiguredResources = allocatedResources(
Resources::parse(slaveFlags.resources.get()).get(), "*");
EXPECT_TRUE(offeredResources.contains(reconfiguredResources));
}
// This test checks that a resource provider triggers an
// `UpdateSlaveMessage` to be sent to the master if an non-speculated
// operation fails in the resource provider.
TEST_F(SlaveTest, ResourceProviderReconciliation)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
slave::Flags slaveFlags = CreateSlaveFlags();
Future<UpdateSlaveMessage> updateSlaveMessage =
FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
StandaloneMasterDetector detector(master.get()->pid);
Try<Owned<cluster::Slave>> slave = StartSlave(&detector, slaveFlags);
ASSERT_SOME(slave);
Clock::settle();
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(updateSlaveMessage);
mesos::v1::ResourceProviderInfo resourceProviderInfo;
resourceProviderInfo.set_type("org.apache.mesos.resource_provider.test");
resourceProviderInfo.set_name("test");
// Register a resource provider with the agent.
v1::Resources resourceProviderResources = v1::createDiskResource(
"200",
"*",
None(),
None(),
v1::createDiskSourceRaw());
v1::TestResourceProvider resourceProvider(
resourceProviderInfo,
resourceProviderResources);
Owned<EndpointDetector> endpointDetector(
resource_provider::createEndpointDetector(slave.get()->pid));
updateSlaveMessage = FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
resourceProvider.start(std::move(endpointDetector), ContentType::PROTOBUF);
AWAIT_READY(updateSlaveMessage);
// Register a framework to excercise operations.
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
Future<v1::scheduler::Event::Offers> offers1;
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers1))
.WillRepeatedly(Return()); // Ignore subsequent offers;
v1::FrameworkInfo frameworkInfo = v1::DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_roles(0, "foo");
// Subscribe the framework.
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(frameworkInfo);
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId = subscribed->framework_id();
AWAIT_READY(offers1);
ASSERT_FALSE(offers1->offers().empty());
// We now perform a `RESERVE` operation on the offered resources,
// but let the operation fail in the resource provider.
Future<v1::resource_provider::Event::ApplyOperation> operation;
EXPECT_CALL(*resourceProvider.process, applyOperation(_))
.WillOnce(FutureArg<0>(&operation));
{
const v1::Offer& offer = offers1->offers(0);
v1::Resources reserved = offer.resources();
reserved = reserved.filter(
[](const v1::Resource& r) { return r.has_provider_id(); });
ASSERT_FALSE(reserved.empty());
reserved = reserved.pushReservation(v1::createDynamicReservationInfo(
frameworkInfo.roles(0), frameworkInfo.principal()));
Call call =
v1::createCallAccept(frameworkId, offer, {v1::RESERVE(reserved)});
call.mutable_accept()->mutable_filters()->set_refuse_seconds(0);
mesos.send(call);
}
AWAIT_READY(operation);
// We expect the agent to send an `UpdateSlaveMessage` since below
// the resource provider responds with an `UPDATE_STATE` call.
updateSlaveMessage = FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
Future<v1::scheduler::Event::Offers> offers2;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers2))
.WillRepeatedly(Return()); // Ignore subsequent offers;
// Fail the operation in the resource provider. This should trigger
// an `UpdateSlaveMessage` to the master.
{
Future<v1::ResourceProviderID> resourceProviderId =
resourceProvider.process->id();
AWAIT_READY(resourceProviderId);
v1::Resources resourceProviderResources_;
foreach (v1::Resource resource, resourceProviderResources) {
resource.mutable_provider_id()->CopyFrom(resourceProviderId.get());
resourceProviderResources_ += resource;
}
// Update the resource version of the resource provider.
id::UUID resourceVersionUuid = id::UUID::random();
v1::resource_provider::Call call;
call.set_type(v1::resource_provider::Call::UPDATE_STATE);
call.mutable_resource_provider_id()->CopyFrom(resourceProviderId.get());
v1::resource_provider::Call::UpdateState* updateState =
call.mutable_update_state();
updateState->mutable_resource_version_uuid()->CopyFrom(
evolve(protobuf::createUUID(resourceVersionUuid)));
updateState->mutable_resources()->CopyFrom(resourceProviderResources_);
mesos::v1::Operation* _operation = updateState->add_operations();
_operation->mutable_framework_id()->CopyFrom(operation->framework_id());
_operation->mutable_info()->CopyFrom(operation->info());
_operation->mutable_uuid()->CopyFrom(operation->operation_uuid());
mesos::v1::OperationStatus* lastStatus =
_operation->mutable_latest_status();
lastStatus->set_state(::mesos::v1::OPERATION_FAILED);
_operation->add_statuses()->CopyFrom(*lastStatus);
AWAIT_READY(resourceProvider.send(call));
}
AWAIT_READY(updateSlaveMessage);
// The reserve operation will be reported as failed, but the `statuses` field
// will remain empty since no operation status update has been received from
// the resource provider.
ASSERT_TRUE(updateSlaveMessage->has_resource_providers());
ASSERT_EQ(1, updateSlaveMessage->resource_providers().providers_size());
auto provider = updateSlaveMessage->resource_providers().providers(0);
ASSERT_TRUE(provider.has_operations());
ASSERT_EQ(1, provider.operations().operations_size());
const Operation& reserve = provider.operations().operations(0);
EXPECT_EQ(Offer::Operation::RESERVE, reserve.info().type());
ASSERT_TRUE(reserve.has_latest_status());
EXPECT_EQ(OPERATION_FAILED, reserve.latest_status().state());
EXPECT_EQ(0, reserve.statuses_size());
// The resources are returned to the available pool and the framework will get
// offered the same resources as in the previous offer cycle.
