refactor(executor): globalize QueriesPipelineExecutor creation method

src/query/service/Cargo.toml

@@ -28,6 +28,8 @@ io-uring = [
     # "databend-common-meta-raft-store/io-uring",
 ]
 
+enable_queries_executor = []
+
 [dependencies]
 # Workspace dependencies
 databend-common-arrow = { path = "../../common/arrow" }
@@ -139,6 +141,7 @@ match-template = { workspace = true }
 metrics = "0.20.1"
 minitrace = { workspace = true }
 naive-cityhash = "0.2.0"
+num_cpus = "1.16.0"
 once_cell = { workspace = true }
 opendal = { workspace = true }
 opensrv-mysql = { version = "0.5.0", features = ["tls"] }

src/query/service/src/global_services.rs

@@ -40,6 +40,8 @@ use crate::auth::AuthMgr;
 use crate::catalogs::DatabaseCatalog;
 use crate::clusters::ClusterDiscovery;
 use crate::locks::LockManager;
+#[cfg(feature = "enable_queries_executor")]
+use crate::pipelines::executor::GlobalQueriesExecutor;
 use crate::servers::http::v1::HttpQueryManager;
 use crate::sessions::QueriesQueueManager;
 use crate::sessions::SessionManager;
@@ -131,6 +133,11 @@ impl GlobalServices {
             CloudControlApiProvider::init(addr, config.query.cloud_control_grpc_timeout).await?;
         }
 
+        #[cfg(feature = "enable_queries_executor")]
+        {
+            GlobalQueriesExecutor::init()?;
+        }
+
         Ok(())
     }
 }

src/query/service/src/pipelines/executor/executor_settings.rs

@@ -22,7 +22,7 @@ use databend_common_exception::Result;
 pub struct ExecutorSettings {
     pub query_id: Arc<String>,
     pub max_threads: u64,
-    pub enable_new_executor: bool,
+    pub enable_queries_executor: bool,
     pub max_execute_time_in_seconds: Duration,
     pub executor_node_id: String,
 }
@@ -35,7 +35,7 @@ impl ExecutorSettings {
         let max_execute_time_in_seconds = settings.get_max_execute_time_in_seconds()?;
 
         Ok(ExecutorSettings {
-            enable_new_executor: settings.get_enable_experimental_queries_executor()?,
+            enable_queries_executor: settings.get_enable_experimental_queries_executor()?,
             query_id: Arc::new(query_id),
             max_execute_time_in_seconds: Duration::from_secs(max_execute_time_in_seconds),
             max_threads,

src/query/service/src/pipelines/executor/global_queries_executor.rs

@@ -0,0 +1,41 @@
+// Copyright 2021 Datafuse Labs
+//
+// Licensed 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.
+
+use std::sync::Arc;
+
+use databend_common_base::base::GlobalInstance;
+use databend_common_base::runtime::Thread;
+use databend_common_exception::Result;
+use log::info;
+
+use crate::pipelines::executor::QueriesPipelineExecutor;
+
+pub struct GlobalQueriesExecutor(pub QueriesPipelineExecutor);
+
+impl GlobalQueriesExecutor {
+    pub fn init() -> Result<()> {
+        let num_cpus = num_cpus::get();
+        GlobalInstance::set(QueriesPipelineExecutor::create(num_cpus)?);
+        Thread::spawn(|| {
+            if let Err(e) = Self::instance().execute() {
+                info!("Executor finished with error: {:?}", e);
+            }
+        });
+        Ok(())
+    }
+
+    pub fn instance() -> Arc<QueriesPipelineExecutor> {
+        GlobalInstance::get()
+    }
+}

