[Core] Cancel lease requests before returning a PG bundle (#45919)

Signed-off-by: Jiajun Yao <[email protected]>

python/ray/tests/test_gcs_fault_tolerance.py

@@ -1,4 +1,5 @@
 import sys
+import asyncio
 import os
 import threading
 from time import sleep
@@ -22,6 +23,8 @@
 )
 from ray.job_submission import JobSubmissionClient, JobStatus
 from ray._raylet import GcsClient
+from ray._private.runtime_env.plugin import RuntimeEnvPlugin
+from ray.util.state import list_placement_groups
 
 import psutil
 
@@ -1213,6 +1216,87 @@ def spawn(self, name, namespace):
         raise ValueError(f"Unknown case: {case}")
 
 
+MyPlugin = "MyPlugin"
+MY_PLUGIN_CLASS_PATH = "ray.tests.test_gcs_fault_tolerance.HangPlugin"
+
+
+class HangPlugin(RuntimeEnvPlugin):
+    name = MyPlugin
+
+    async def create(
+        self,
+        uri,
+        runtime_env,
+        ctx,
+        logger,  # noqa: F821
+    ) -> float:
+        while True:
+            await asyncio.sleep(1)
+
+    @staticmethod
+    def validate(runtime_env_dict: dict) -> str:
+        return 1
+
+
+@pytest.mark.parametrize(
+    "ray_start_regular_with_external_redis",
+    [
+        generate_system_config_map(
+            gcs_rpc_server_reconnect_timeout_s=60,
+            testing_asio_delay_us="NodeManagerService.grpc_server.CancelResourceReserve=500000000:500000000",  # noqa: E501
+        ),
+    ],
+    indirect=True,
+)
+@pytest.mark.parametrize(
+    "set_runtime_env_plugins",
+    [
+        '[{"class":"' + MY_PLUGIN_CLASS_PATH + '"}]',
+    ],
+    indirect=True,
+)
+def test_placement_group_removal_after_gcs_restarts(
+    set_runtime_env_plugins, ray_start_regular_with_external_redis
+):
+    @ray.remote
+    def task():
+        pass
+
+    pg = ray.util.placement_group(bundles=[{"CPU": 1}])
+    _ = task.options(
+        max_retries=0,
+        num_cpus=1,
+        scheduling_strategy=PlacementGroupSchedulingStrategy(
+            placement_group=pg,
+        ),
+        runtime_env={
+            MyPlugin: {"name": "f2"},
+            "config": {"setup_timeout_seconds": -1},
+        },
+    ).remote()
+
+    # The task should be popping worker
+    # TODO(jjyao) Use a more determinstic way to
+    # decide whether the task is popping worker
+    sleep(5)
+
+    ray.util.remove_placement_group(pg)
+    # The PG is marked as REMOVED in redis but not removed yet from raylet
+    # due to the injected delay of CancelResourceReserve rpc
+    wait_for_condition(lambda: list_placement_groups()[0].state == "REMOVED")
+
+    ray._private.worker._global_node.kill_gcs_server()
+    # After GCS restarts, it will try to remove the PG resources
+    # again via ReleaseUnusedBundles rpc
+    ray._private.worker._global_node.start_gcs_server()
+
+    def verify_pg_resources_cleaned():
+        r_keys = ray.available_resources().keys()
+        return all("group" not in k for k in r_keys)
+
+    wait_for_condition(verify_pg_resources_cleaned, timeout=30)
+
+
 if __name__ == "__main__":
 
     import pytest

python/ray/tests/test_placement_group_5.py

@@ -470,10 +470,9 @@ async def create(
     ) -> float:
         await asyncio.sleep(PLUGIN_TIMEOUT)
 
-
-@staticmethod
-def validate(runtime_env_dict: dict) -> str:
-    return 1
+    @staticmethod
+    def validate(runtime_env_dict: dict) -> str:
+        return 1
 
 
 @pytest.mark.parametrize(

src/ray/core_worker/core_worker_process.cc

@@ -233,6 +233,7 @@ void CoreWorkerProcessImpl::InitializeSystemConfig() {
   thread.join();
 