Clock::advance(masterFlags.allocation_interval);
Clock::settle();
AWAIT_READY(offers2);
ASSERT_EQ(1, offers2->offers_size());
const v1::Offer& offer1 = offers1->offers(0);
const v1::Offer& offer2 = offers2->offers(0);
EXPECT_EQ(
v1::Resources(offer1.resources()), v1::Resources(offer2.resources()));
}
// This test verifies that the agent checks resource versions received when
// launching tasks against its own state of the used resource providers and
// rejects tasks assuming incompatible state.
TEST_F(SlaveTest, RunTaskResourceVersions)
{
Clock::pause();
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
slave::Flags slaveFlags = CreateSlaveFlags();
Future<UpdateSlaveMessage> updateSlaveMessage =
FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
StandaloneMasterDetector detector(master.get()->pid);
Try<Owned<cluster::Slave>> slave = StartSlave(&detector, slaveFlags);
ASSERT_SOME(slave);
Clock::settle();
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(updateSlaveMessage);
// Register a resource provider with the agent.
mesos::v1::ResourceProviderInfo resourceProviderInfo;
resourceProviderInfo.set_type("org.apache.mesos.resource_provider.test");
resourceProviderInfo.set_name("test");
v1::Resources resourceProviderResources = v1::createDiskResource(
"200",
"*",
None(),
None(),
v1::createDiskSourceRaw());
v1::TestResourceProvider resourceProvider(
resourceProviderInfo,
resourceProviderResources);
Owned<EndpointDetector> endpointDetector(
resource_provider::createEndpointDetector(slave.get()->pid));
updateSlaveMessage = FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
resourceProvider.start(std::move(endpointDetector), ContentType::PROTOBUF);
AWAIT_READY(updateSlaveMessage);
// Start a framework to launch a task.
MockScheduler sched;
MesosSchedulerDriver driver(
&sched,
DEFAULT_FRAMEWORK_INFO,
master.get()->pid,
false,
DEFAULT_CREDENTIAL);
EXPECT_CALL(sched, registered(_, _, _));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(_, _))
.WillOnce(FutureArg<1>(&offers));
driver.start();
// Below we update the agent's resource version of the registered
// resource provider. We prevent this update from propagating to the
// master to simulate a race between the agent updating its state
// and the master launching a task.
updateSlaveMessage = DROP_PROTOBUF(UpdateSlaveMessage(), _, _);
// Update resource version of the resource provider.
{
Future<v1::ResourceProviderID> resourceProviderId =
resourceProvider.process->id();
AWAIT_READY(resourceProviderId);
v1::Resources resourceProviderResources_;
foreach (v1::Resource resource, resourceProviderResources) {
resource.mutable_provider_id()->CopyFrom(resourceProviderId.get());
resourceProviderResources_ += resource;
}
v1::resource_provider::Call call;
call.set_type(v1::resource_provider::Call::UPDATE_STATE);
call.mutable_resource_provider_id()->CopyFrom(resourceProviderId.get());
v1::resource_provider::Call::UpdateState* updateState =
call.mutable_update_state();
updateState->mutable_resource_version_uuid()->CopyFrom(
evolve(protobuf::createUUID()));
updateState->mutable_resources()->CopyFrom(resourceProviderResources_);
AWAIT_READY(resourceProvider.send(call));
}
AWAIT_READY(updateSlaveMessage);
// Launch a task on the offered resources. Since the agent will only check
// resource versions from resource providers used in the task launch, we
// explicitly confirm that the offer included resource provider resources.
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
const Resources& offeredResources = offers->front().resources();
ASSERT_TRUE(std::any_of(
offeredResources.begin(), offeredResources.end(), [](const Resource& r) {
return r.has_provider_id();
}));
TaskInfo task = createTask(offers->front(), "sleep 1000");
Future<TaskStatus> statusUpdate;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusUpdate));
driver.launchTasks(offers->front().id(), {task});
AWAIT_READY(statusUpdate);
EXPECT_EQ(TASK_LOST, statusUpdate->state());
EXPECT_EQ(TaskStatus::SOURCE_SLAVE, statusUpdate->source());
EXPECT_EQ(TaskStatus::REASON_INVALID_OFFERS, statusUpdate->reason());
}
// This test verifies that on agent restarts, unacknowledged operation status
// updates are resent to the master. This verifies that the agent's
// checkpointing/recovery logic works.
//
// To accomplish this:
// 1. Sends a `RESERVE` operation.
// 2. Verifies that the framework receives an operation status update. The
// status update is not acknowledged.
// 3. Restarts the agent.
// 4. Verifies that the agent resends the operation status update.
TEST_F(SlaveTest, RetryOperationStatusUpdateAfterRecovery)
{
Clock::pause();
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Owned<MasterDetector> detector = master.get()->createDetector();
slave::Flags slaveFlags = CreateSlaveFlags();
Try<Owned<cluster::Slave>> slave = StartSlave(detector.get(), slaveFlags);
ASSERT_SOME(slave);
// Advance the clock to trigger agent registration.
Clock::advance(slaveFlags.registration_backoff_factor);
// Register a framework to exercise an operation.
v1::FrameworkInfo frameworkInfo = v1::DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_roles(0, DEFAULT_TEST_ROLE);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(v1::scheduler::SendSubscribe(frameworkInfo));
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
// Set an expectation for the first offer.