src/query/service/src/pipelines/executor/mod.rs

@@ -18,6 +18,7 @@ mod executor_condvar;
 mod executor_graph;
 mod executor_settings;
 mod executor_worker_context;
+mod global_queries_executor;
 mod pipeline_complete_executor;
 mod pipeline_executor;
 mod pipeline_pulling_executor;
@@ -35,6 +36,7 @@ pub use executor_settings::ExecutorSettings;
 pub use executor_worker_context::CompletedAsyncTask;
 pub use executor_worker_context::ExecutorTask;
 pub use executor_worker_context::ExecutorWorkerContext;
+pub use global_queries_executor::GlobalQueriesExecutor;
 pub use pipeline_complete_executor::PipelineCompleteExecutor;
 pub use pipeline_executor::PipelineExecutor;
 pub use pipeline_pulling_executor::PipelinePullingExecutor;

src/query/service/src/pipelines/executor/pipeline_executor.rs

@@ -30,7 +30,7 @@ use parking_lot::Condvar;
 use parking_lot::Mutex;
 
 use crate::pipelines::executor::ExecutorSettings;
-use crate::pipelines::executor::QueriesPipelineExecutor;
+use crate::pipelines::executor::GlobalQueriesExecutor;
 use crate::pipelines::executor::QueryPipelineExecutor;
 use crate::pipelines::executor::RunningGraph;
 
@@ -40,8 +40,6 @@ pub type FinishedCallback =
     Box<dyn FnOnce(&Result<Vec<PlanProfile>, ErrorCode>) -> Result<()> + Send + Sync + 'static>;
 
 pub struct QueryWrapper {
-    // TODO: will remove it after refactoring queries pipeline executor
-    executor: Arc<QueriesPipelineExecutor>,
     graph: Arc<RunningGraph>,
     settings: ExecutorSettings,
     on_init_callback: Mutex<Option<InitCallback>>,
@@ -58,7 +56,7 @@ pub enum PipelineExecutor {
 
 impl PipelineExecutor {
     pub fn create(mut pipeline: Pipeline, settings: ExecutorSettings) -> Result<Self> {
-        if !settings.enable_new_executor {
+        if !settings.enable_queries_executor {
             Ok(PipelineExecutor::QueryPipelineExecutor(
                 QueryPipelineExecutor::create(pipeline, settings)?,
             ))
@@ -78,7 +76,6 @@ impl PipelineExecutor {
             )?;
 
             Ok(PipelineExecutor::QueriesPipelineExecutor(QueryWrapper {
-                executor: QueriesPipelineExecutor::create(settings.clone())?,
                 graph,
                 settings,
                 on_init_callback: Mutex::new(on_init_callback),
@@ -93,7 +90,7 @@ impl PipelineExecutor {
         mut pipelines: Vec<Pipeline>,
         settings: ExecutorSettings,
     ) -> Result<Self> {
-        if !settings.enable_new_executor {
+        if !settings.enable_queries_executor {
             Ok(PipelineExecutor::QueryPipelineExecutor(
                 QueryPipelineExecutor::from_pipelines(pipelines, settings)?,
             ))
@@ -141,7 +138,6 @@ impl PipelineExecutor {
             )?;
 
             Ok(PipelineExecutor::QueriesPipelineExecutor(QueryWrapper {
-                executor: QueriesPipelineExecutor::create(settings.clone())?,
                 graph,
                 settings,
                 on_init_callback: Mutex::new(on_init_callback),
@@ -181,9 +177,7 @@ impl PipelineExecutor {
                     &query_wrapper.on_init_callback,
                     &query_wrapper.settings.query_id,
                 )?;
-                query_wrapper
-                    .executor
-                    .send_graph(query_wrapper.graph.clone())?;
+                GlobalQueriesExecutor::instance().send_graph(query_wrapper.graph.clone())?;
 
                 let (lock, cvar) = &*query_wrapper.finish_condvar_wait;
                 let mut finished = lock.lock();
@@ -192,24 +186,25 @@ impl PipelineExecutor {
                 }
 