   RayConfig::instance().initialize(promise.get_future().get());
+  ray::asio::testing::init();
 }
 
 void CoreWorkerProcessImpl::RunWorkerTaskExecutionLoop() {

src/ray/gcs/gcs_server/gcs_placement_group_scheduler.cc

@@ -243,9 +243,9 @@ void GcsPlacementGroupScheduler::CancelResourceReserve(
   auto node_id = NodeID::FromBinary(node.value()->node_id());
 
   if (max_retry == current_retry_cnt) {
-    RAY_LOG(INFO) << "Failed to cancel resource reserved for bundle because the max "
-                     "retry count is reached. "
-                  << bundle_spec->DebugString() << " at node " << node_id;
+    RAY_LOG(ERROR) << "Failed to cancel resource reserved for bundle because the max "
+                      "retry count is reached. "
+                   << bundle_spec->DebugString() << " at node " << node_id;
     return;
   }
 
@@ -261,11 +261,10 @@ void GcsPlacementGroupScheduler::CancelResourceReserve(
           RAY_LOG(INFO) << "Finished cancelling the resource reserved for bundle: "
                         << bundle_spec->DebugString() << " at node " << node_id;
         } else {
-          // We couldn't delete the pg resources either becuase it is in use
-          // or network issue. Retry.
-          RAY_LOG(INFO) << "Failed to cancel the resource reserved for bundle: "
-                        << bundle_spec->DebugString() << " at node " << node_id
-                        << ". Status: " << status;
+          // We couldn't delete the pg resources because of network issue. Retry.
+          RAY_LOG(WARNING) << "Failed to cancel the resource reserved for bundle: "
+                           << bundle_spec->DebugString() << " at node " << node_id
+                           << ". Status: " << status;
           execute_after(
               io_context_,
               [this, bundle_spec, node, max_retry, current_retry_cnt] {
@@ -568,14 +567,10 @@ void GcsPlacementGroupScheduler::DestroyPlacementGroupPreparedBundleResources(
     for (const auto &iter : *(leasing_bundle_locations)) {
       auto &bundle_spec = iter.second.second;
       auto &node_id = iter.second.first;
-      CancelResourceReserve(
-          bundle_spec,
-          gcs_node_manager_.GetAliveNode(node_id),
-          // Retry 10 * worker registeration timeout to avoid race condition.
-          // See https://github.com/ray-project/ray/pull/42942
-          // for more details.
-          /*max_retry*/ RayConfig::instance().worker_register_timeout_seconds() * 10,
-          /*num_retry*/ 0);
+      CancelResourceReserve(bundle_spec,
+                            gcs_node_manager_.GetAliveNode(node_id),
+                            /*max_retry*/ 5,
+                            /*num_retry*/ 0);
     }
   }
 }
@@ -594,14 +589,10 @@ void GcsPlacementGroupScheduler::DestroyPlacementGroupCommittedBundleResources(
     for (const auto &iter : *(committed_bundle_locations)) {
       auto &bundle_spec = iter.second.second;
       auto &node_id = iter.second.first;
-      CancelResourceReserve(
-          bundle_spec,
-          gcs_node_manager_.GetAliveNode(node_id),
-          // Retry 10 * worker registeration timeout to avoid race condition.
-          // See https://github.com/ray-project/ray/pull/42942
-          // for more details.
-          /*max_retry*/ RayConfig::instance().worker_register_timeout_seconds() * 10,
-          /*num_retry*/ 0);
+      CancelResourceReserve(bundle_spec,
+                            gcs_node_manager_.GetAliveNode(node_id),
+                            /*max_retry*/ 5,
+                            /*num_retry*/ 0);
     }
     committed_bundle_location_index_.Erase(placement_group_id);
     cluster_resource_scheduler_.GetClusterResourceManager()

src/ray/gcs/gcs_server/gcs_server_main.cc

@@ -62,6 +62,7 @@ int main(int argc, char *argv[]) {
   gflags::ShutDownCommandLineFlags();
 
   RayConfig::instance().initialize(config_list);
+  ray::asio::testing::init();
 
   // IO Service for main loop.
   instrumented_io_context main_service;

src/ray/raylet/local_task_manager.cc

@@ -546,21 +546,15 @@ bool LocalTaskManager::PoppedWorkerHandler(
     not_detached_with_owner_failed = true;
   }
 