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(subscribed);
const v1::FrameworkID& frameworkId = subscribed->framework_id();
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
// Reserve resources.
v1::OperationID operationId;
operationId.set_value("operation");
ASSERT_FALSE(offer.resources().empty());
v1::Resource reservedResources(*(offer.resources().begin()));
reservedResources.add_reservations()->CopyFrom(
v1::createDynamicReservationInfo(
frameworkInfo.roles(0), DEFAULT_CREDENTIAL.principal()));
Future<v1::scheduler::Event::UpdateOperationStatus> update;
EXPECT_CALL(*scheduler, updateOperationStatus(_, _))
.WillOnce(FutureArg<1>(&update));
mesos.send(v1::createCallAccept(
frameworkId,
offer,
{v1::RESERVE(reservedResources, operationId)}));
AWAIT_READY(update);
EXPECT_EQ(operationId, update->status().operation_id());
EXPECT_EQ(v1::OPERATION_FINISHED, update->status().state());
EXPECT_TRUE(metricEquals("master/operations/finished", 1));
// Restart the agent.
slave.get()->terminate();
// Once the agent is restarted, the agent should resend the unacknowledged
// operation status update.
Future<v1::scheduler::Event::UpdateOperationStatus> retriedUpdate;
EXPECT_CALL(*scheduler, updateOperationStatus(_, _))
.WillOnce(FutureArg<1>(&retriedUpdate));
slave = StartSlave(detector.get(), slaveFlags);
ASSERT_SOME(slave);
// Advance the clock to trigger agent registration.
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(retriedUpdate);
EXPECT_EQ(operationId, retriedUpdate->status().operation_id());
EXPECT_EQ(v1::OPERATION_FINISHED, retriedUpdate->status().state());
// The scheduler will acknowledge the operation status update, so the agent
// should receive an acknowledgement.
Future<AcknowledgeOperationStatusMessage> acknowledgeOperationStatusMessage =
FUTURE_PROTOBUF(
AcknowledgeOperationStatusMessage(), master.get()->pid, slave.get()->pid);
mesos.send(v1::createCallAcknowledgeOperationStatus(
frameworkId, offer.agent_id(), None(), retriedUpdate.get()));
AWAIT_READY(acknowledgeOperationStatusMessage);
// The master has acknowledged the operation status update, so the SLRP
// shouldn't send further operation status updates.
EXPECT_NO_FUTURE_PROTOBUFS(UpdateOperationStatusMessage(), _, _);
Clock::advance(slave::STATUS_UPDATE_RETRY_INTERVAL_MIN);
Clock::settle();
}
// This test verifies that on agent failover HTTP-based executors using resource
// provider resources can resubscribe without crashing the agent or killing the
// executor. This is a regression test for MESOS-9667 and MESOS-9711.
TEST_F(
SlaveTest, DISABLED_AgentFailoverHTTPExecutorUsingResourceProviderResources)
{
// This test is run with paused clock to avoid
// dealing with retried task status updates.
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
Future<UpdateSlaveMessage> updateSlaveMessage =
FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
slave::Flags slaveFlags = CreateSlaveFlags();
// Use the same process ID so the executor can resubscribe.
string processId = process::ID::generate("slave");
StandaloneMasterDetector detector(master.get()->pid);
Try<Owned<cluster::Slave>> slave =
StartSlave(&detector, processId, slaveFlags);
ASSERT_SOME(slave);
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(updateSlaveMessage);
mesos::v1::ResourceProviderInfo resourceProviderInfo;
resourceProviderInfo.set_type("org.apache.mesos.resource_provider.test");
resourceProviderInfo.set_name("test");
// Register a resource provider with the agent.
v1::Resources resourceProviderResources = v1::createDiskResource(
"200",
"*",
None(),
None(),
v1::createDiskSourceRaw());
Owned<v1::TestResourceProvider> resourceProvider(new v1::TestResourceProvider(
resourceProviderInfo, resourceProviderResources));
Owned<EndpointDetector> endpointDetector(
resource_provider::createEndpointDetector(slave.get()->pid));
updateSlaveMessage = FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
resourceProvider->start(std::move(endpointDetector), ContentType::PROTOBUF);
AWAIT_READY(updateSlaveMessage);
// Register a framework to exercise operations.
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
Future<Nothing> connected;
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(FutureSatisfy(&connected));
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(connected);
Future<v1::scheduler::Event::Subscribed> subscribed;
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
EXPECT_CALL(*scheduler, offers(_, _))
.WillRepeatedly(v1::scheduler::DeclineOffers());
EXPECT_CALL(
*scheduler,
offers(
_,
v1::scheduler::OffersHaveAnyResource(
&v1::Resources::hasResourceProvider)))
.WillOnce(FutureArg<1>(&offers));
v1::FrameworkInfo frameworkInfo = v1::DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_roles(0, "foo");
frameworkInfo.set_checkpoint(true);
frameworkInfo.set_failover_timeout(Days(365).secs());
// Subscribe the framework.