                 let may_error = query_wrapper.graph.get_error();
-                match may_error {
+                return match may_error {
                     None => {
                         let guard = query_wrapper.on_finished_callback.lock().take();
                         if let Some(on_finished_callback) = guard {
                             catch_unwind(move || {
                                 on_finished_callback(&Ok(self.get_plans_profile()))
                             })??;
                         }
+                        Ok(())
                     }
                     Some(cause) => {
                         let guard = query_wrapper.on_finished_callback.lock().take();
+                        let cause_clone = cause.clone();
                         if let Some(on_finished_callback) = guard {
-                            catch_unwind(move || on_finished_callback(&Err(cause)))??;
+                            catch_unwind(move || on_finished_callback(&Err(cause_clone)))??;
                         }
+                        Err(cause)
                     }
-                }
-
-                Ok(())
+                };
             }
         }
     }

src/query/service/src/pipelines/executor/queries_executor_tasks.rs

@@ -136,21 +136,14 @@ impl QueriesExecutorTasksQueue {
         }
 
         let workers_condvar = context.get_workers_condvar();
-        if !workers_condvar.has_waiting_async_task()
-            && workers_tasks.workers_waiting_status.is_last_active_worker()
-        {
-            drop(workers_tasks);
-            self.finish(workers_condvar.clone());
-            return;
-        }
 
         let worker_id = context.get_worker_id();
         workers_tasks.workers_waiting_status.wait_worker(worker_id);
         drop(workers_tasks);
         workers_condvar.wait(worker_id, self.finished.clone());
     }
 
-    pub fn init_sync_tasks(&self, tasks: VecDeque<ProcessorWrapper>) {
+    pub fn init_sync_tasks(&self, tasks: VecDeque<ProcessorWrapper>, condvar: Arc<WorkersCondvar>) {
         let mut workers_tasks = self.workers_tasks.lock();
 
         let mut worker_id = 0;
@@ -163,6 +156,40 @@ impl QueriesExecutorTasksQueue {
             if worker_id == workers_tasks.next_tasks.workers_sync_tasks.len() {
                 worker_id = 0;
             }
+
+            if workers_tasks.workers_waiting_status.is_waiting(worker_id) {
+                workers_tasks
+                    .workers_waiting_status
+                    .wakeup_worker(worker_id);
+                condvar.wakeup(worker_id);
+            }
+        }
+    }
+
+    pub fn init_async_tasks(
+        &self,
+        tasks: VecDeque<ProcessorWrapper>,
+        condvar: Arc<WorkersCondvar>,
+    ) {
+        let mut workers_tasks = self.workers_tasks.lock();
+
+        let mut worker_id = 0;
+        for proc in tasks.into_iter() {
+            workers_tasks
+                .next_tasks
+                .push_task(worker_id, ExecutorTask::Async(proc));
+
+            worker_id += 1;
+            if worker_id == workers_tasks.next_tasks.workers_sync_tasks.len() {
+                worker_id = 0;
+            }
+
+            if workers_tasks.workers_waiting_status.is_waiting(worker_id) {
+                workers_tasks
+                    .workers_waiting_status
+                    .wakeup_worker(worker_id);
+                condvar.wakeup(worker_id);
+            }
         }
     }
 

src/query/service/src/pipelines/executor/queries_pipeline_executor.rs

@@ -12,7 +12,6 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-use std::collections::VecDeque;
 use std::sync::atomic::AtomicU32;
 use std::sync::atomic::Ordering;
 use std::sync::Arc;
@@ -31,8 +30,6 @@ use minitrace::full_name;
 use minitrace::prelude::*;
 use parking_lot::Mutex;
 
