C++ Examples not works well

Ray Core
1

Hi, my friends, Now days I try to use cpp in my project, but I can’t pass the examples and it’s has puzzle me a lot.

The version of ray i used is: 2.5.1.

The code is as follows:

#include <ray/api.h>

/// common function
int Plus(int x, int y) { return x + y; }
/// Declare remote function
RAY_REMOTE(Plus);

/// class
class Counter {
 public:
  int count;

  Counter(int init) { count = init; }
  /// static factory method
  static Counter *FactoryCreate(int init) { return new Counter(init); }

  /// non static function
  int Add(int x) {
    count += x;
    return count;
  }
};

/// Declare remote function
RAY_REMOTE(Counter::FactoryCreate, &Counter::Add);

int main(int argc, char **argv) {
  /// initialization
  ray::RayConfig config;
  config.address="192.168.10.61:9700";
  config.code_search_path.push_back("/home/mic_rd/test/example.so");
  ray::Init(config);

  /// put and get object
  auto object = ray::Put(100);
  auto put_get_result = *(ray::Get(object));
  std::cout << "put_get_result = " << put_get_result << std::endl;

  /// common task
  auto task_object = ray::Task(Plus).Remote(1, 2);
  int task_result = *(ray::Get(task_object));
  std::cout << "task_result = " << task_result << std::endl;
  /// actor
  std::vector<ray::ObjectRef<int> > actors;

  for(int i=0;i<100;i++) {
    ray::ActorHandle<Counter> actor = ray::Actor(Counter::FactoryCreate).Remote(0);
    /// actor task
    actors.push_back(actor.Task(&Counter::Add).Remote(3));
  }
  for(int i=0;i<100;i++) {
    int actor_task_result = *(ray::Get(actors[i]));
    std::cout << "actor_task_result = " << actor_task_result << std::endl;
  }

  /// actor task with reference argument
/*  auto actor_object2 = actor.Task(&Counter::Add).Remote(task_object);
  int actor_task_result2 = *(ray::Get(actor_object2));
  std::cout << "actor_task_result2 = " << actor_task_result2 << std::endl;
*/
  /// shutdown
  ray::Shutdown();
  return 0;
}

The code is works well for a cluster with only one node, but it’s not work for two or more nodes, And the exception is as follows:

[2023-07-14 17:06:22,016 E 2026113 2026113] logging.cc:97: Unhandled exception: N3ray8internal18RayWorkerExceptionE. what():
[2023-07-14 17:06:22,027 E 2026113 2026113] logging.cc:104: Stack trace:
 /home/....../lib/python3.10/site-packages/ray/cpp/lib/libray_api.so(+0xd6fe6a) [0x7f3007a6fe6a] ray::operator<<()
/home/....../lib/python3.10/site-packages/ray/cpp/lib/libray_api.so(_ZN3ray16TerminateHandlerEv+0xd8) [0x7f3007a72628] ray::TerminateHandler()
/lib64/libstdc++.so.6(+0xad37c) [0x7f3006b8037c]
/lib64/libstdc++.so.6(+0xad3e7) [0x7f3006b803e7]
/lib64/libstdc++.so.6(+0xad649) [0x7f3006b80649]
/home/....../lib/python3.10/site-packages/ray/cpp/lib/libray_api.so(+0x48d32c) [0x7f300718d32c] ray::internal::NativeObjectStore::CheckException()
/home/....../lib/python3.10/site-packages/ray/cpp/lib/libray_api.so(_ZN3ray8internal17NativeObjectStore6GetRawERKSt6vectorINS_8ObjectIDESaIS3_EEi+0x27c) [0x7f300726009c] ray::internal::NativeObjectStore::GetRaw()
/home/....../lib/python3.10/site-packages/ray/cpp/lib/libray_api.so(_ZN3ray8internal17NativeObjectStore6GetRawERKNS_8ObjectIDEi+0x96) [0x7f300725fab6] ray::internal::NativeObjectStore::GetRaw()
/home/....../lib/python3.10/site-packages/ray/cpp/lib/libray_api.so(_ZN3ray8internal11ObjectStore3GetERKNS_8ObjectIDEi+0x13) [0x7f3007260d83] ray::internal::ObjectStore::Get()
/.../lib/python3.10/site-packages/ray/cpp/lib/libray_api.so(_ZN3ray8internal18AbstractRayRuntime3GetERKSs+0x49) [0x7f3007259649] ray::internal::AbstractRayRuntime::Get()
./example() [0x40ee0e] ray::GetFromRuntime<>()
./example() [0x40b83c] ray::Get<>()
./example() [0x404ab8] main
/lib64/libc.so.6(+0x3feb0) [0x7f3006813eb0] __libc_start_call_main
/lib64/libc.so.6(__libc_start_main+0x80) [0x7f3006813f60] __libc_start_main@GLIBC_2.2.5
./example() [0x404665] _start

*** SIGABRT received at time=1689325582 on cpu 7 ***
PC: @     0x7f300687554c  (unknown)  __pthread_kill_implementation
    @     0x7f3006828df0  (unknown)  (unknown)
[2023-07-14 17:06:22,028 E 2026113 2026113] logging.cc:361: *** SIGABRT received at time=1689325582 on cpu 7 ***
[2023-07-14 17:06:22,028 E 2026113 2026113] logging.cc:361: PC: @     0x7f300687554c  (unknown)  __pthread_kill_implementation
[2023-07-14 17:06:22,028 E 2026113 2026113] logging.cc:361:     @     0x7f3006828df0  (unknown)  (unknown)
run.sh: line 17: 2026113 Aborted                 (core dumped) LD_LIBRARY_PATH=/.../lib ./example

So, how to solve it? Thank you~

1 Like
2

Did you distribute this library to all of your ray node manually?

3

Hi @matrix , were you able to fix this problem? The suggestion provided by @GuyangSong did not work for me.

4

Yeah.
I‘m calling C++ through Python, with the main algorithm still in C++, but the Ray scheduling uses Python code, it’s works for me. And the performance is ok. I hopes this is helpful to you.