{
Call call;
call.set_type(Call::SUBSCRIBE);
Call::Subscribe* subscribe = call.mutable_subscribe();
subscribe->mutable_framework_info()->CopyFrom(frameworkInfo);
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
mesos.send(call);
}
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId = subscribed->framework_id();
AWAIT_READY(offers);
ASSERT_EQ(1, offers->offers_size());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID agentId = offer.agent_id();
const v1::Resources resources = offer.resources();
ASSERT_FALSE(resources.filter(&v1::Resources::hasResourceProvider).empty())
<< "Offer does not contain resource provider resources: " << resources;
v1::TaskID taskId;
taskId.set_value(id::UUID::random().toString());
Future<v1::scheduler::Event::Update> taskStarting;
Future<v1::scheduler::Event::Update> taskRunning;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(DoAll(
v1::scheduler::SendAcknowledge(frameworkId, agentId),
FutureArg<1>(&taskStarting)))
.WillOnce(DoAll(
v1::scheduler::SendAcknowledge(frameworkId, agentId),
FutureArg<1>(&taskRunning)));
// The following futures will ensure that the task status update manager has
// checkpointed the status update acknowledgements so there will be no retry.
//
// NOTE: The order of the two `FUTURE_DISPATCH`s is reversed because Google
// Mock will search the expectations in reverse order.
Future<Nothing> _taskRunningAcknowledgement =
FUTURE_DISPATCH(_, &Slave::_statusUpdateAcknowledgement);
Future<Nothing> _taskStartingAcknowledgement =
FUTURE_DISPATCH(_, &Slave::_statusUpdateAcknowledgement);
{
v1::Resources executorResources =
*v1::Resources::parse("cpus:0.1;mem:32;disk:32");
executorResources.allocate(frameworkInfo.roles(0));
v1::ExecutorInfo executorInfo;
executorInfo.set_type(v1::ExecutorInfo::DEFAULT);
executorInfo.mutable_executor_id()->CopyFrom(v1::DEFAULT_EXECUTOR_ID);
executorInfo.mutable_framework_id()->CopyFrom(frameworkId);
executorInfo.mutable_resources()->CopyFrom(executorResources);
ASSERT_TRUE(v1::Resources(offer.resources()).contains(executorResources));
v1::Resources taskResources = offer.resources() - executorResources;
v1::TaskInfo taskInfo =
v1::createTask(agentId, taskResources, SLEEP_COMMAND(1000));
Call call = v1::createCallAccept(
frameworkId,
offer,
{v1::LAUNCH_GROUP(executorInfo, v1::createTaskGroupInfo({taskInfo}))});
mesos.send(call);
}
AWAIT_READY(taskStarting);
ASSERT_EQ(v1::TaskState::TASK_STARTING, taskStarting->status().state());
ASSERT_EQ(v1::TaskStatus::SOURCE_EXECUTOR, taskStarting->status().source());
AWAIT_READY(_taskStartingAcknowledgement);
AWAIT_READY(taskRunning);
ASSERT_EQ(v1::TaskState::TASK_RUNNING, taskRunning->status().state());
ASSERT_EQ(v1::TaskStatus::SOURCE_EXECUTOR, taskRunning->status().source());
AWAIT_READY(_taskRunningAcknowledgement);
// Fail over the agent. We expect the executor to resubscribe successfully
// even if the resource provider does not resubscribe.
EXPECT_CALL(*resourceProvider->process, disconnected())
.Times(AtMost(1));
EXPECT_NO_FUTURE_DISPATCHES(_, &Slave::executorTerminated);
slave.get()->terminate();
// Terminate the mock resource provider so it won't resubscribe.
resourceProvider.reset();
// The following future will be satisfied when an HTTP executor subscribes.
Future<Nothing> executorSubscribed = FUTURE_DISPATCH(_, &Slave::___run);
slave = StartSlave(&detector, processId, slaveFlags);
ASSERT_SOME(slave);
// Resume the clock so when the regression happens, we'll see the executor
// termination and the test will likely (but not 100% reliable) fail.
Clock::resume();
AWAIT_READY(executorSubscribed);
}
// When an agent receives a `DrainSlaveMessage`, it should kill running tasks.
// Agent API outputs related to draining are also verified.
TEST_F(SlaveTest, DrainAgentKillsRunningTask)
{
Clock::pause();
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Future<UpdateSlaveMessage> updateSlaveMessage =
FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
StandaloneMasterDetector detector(master.get()->pid);
slave::Flags slaveFlags = CreateSlaveFlags();
Try<Owned<cluster::Slave>> slave = StartSlave(&detector, slaveFlags, true);
ASSERT_SOME(slave);
slave.get()->start();
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(updateSlaveMessage);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(v1::scheduler::SendSubscribe(v1::DEFAULT_FRAMEWORK_INFO));
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
Future<v1::scheduler::Event::Update> startingUpdate;
Future<v1::scheduler::Event::Update> runningUpdate;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(DoAll(
FutureArg<1>(&startingUpdate),
v1::scheduler::SendAcknowledge(frameworkId, agentId)))
.WillOnce(DoAll(
FutureArg<1>(&runningUpdate),
v1::scheduler::SendAcknowledge(frameworkId, agentId)));
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::TaskInfo taskInfo =
v1::createTask(agentId, resources, SLEEP_COMMAND(1000));
v1::Offer::Operation launch = v1::LAUNCH({taskInfo});
mesos.send(
v1::createCallAccept(
frameworkId,
offer,
{launch}));
AWAIT_READY(startingUpdate);
EXPECT_EQ(v1::TASK_STARTING, startingUpdate->status().state());
AWAIT_READY(runningUpdate);
EXPECT_EQ(v1::TASK_RUNNING, runningUpdate->status().state());
Future<v1::scheduler::Event::Update> killedUpdate;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(FutureArg<1>(&killedUpdate));
// Simulate the master sending a `DrainSlaveMessage` to the agent.
// Immediately kill the task forcefully.