-  const auto &required_resource =
-      task.GetTaskSpecification().GetRequiredResources().GetResourceMap();
-  for (auto &entry : required_resource) {
-    if (!cluster_resource_scheduler_->GetLocalResourceManager().ResourcesExist(
-            scheduling::ResourceID(entry.first))) {
-      RAY_CHECK(task.GetTaskSpecification().PlacementGroupBundleId().first !=
-                PlacementGroupID::Nil());
-      RAY_LOG(DEBUG) << "The placement group: "
-                     << task.GetTaskSpecification().PlacementGroupBundleId().first
-                     << " was removed when poping workers for task: " << task_id
-                     << ", will cancel the task.";
-      CancelTask(
-          task_id,
-          rpc::RequestWorkerLeaseReply::SCHEDULING_CANCELLED_PLACEMENT_GROUP_REMOVED);
-      canceled = true;
+  if (!canceled) {
+    const auto &required_resource =
+        task.GetTaskSpecification().GetRequiredResources().GetResourceMap();
+    for (auto &entry : required_resource) {
+      // This is to make sure PG resource is not deleted during popping worker
+      // unless the lease request is cancelled.
+      RAY_CHECK(cluster_resource_scheduler_->GetLocalResourceManager().ResourcesExist(
+          scheduling::ResourceID(entry.first)))
+          << entry.first;
     }
   }
 
@@ -855,7 +849,7 @@ void LocalTaskManager::ReleaseTaskArgs(const TaskID &task_id) {
 }
 
 namespace {
-void ReplyCancelled(std::shared_ptr<internal::Work> &work,
+void ReplyCancelled(const std::shared_ptr<internal::Work> &work,
                     rpc::RequestWorkerLeaseReply::SchedulingFailureType failure_type,
                     const std::string &scheduling_failure_message) {
   auto reply = work->reply;
@@ -867,55 +861,67 @@ void ReplyCancelled(std::shared_ptr<internal::Work> &work,
 }
 }  // namespace
 
-bool LocalTaskManager::CancelTask(
-    const TaskID &task_id,
+bool LocalTaskManager::CancelTasks(
+    std::function<bool(const std::shared_ptr<internal::Work> &)> predicate,
     rpc::RequestWorkerLeaseReply::SchedulingFailureType failure_type,
     const std::string &scheduling_failure_message) {
-  for (auto shapes_it = tasks_to_dispatch_.begin(); shapes_it != tasks_to_dispatch_.end();
-       shapes_it++) {
-    auto &work_queue = shapes_it->second;
-    for (auto work_it = work_queue.begin(); work_it != work_queue.end(); work_it++) {
-      const auto &task = (*work_it)->task;
-      if (task.GetTaskSpecification().TaskId() == task_id) {
-        RAY_LOG(DEBUG) << "Canceling task " << task_id << " from dispatch queue.";
-        ReplyCancelled(*work_it, failure_type, scheduling_failure_message);
-        if ((*work_it)->GetState() == internal::WorkStatus::WAITING_FOR_WORKER) {
-          // We've already acquired resources so we need to release them.
-          cluster_resource_scheduler_->GetLocalResourceManager().ReleaseWorkerResources(
-              (*work_it)->allocated_instances);
-          // Release pinned task args.
-          ReleaseTaskArgs(task_id);
-        }
-        if (!task.GetTaskSpecification().GetDependencies().empty()) {
-          task_dependency_manager_.RemoveTaskDependencies(
-              task.GetTaskSpecification().TaskId());
+  bool tasks_cancelled = false;
+
+  ray::erase_if<SchedulingClass, std::shared_ptr<internal::Work>>(
+      tasks_to_dispatch_, [&](const std::shared_ptr<internal::Work> &work) {
+        if (predicate(work)) {
+          const TaskID task_id = work->task.GetTaskSpecification().TaskId();
+          RAY_LOG(DEBUG) << "Canceling task " << task_id << " from dispatch queue.";
+          ReplyCancelled(work, failure_type, scheduling_failure_message);
+          if (work->GetState() == internal::WorkStatus::WAITING_FOR_WORKER) {
+            // We've already acquired resources so we need to release them.
+            cluster_resource_scheduler_->GetLocalResourceManager().ReleaseWorkerResources(
+                work->allocated_instances);
+            // Release pinned task args.
+            ReleaseTaskArgs(task_id);
+          }
+          if (!work->task.GetTaskSpecification().GetDependencies().empty()) {
+            task_dependency_manager_.RemoveTaskDependencies(
+                work->task.GetTaskSpecification().TaskId());
+          }
+          RemoveFromRunningTasksIfExists(work->task);
+          work->SetStateCancelled();
+          tasks_cancelled = true;
+          return true;
+        } else {
+          return false;
         }
-        RemoveFromRunningTasksIfExists(task);
-        (*work_it)->SetStateCancelled();
-        work_queue.erase(work_it);
-        if (work_queue.empty()) {
-          tasks_to_dispatch_.erase(shapes_it);
+      });
+
+  ray::erase_if<std::shared_ptr<internal::Work>>(
+      waiting_task_queue_, [&](const std::shared_ptr<internal::Work> &work) {
+        if (predicate(work)) {
+          ReplyCancelled(work, failure_type, scheduling_failure_message);
+          if (!work->task.GetTaskSpecification().GetDependencies().empty()) {
+            task_dependency_manager_.RemoveTaskDependencies(
+                work->task.GetTaskSpecification().TaskId());
+          }
+          waiting_tasks_index_.erase(work->task.GetTaskSpecification().TaskId());
+          tasks_cancelled = true;
+          return true;
+        } else {
+          return false;
         }
-        return true;
-      }
-    }
-  }
+      });
 
