[MESOS-445] Scheduler driver destructor waits forever

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Fix Version/s: 0.20.0
• Assignee: Unassigned
• Reporter: Vinod Kone
• Created: 22/Apr/13 19:50
• Updated: 26/Jun/14 17:28
• Resolved: 26/Jun/14 17:28

Description

When tests fail (e.g., ASSERT failures), the scheduler driver's destructor blocks because the underlying scheduler process is never stopped. We need to make the destructor handle this case.

Thread 1 (Thread 0x7fba1e774720 (LWP 22679)):
#0  0x00007fba1c3d1ee9 in pthread_cond_wait@@GLIBC_2.3.2 () from /lib64/libpthread.so.0
#1  0x00007fba1d36c8de in Gate::arrive (this=0x1e1c800, old=0) at ../../../third_party/libprocess/src/gate.hpp:73
#2  0x00007fba1d32282a in process::ProcessManager::wait (this=0x1ccfd00, pid=...) at ../../../third_party/libprocess/src/process.cpp:2665
#3  0x00007fba1d329a73 in process::wait (pid=..., duration=...) at ../../../third_party/libprocess/src/process.cpp:3105
#4  0x00000000005d19b3 in process::wait (process=0x7fba14082f28, duration=...) at ../../third_party/libprocess/include/process/process.hpp:357
#5  0x00007fba1cf4a9d8 in mesos::MesosSchedulerDriver::~MesosSchedulerDriver (this=0x7fff87544cf0, __in_chrg=<value optimized out>) at ../../src/sched/sched.cpp:757
#6  0x000000000097e3ab in AllocatorZooKeeperTest_FrameworkReregistersFirst_Test<mesos::internal::master::HierarchicalAllocatorProcess<mesos::internal::master::DRFSorter, mesos::internal::master::DRFSorter> >::TestBody (
    this=0x7fba0c0f1830) at ../../src/tests/allocator_zookeeper_tests.cpp:226
#7  0x00000000009987fe in testing::internal::HandleSehExceptionsInMethodIfSupported<testing::Test, void> (object=0x7fba0c0f1830, method=&virtual testing::Test::TestBody(), location=0xab63d3 "the test body")
    at gmock-1.6.0/gtest/src/gtest.cc:2090
#8  0x00000000009a8cfc in testing::internal::HandleExceptionsInMethodIfSupported<testing::Test, void> (object=0x7fba0c0f1830, method=&virtual testing::Test::TestBody(), location=0xab63d3 "the test body")
    at gmock-1.6.0/gtest/src/gtest.cc:2126
#9  0x0000000000990487 in testing::Test::Run (this=0x7fba0c0f1830) at gmock-1.6.0/gtest/src/gtest.cc:2162
#10 0x0000000000997210 in testing::TestInfo::Run (this=0x1c99520) at gmock-1.6.0/gtest/src/gtest.cc:2338
#11 0x0000000000997363 in testing::TestCase::Run (this=0x1c9b080) at gmock-1.6.0/gtest/src/gtest.cc:2445
#12 0x0000000000997666 in testing::internal::UnitTestImpl::RunAllTests (this=0x1c996f0) at gmock-1.6.0/gtest/src/gtest.cc:4237
#13 0x0000000000998d6c in testing::internal::HandleSehExceptionsInMethodIfSupported<testing::internal::UnitTestImpl, bool> (object=0x1c996f0, method=
    (bool (testing::internal::UnitTestImpl::*)(testing::internal::UnitTestImpl *)) 0x9973fa <testing::internal::UnitTestImpl::RunAllTests()>, location=0xab60b8 "auxiliary test code (environments or event listeners)")
    at gmock-1.6.0/gtest/src/gtest.cc:2090
#14 0x00000000009a8776 in testing::internal::HandleExceptionsInMethodIfSupported<testing::internal::UnitTestImpl, bool> (object=0x1c996f0, method=
    (bool (testing::internal::UnitTestImpl::*)(testing::internal::UnitTestImpl *)) 0x9973fa <testing::internal::UnitTestImpl::RunAllTests()>, location=0xab60b8 "auxiliary test code (environments or event listeners)")
    at gmock-1.6.0/gtest/src/gtest.cc:2126
#15 0x000000000098fc14 in testing::UnitTest::Run (this=0xe986a0) at gmock-1.6.0/gtest/src/gtest.cc:3874
#16 0x00000000005c9682 in main (argc=1, argv=0x7fff875465a8) at ../../src/tests/main.cpp:103

Issue Links

• is duplicated by: MESOS-1534 Scheduler process is not explicitly terminated in the destructor of MesosSchedulerDriver.