DurationInfo maxGracePeriod;
maxGracePeriod.set_nanoseconds(0);
DrainConfig drainConfig;
drainConfig.mutable_max_grace_period()->CopyFrom(maxGracePeriod);
DrainSlaveMessage drainSlaveMessage;
drainSlaveMessage.mutable_config()->CopyFrom(drainConfig);
process::post(master.get()->pid, slave.get()->pid, drainSlaveMessage);
AWAIT_READY(killedUpdate);
EXPECT_EQ(v1::TASK_KILLED, killedUpdate->status().state());
EXPECT_EQ(
v1::TaskStatus::REASON_AGENT_DRAINING, killedUpdate->status().reason());
// Since the scheduler has not acknowledged the terminal task status update,
// the agent should still be in the draining state. Confirm that its drain
// info appears in API outputs.
{
v1::agent::Call call;
call.set_type(v1::agent::Call::GET_AGENT);
const ContentType contentType = ContentType::PROTOBUF;
process::http::Headers headers = createBasicAuthHeaders(DEFAULT_CREDENTIAL);
headers["Accept"] = stringify(contentType);
Future<process::http::Response> httpResponse =
process::http::post(
slave.get()->pid,
"api/v1",
headers,
serialize(contentType, call),
stringify(contentType));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, httpResponse);
Future<v1::agent::Response> responseMessage =
deserialize<v1::agent::Response>(contentType, httpResponse->body);
AWAIT_READY(responseMessage);
ASSERT_TRUE(responseMessage->get_agent().has_drain_config());
EXPECT_EQ(
drainConfig,
devolve(responseMessage->get_agent().drain_config()));
}
{
Future<Response> response = process::http::get(
slave.get()->pid,
"state",
None(),
createBasicAuthHeaders(DEFAULT_CREDENTIAL));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, response);
AWAIT_EXPECT_RESPONSE_HEADER_EQ(APPLICATION_JSON, "Content-Type", response);
Try<JSON::Object> state = JSON::parse<JSON::Object>(response->body);
ASSERT_SOME(state);
EXPECT_EQ(JSON::protobuf(drainConfig), state->values["drain_config"]);
}
// Now acknowledge the terminal update and confirm that the agent's drain info
// is gone.
Future<StatusUpdateAcknowledgementMessage> terminalAcknowledgement =
FUTURE_PROTOBUF(StatusUpdateAcknowledgementMessage(), _, _);
// The agent won't complete draining until the framework has been removed.
// Set up an expectation to await on this event.
Future<Nothing> removeFramework;
EXPECT_CALL(*slave.get()->mock(), removeFramework(_))
.WillOnce(DoAll(Invoke(slave.get()->mock(),
&MockSlave::unmocked_removeFramework),
FutureSatisfy(&removeFramework)));
{
Call call;
call.mutable_framework_id()->CopyFrom(frameworkId);
call.set_type(Call::ACKNOWLEDGE);
Call::Acknowledge* acknowledge = call.mutable_acknowledge();
acknowledge->mutable_task_id()->CopyFrom(killedUpdate->status().task_id());
acknowledge->mutable_agent_id()->CopyFrom(offer.agent_id());
acknowledge->set_uuid(killedUpdate->status().uuid());
mesos.send(call);
}
// Resume the clock so that the timer used by `delay()` in `os::reap()` can
// elapse and allow the executor process to be reaped.
Clock::resume();
AWAIT_READY(terminalAcknowledgement);
AWAIT_READY(removeFramework);
{
v1::agent::Call call;
call.set_type(v1::agent::Call::GET_AGENT);
const ContentType contentType = ContentType::PROTOBUF;
process::http::Headers headers = createBasicAuthHeaders(DEFAULT_CREDENTIAL);
headers["Accept"] = stringify(contentType);
Future<process::http::Response> httpResponse =
process::http::post(
slave.get()->pid,
"api/v1",
headers,
serialize(contentType, call),
stringify(contentType));
AWAIT_EXPECT_RESPONSE_STATUS_EQ(OK().status, httpResponse);
Future<v1::agent::Response> responseMessage =
deserialize<v1::agent::Response>(contentType, httpResponse->body);
AWAIT_READY(responseMessage);
ASSERT_FALSE(responseMessage->get_agent().has_drain_config());
}
}
// When the agent receives a `DrainSlaveMessage`, it should kill queued tasks.
TEST_F(SlaveTest, DrainAgentKillsQueuedTask)
{
Clock::pause();
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Future<UpdateSlaveMessage> updateSlaveMessage =
FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
MockContainerizer mockContainerizer;
StandaloneMasterDetector detector(master.get()->pid);
slave::Flags slaveFlags = CreateSlaveFlags();
EXPECT_CALL(mockContainerizer, recover(_))
.WillOnce(Return(Nothing()));
EXPECT_CALL(mockContainerizer, containers())
.WillOnce(Return(hashset<ContainerID>()));
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&mockContainerizer,
slaveFlags);
ASSERT_SOME(slave);
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(updateSlaveMessage);
// Set the partition-aware capability to ensure that the terminal update state
// is TASK_GONE_BY_OPERATOR, since we will set `mark_gone = true`.
v1::FrameworkInfo frameworkInfo = v1::DEFAULT_FRAMEWORK_INFO;
frameworkInfo.add_capabilities()->set_type(
v1::FrameworkInfo::Capability::PARTITION_AWARE);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(v1::scheduler::SendSubscribe(frameworkInfo));
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::TaskInfo taskInfo =
v1::createTask(agentId, resources, SLEEP_COMMAND(1000));
v1::Offer::Operation launch = v1::LAUNCH({taskInfo});
// Return a pending future from the containerizer when launching the executor
// container so that the task remains pending.