-  auto iter = waiting_tasks_index_.find(task_id);
-  if (iter != waiting_tasks_index_.end()) {
-    const auto &task = (*iter->second)->task;
-    ReplyCancelled(*iter->second, failure_type, scheduling_failure_message);
-    if (!task.GetTaskSpecification().GetDependencies().empty()) {
-      task_dependency_manager_.RemoveTaskDependencies(
-          task.GetTaskSpecification().TaskId());
-    }
-    waiting_task_queue_.erase(iter->second);
-    waiting_tasks_index_.erase(iter);
-
-    return true;
-  }
+  return tasks_cancelled;
+}
 
-  return false;
+bool LocalTaskManager::CancelTask(
+    const TaskID &task_id,
+    rpc::RequestWorkerLeaseReply::SchedulingFailureType failure_type,
+    const std::string &scheduling_failure_message) {
+  return CancelTasks(
+      [task_id](const std::shared_ptr<internal::Work> &work) {
+        return work->task.GetTaskSpecification().TaskId() == task_id;
+      },
+      failure_type,
+      scheduling_failure_message);
 }
 
 bool LocalTaskManager::AnyPendingTasksForResourceAcquisition(

src/ray/raylet/local_task_manager.h

@@ -111,17 +111,15 @@ class LocalTaskManager : public ILocalTaskManager {
   /// \param task: Output parameter.
   void TaskFinished(std::shared_ptr<WorkerInterface> worker, RayTask *task);
 
-  /// Attempt to cancel an already queued task.
+  /// Attempt to cancel all queued tasks that match the predicate.
   ///
-  /// \param task_id: The id of the task to remove.
-  /// \param failure_type: The failure type.
-  ///
-  /// \return True if task was successfully removed. This function will return
-  /// false if the task is already running.
-  bool CancelTask(const TaskID &task_id,
-                  rpc::RequestWorkerLeaseReply::SchedulingFailureType failure_type =
-                      rpc::RequestWorkerLeaseReply::SCHEDULING_CANCELLED_INTENDED,
-                  const std::string &scheduling_failure_message = "") override;
+  /// \param predicate: A function that returns true if a task needs to be cancelled.
+  /// \param failure_type: The reason for cancellation.
+  /// \param scheduling_failure_message: The reason message for cancellation.
+  /// \return True if any task was successfully cancelled.
+  bool CancelTasks(std::function<bool(const std::shared_ptr<internal::Work> &)> predicate,
+                   rpc::RequestWorkerLeaseReply::SchedulingFailureType failure_type,
+                   const std::string &scheduling_failure_message) override;
 
   /// Return if any tasks are pending resource acquisition.
   ///
@@ -203,6 +201,18 @@ class LocalTaskManager : public ILocalTaskManager {
                            const rpc::Address &owner_address,
                            const std::string &runtime_env_setup_error_message);
 
+  /// Attempt to cancel an already queued task.
+  ///
+  /// \param task_id: The id of the task to remove.
+  /// \param failure_type: The failure type.
+  ///
+  /// \return True if task was successfully removed. This function will return
+  /// false if the task is already running.
+  bool CancelTask(const TaskID &task_id,
+                  rpc::RequestWorkerLeaseReply::SchedulingFailureType failure_type =
+                      rpc::RequestWorkerLeaseReply::SCHEDULING_CANCELLED_INTENDED,
+                  const std::string &scheduling_failure_message = "");
+
   /// Attempts to dispatch all tasks which are ready to run. A task
   /// will be dispatched if it is on `tasks_to_dispatch_` and there are still
   /// available resources on the node.