-use crate::pipelines::executor::ExecutorSettings;
-use crate::pipelines::executor::ExecutorTask;
 use crate::pipelines::executor::ExecutorWorkerContext;
 use crate::pipelines::executor::QueriesExecutorTasksQueue;
 use crate::pipelines::executor::RunningGraph;
@@ -51,9 +48,7 @@ pub struct QueriesPipelineExecutor {
 }
 
 impl QueriesPipelineExecutor {
-    pub fn create(settings: ExecutorSettings) -> Result<Arc<QueriesPipelineExecutor>> {
-        let threads_num = settings.max_threads as usize;
-
+    pub fn create(threads_num: usize) -> Result<Arc<QueriesPipelineExecutor>> {
         let workers_condvar = WorkersCondvar::create(threads_num);
         let global_tasks_queue = QueriesExecutorTasksQueue::create(threads_num);
 
@@ -97,22 +92,13 @@ impl QueriesPipelineExecutor {
         unsafe {
             let mut init_schedule_queue = graph.init_schedule_queue(self.threads_num)?;
 
-            let mut wakeup_worker_id = 0;
-            while let Some(proc) = init_schedule_queue.async_queue.pop_front() {
-                let mut tasks = VecDeque::with_capacity(1);
-                tasks.push_back(ExecutorTask::Async(proc));
-                self.global_tasks_queue
-                    .push_tasks(wakeup_worker_id, None, tasks);
-
-                wakeup_worker_id += 1;
-                if wakeup_worker_id == self.threads_num {
-                    wakeup_worker_id = 0;
-                }
-            }
+            let async_queue = std::mem::take(&mut init_schedule_queue.async_queue);
+            self.global_tasks_queue
+                .init_async_tasks(async_queue, self.workers_condvar.clone());
 
             let sync_queue = std::mem::take(&mut init_schedule_queue.sync_queue);
-            self.global_tasks_queue.init_sync_tasks(sync_queue);
-            self.execute()?;
+            self.global_tasks_queue
+                .init_sync_tasks(sync_queue, self.workers_condvar.clone());
             Ok(())
         }
     }
@@ -203,10 +189,6 @@ impl QueriesPipelineExecutor {
                                 }
                             }
                         }
-                        if graph.is_should_finish() {
-                            // TODO: temporary finish method, will remove after change executor to a global service
-                            self.finish(None);
-                        }
                         if graph.is_all_nodes_finished() {
                             graph.should_finish(Ok(()))?;
                         }
@@ -217,7 +199,6 @@ impl QueriesPipelineExecutor {
                         if let Some(graph) = graph {
                             graph.should_finish(Err(cause.clone()))?;
                         }
-                        self.finish(Some(cause));
                     }
                 }
             }

src/query/service/tests/it/pipelines/executor/executor_graph.rs

@@ -495,7 +495,7 @@ async fn create_executor_with_simple_pipeline(
     let settings = ExecutorSettings {
         query_id: Arc::new("".to_string()),
         max_execute_time_in_seconds: Default::default(),
-        enable_new_executor: false,
+        enable_queries_executor: false,
         max_threads: 8,
         executor_node_id: "".to_string(),
     };

src/query/service/tests/it/pipelines/executor/pipeline_executor.rs

@@ -43,7 +43,7 @@ async fn test_always_call_on_finished() -> Result<()> {
     let settings = ExecutorSettings {
         query_id: Arc::new("".to_string()),
         max_execute_time_in_seconds: Default::default(),
-        enable_new_executor: false,
+        enable_queries_executor: false,
         max_threads: 8,
         executor_node_id: "".to_string(),
     };

src/query/service/tests/it/storages/fuse/operations/truncate.rs

@@ -188,7 +188,8 @@ async fn truncate_table(ctx: Arc<QueryContext>, table: Arc<dyn Table>) -> Result
     table.truncate(ctx.clone(), &mut pipeline).await?;
     if !pipeline.is_empty() {
         pipeline.set_max_threads(1);
-        let executor_settings = ExecutorSettings::try_create(ctx.clone())?;
+        let mut executor_settings = ExecutorSettings::try_create(ctx.clone())?;
+        executor_settings.enable_queries_executor = false;
         let executor = PipelineCompleteExecutor::try_create(pipeline, executor_settings)?;
         ctx.set_executor(executor.get_inner())?;
         executor.execute()?;