Promise<slave::Containerizer::LaunchResult> launchResult;
Future<Nothing> launched;
EXPECT_CALL(mockContainerizer, launch(_, _, _, _))
.WillOnce(DoAll(
FutureSatisfy(&launched),
Return(launchResult.future())));
EXPECT_CALL(mockContainerizer, update(_, _, _))
.WillOnce(Return(Nothing()));
mesos.send(
v1::createCallAccept(
frameworkId,
offer,
{launch}));
AWAIT_READY(launched);
Future<v1::scheduler::Event::Update> killedUpdate;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(FutureArg<1>(&killedUpdate));
// Simulate the master sending a `DrainSlaveMessage` to the agent.
// Immediately kill the task forcefully.
DurationInfo maxGracePeriod;
maxGracePeriod.set_nanoseconds(0);
DrainConfig drainConfig;
drainConfig.set_mark_gone(true);
drainConfig.mutable_max_grace_period()->CopyFrom(maxGracePeriod);
DrainSlaveMessage drainSlaveMessage;
drainSlaveMessage.mutable_config()->CopyFrom(drainConfig);
EXPECT_CALL(mockContainerizer, destroy(_))
.WillOnce(Return(None()));
process::post(master.get()->pid, slave.get()->pid, drainSlaveMessage);
AWAIT_READY(killedUpdate);
EXPECT_EQ(v1::TASK_GONE_BY_OPERATOR, killedUpdate->status().state());
EXPECT_EQ(
v1::TaskStatus::REASON_AGENT_DRAINING, killedUpdate->status().reason());
}
// When the agent receives a `DrainSlaveMessage`, it should kill pending tasks.
TEST_F(SlaveTest, DrainAgentKillsPendingTask)
{
Clock::pause();
Try<Owned<cluster::Master>> master = StartMaster();
ASSERT_SOME(master);
Future<UpdateSlaveMessage> updateSlaveMessage =
FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
StandaloneMasterDetector detector(master.get()->pid);
slave::Flags slaveFlags = CreateSlaveFlags();
MockAuthorizer mockAuthorizer;
Try<Owned<cluster::Slave>> slave = StartSlave(
&detector,
&mockAuthorizer,
slaveFlags);
ASSERT_SOME(slave);
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(updateSlaveMessage);
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(v1::scheduler::SendSubscribe(v1::DEFAULT_FRAMEWORK_INFO));
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
Future<v1::scheduler::Event::Offers> offers;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
v1::scheduler::TestMesos mesos(
master.get()->pid,
ContentType::PROTOBUF,
scheduler);
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
AWAIT_READY(offers);
ASSERT_FALSE(offers->offers().empty());
const v1::Offer& offer = offers->offers(0);
const v1::AgentID& agentId = offer.agent_id();
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::TaskInfo taskInfo =
v1::createTask(agentId, resources, SLEEP_COMMAND(1000));
v1::Offer::Operation launch = v1::LAUNCH({taskInfo});
// Intercept authorization so that the task remains pending.
Future<Nothing> authorized;
Promise<bool> promise; // Never satisfied.
EXPECT_CALL(mockAuthorizer, authorized(_))
.WillOnce(DoAll(FutureSatisfy(&authorized),
Return(promise.future())));
mesos.send(
v1::createCallAccept(
frameworkId,
offer,
{launch}));
AWAIT_READY(authorized);
Future<v1::scheduler::Event::Update> killedUpdate;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(FutureArg<1>(&killedUpdate));
// Simulate the master sending a `DrainSlaveMessage` to the agent.
// Immediately kill the task forcefully.
DurationInfo maxGracePeriod;
maxGracePeriod.set_nanoseconds(0);
DrainConfig drainConfig;
drainConfig.set_mark_gone(true);
drainConfig.mutable_max_grace_period()->CopyFrom(maxGracePeriod);
DrainSlaveMessage drainSlaveMessage;
drainSlaveMessage.mutable_config()->CopyFrom(drainConfig);
process::post(master.get()->pid, slave.get()->pid, drainSlaveMessage);
AWAIT_READY(killedUpdate);
// The terminal update state in this case should be TASK_KILLED because the
// scheduler is not partition-aware.
EXPECT_EQ(v1::TASK_KILLED, killedUpdate->status().state());
EXPECT_EQ(
v1::TaskStatus::REASON_AGENT_DRAINING, killedUpdate->status().reason());
}
// This test validates that a draining agent fails further task launch
// attempts to protect its internal draining invariants, and that the
// agent leaves the draining state on its own once all tasks have
// terminated and their status updates have been acknowledged.
TEST_F(SlaveTest, DrainingAgentRejectLaunch)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
ASSERT_SOME(master);
Future<UpdateSlaveMessage> updateSlaveMessage =
FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
StandaloneMasterDetector detector(master.get()->pid);
slave::Flags slaveFlags = CreateSlaveFlags();
Try<Owned<cluster::Slave>> slave = StartSlave(&detector, slaveFlags);
ASSERT_SOME(slave);
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(updateSlaveMessage);
// Register a scheduler to launch tasks.
auto scheduler = std::make_shared<v1::MockHTTPScheduler>();
EXPECT_CALL(*scheduler, connected(_))
.WillOnce(v1::scheduler::SendSubscribe(v1::DEFAULT_FRAMEWORK_INFO));
Future<v1::scheduler::Event::Subscribed> subscribed;
EXPECT_CALL(*scheduler, subscribed(_, _))
.WillOnce(FutureArg<1>(&subscribed));
Future<v1::scheduler::Event::Offers> offers1;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers1))
.WillRepeatedly(Return()); // Ignore subsequent offers.
EXPECT_CALL(*scheduler, heartbeat(_))
.WillRepeatedly(Return()); // Ignore heartbeats.
v1::scheduler::TestMesos mesos(
master.get()->pid, ContentType::PROTOBUF, scheduler);
AWAIT_READY(subscribed);
v1::FrameworkID frameworkId(subscribed->framework_id());
AWAIT_READY(offers1);
ASSERT_FALSE(offers1->offers().empty());
v1::Offer offer = offers1->offers(0);
v1::AgentID agentId = offer.agent_id();
// Launch a task. When the agent is put into draining state this task will be
// killed, but we will leave the draining state open even after the task is
// killed by not acknowledging the terminal task status update.
v1::Resources resources =
v1::Resources::parse("cpus:0.1;mem:32;disk:32").get();
v1::TaskInfo taskInfo1 =
v1::createTask(agentId, resources, SLEEP_COMMAND(1000), None());
// We do not acknowledge the KILLED update to control
// when the agent finishes draining.
Future<v1::scheduler::Event::Update> runningUpdate1;
Future<v1::scheduler::Event::Update> killedUpdate1;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(v1::scheduler::SendAcknowledge(frameworkId, agentId)) // Starting.
.WillOnce(DoAll(
v1::scheduler::SendAcknowledge(frameworkId, agentId),
FutureArg<1>(&runningUpdate1)))
.WillOnce(FutureArg<1>(&killedUpdate1));
mesos.send(
v1::createCallAccept(frameworkId, offer, {v1::LAUNCH({taskInfo1})}));
AWAIT_READY(runningUpdate1);
// Simulate the master sending a `DrainSlaveMessage` to the agent.
DurationInfo maxGracePeriod;
maxGracePeriod.set_nanoseconds(0);
DrainConfig drainConfig;
drainConfig.set_mark_gone(false);
drainConfig.mutable_max_grace_period()->CopyFrom(maxGracePeriod);
DrainSlaveMessage drainSlaveMessage;
drainSlaveMessage.mutable_config()->CopyFrom(drainConfig);
// Explicitly wait for the executor to be terminated.
Future<Nothing> executorTerminated =
FUTURE_DISPATCH(_, &Slave::executorTerminated);
process::post(master.get()->pid, slave.get()->pid, drainSlaveMessage);
// Wait until we have received the terminal task status update
// (which we did not acknowledge) before continuing. The agent will
// subsequentially be left in a draining state.
AWAIT_READY(killedUpdate1);
ASSERT_EQ(v1::TASK_KILLED, killedUpdate1->status().state());
Future<v1::scheduler::Event::Offers> offers2;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers2))
.WillRepeatedly(Return()); // Ignore subsequent offers.
// Resume the clock so the containerizer can detect the terminated executor.
Clock::resume();
AWAIT_READY(executorTerminated);
Clock::pause();
Clock::settle();
Clock::advance(masterFlags.allocation_interval);
AWAIT_READY(offers2);
ASSERT_FALSE(offers2->offers().empty());
offer = offers2->offers(0);
agentId = offer.agent_id();
// Launch another task. Since the agent is in draining
// state the task will be rejected by the agent.
Future<v1::scheduler::Event::Update> lostUpdate;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(DoAll(
v1::scheduler::SendAcknowledge(frameworkId, agentId),
FutureArg<1>(&lostUpdate)));
v1::TaskInfo taskInfo2 =
v1::createTask(agentId, resources, SLEEP_COMMAND(1000), None());
mesos.send(
v1::createCallAccept(frameworkId, offer, {v1::LAUNCH({taskInfo2})}));
AWAIT_READY(lostUpdate);
ASSERT_EQ(taskInfo2.task_id(), lostUpdate->status().task_id());
ASSERT_EQ(v1::TASK_LOST, lostUpdate->status().state());
ASSERT_EQ(
v1::TaskStatus::REASON_AGENT_DRAINING, lostUpdate->status().reason());
// Acknowledge the pending task status update. Once the acknowledgement has
// been processed the agent will leave its draining state and accept task
// launches again.
Future<Nothing> statusUpdateAcknowledgement =
FUTURE_DISPATCH(_, &Slave::_statusUpdateAcknowledgement);
{
v1::scheduler::Call call;
call.set_type(v1::scheduler::Call::ACKNOWLEDGE);
call.mutable_framework_id()->CopyFrom(frameworkId);
v1::scheduler::Call::Acknowledge* acknowledge = call.mutable_acknowledge();
acknowledge->mutable_task_id()->CopyFrom(killedUpdate1->status().task_id());
acknowledge->mutable_agent_id()->CopyFrom(agentId);
acknowledge->set_uuid(killedUpdate1->status().uuid());
mesos.send(call);
}
AWAIT_READY(statusUpdateAcknowledgement);
Future<v1::scheduler::Event::Offers> offers3;
EXPECT_CALL(*scheduler, offers(_, _))
.WillOnce(FutureArg<1>(&offers3))
.WillRepeatedly(Return()); // Ignore subsequent offers.
// Trigger another allocation.
Clock::advance(masterFlags.allocation_interval);
AWAIT_READY(offers3);
ASSERT_FALSE(offers3->offers().empty());
offer = offers3->offers(0);
agentId = offer.agent_id();
// The agent should have left its running state and now accept task launches.
Future<v1::scheduler::Event::Update> runningUpdate2;
EXPECT_CALL(*scheduler, update(_, _))
.WillOnce(v1::scheduler::SendAcknowledge(frameworkId, agentId)) // Starting.
.WillOnce(FutureArg<1>(&runningUpdate2));
mesos.send(
v1::createCallAccept(frameworkId, offer, {v1::LAUNCH({taskInfo2})}));
AWAIT_READY(runningUpdate2);
EXPECT_EQ(taskInfo2.task_id(), runningUpdate2->status().task_id());
EXPECT_EQ(v1::TASK_RUNNING, runningUpdate2->status().state());
}
// This test verifies that if the agent recovers that it is in
// draining state any tasks after the restart are killed.
TEST_F(SlaveTest, CheckpointedDrainInfo)
{
Clock::pause();
master::Flags masterFlags = CreateMasterFlags();
Try<Owned<cluster::Master>> master = StartMaster(masterFlags);
slave::Flags slaveFlags = CreateSlaveFlags();
// Make the executor reregistration timeout less than the agent's
// registration backoff factor to avoid resent status updates.
slaveFlags.executor_reregistration_timeout = Milliseconds(2);
ExecutorID executorId = DEFAULT_EXECUTOR_ID;
MockExecutor exec(executorId);
TestContainerizer containerizer(&exec);
Future<UpdateSlaveMessage> updateSlaveMessage =
FUTURE_PROTOBUF(UpdateSlaveMessage(), _, _);
StandaloneMasterDetector detector(master.get()->pid);
Try<Owned<cluster::Slave>> slave =
StartSlave(&detector, &containerizer, slaveFlags);
ASSERT_SOME(slave);
// Advance the clock to trigger the agent registration.
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(updateSlaveMessage);
// Start a framework.
FrameworkInfo frameworkInfo = DEFAULT_FRAMEWORK_INFO;
frameworkInfo.set_checkpoint(true);
MockScheduler sched;
MesosSchedulerDriver driver(
&sched, frameworkInfo, master.get()->pid, DEFAULT_CREDENTIAL);
Future<Nothing> frameworkId;
EXPECT_CALL(sched, registered(_, _, _))
.WillOnce(FutureSatisfy(&frameworkId));
Future<vector<Offer>> offers;
EXPECT_CALL(sched, resourceOffers(_, _))
.WillOnce(FutureArg<1>(&offers))
.WillRepeatedly(Return()); // Ignore subsequent offers.
driver.start();
AWAIT_READY(frameworkId);
AWAIT_READY(offers);
ASSERT_FALSE(offers->empty());
SlaveID slaveId = offers.get()[0].slave_id();
TaskInfo task = createTask(
slaveId,
Resources::parse("cpus:1;mem:32").get(),
SLEEP_COMMAND(1000),
executorId);
EXPECT_CALL(exec, registered(_, _, _, _));
driver.launchTasks(offers.get()[0].id(), {task});
constexpr int GRACE_PERIOD_NANOS = 1000000;
DurationInfo maxGracePeriod;
maxGracePeriod.set_nanoseconds(GRACE_PERIOD_NANOS);
// We do not mark the agent as gone in contrast to some other tests here to
// validate that we observe `TASK_KILLED` instead of `TASK_GONE_BY_OPERATOR`.
DrainConfig drainConfig;
drainConfig.set_mark_gone(false);
drainConfig.mutable_max_grace_period()->CopyFrom(maxGracePeriod);
DrainSlaveMessage drainSlaveMessage;
drainSlaveMessage.mutable_config()->CopyFrom(drainConfig);
EXPECT_CALL(exec, launchTask(_, _))
.WillOnce(SendStatusUpdateFromTask(TASK_RUNNING));
Future<TaskStatus> statusRunning;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusRunning));
AWAIT_READY(statusRunning);
ASSERT_EQ(TaskState::TASK_RUNNING, statusRunning->state());
// We expect a request to kill the task when the drain request is initially
// received. The executor ignores the request and reregisters after agent
// restart.
Future<Nothing> killTask1;
EXPECT_CALL(exec, killTask(_, _))
.WillOnce(FutureSatisfy(&killTask1));
process::post(master.get()->pid, slave.get()->pid, drainSlaveMessage);
AWAIT_READY(killTask1);
Future<Nothing> reregistered;
EXPECT_CALL(exec, reregistered(_, _))
.WillOnce(DoAll(
Invoke(&exec, &MockExecutor::reregistered),
FutureSatisfy(&reregistered)))
.WillRepeatedly(DoDefault());
// Once the agent has finished recovering executors it should send
// another task kill request to the executor.
EXPECT_CALL(exec, killTask(_, _))
.WillOnce(SendStatusUpdateFromTaskID(TASK_KILLED));
// Restart the agent.
slave.get()->terminate();
Future<TaskStatus> statusKilled;
EXPECT_CALL(sched, statusUpdate(&driver, _))
.WillOnce(FutureArg<1>(&statusKilled))
.WillRepeatedly(Return()); // Ignore resent updates.
slave = StartSlave(&detector, &containerizer, slaveFlags);
AWAIT_READY(reregistered);
// Advance the clock to finish the executor and agent reregistration phases.
Clock::advance(slaveFlags.executor_reregistration_timeout);
Clock::settle();
Clock::advance(slaveFlags.registration_backoff_factor);
AWAIT_READY(statusKilled);
EXPECT_EQ(TASK_KILLED, statusKilled->state());
EXPECT_EQ(TaskStatus::REASON_SLAVE_DRAINING, statusKilled->reason());
}
} // namespace tests {
} // namespace internal {
} // namespace mesos {