/root/doris/be/src/pipeline/pipeline_task.h
Line | Count | Source (jump to first uncovered line) |
1 | | // Licensed to the Apache Software Foundation (ASF) under one |
2 | | // or more contributor license agreements. See the NOTICE file |
3 | | // distributed with this work for additional information |
4 | | // regarding copyright ownership. The ASF licenses this file |
5 | | // to you under the Apache License, Version 2.0 (the |
6 | | // "License"); you may not use this file except in compliance |
7 | | // with the License. You may obtain a copy of the License at |
8 | | // |
9 | | // http://www.apache.org/licenses/LICENSE-2.0 |
10 | | // |
11 | | // Unless required by applicable law or agreed to in writing, |
12 | | // software distributed under the License is distributed on an |
13 | | // "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY |
14 | | // KIND, either express or implied. See the License for the |
15 | | // specific language governing permissions and limitations |
16 | | // under the License. |
17 | | |
18 | | #pragma once |
19 | | |
20 | | #include <stdint.h> |
21 | | |
22 | | #include <memory> |
23 | | #include <string> |
24 | | #include <vector> |
25 | | |
26 | | #include "common/status.h" |
27 | | #include "pipeline/dependency.h" |
28 | | #include "pipeline/exec/operator.h" |
29 | | #include "pipeline/pipeline.h" |
30 | | #include "util/runtime_profile.h" |
31 | | #include "util/stopwatch.hpp" |
32 | | #include "vec/core/block.h" |
33 | | |
34 | | namespace doris { |
35 | | class QueryContext; |
36 | | class RuntimeState; |
37 | | namespace pipeline { |
38 | | class PipelineFragmentContext; |
39 | | } // namespace pipeline |
40 | | } // namespace doris |
41 | | |
42 | | namespace doris::pipeline { |
43 | | |
44 | | class MultiCoreTaskQueue; |
45 | | class PriorityTaskQueue; |
46 | | class Dependency; |
47 | | |
48 | | class PipelineTask { |
49 | | public: |
50 | | PipelineTask(PipelinePtr& pipeline, uint32_t task_id, RuntimeState* state, |
51 | | PipelineFragmentContext* fragment_context, RuntimeProfile* parent_profile, |
52 | | std::map<int, std::pair<std::shared_ptr<LocalExchangeSharedState>, |
53 | | std::shared_ptr<Dependency>>> |
54 | | le_state_map, |
55 | | int task_idx); |
56 | | |
57 | | Status prepare(const TPipelineInstanceParams& local_params, const TDataSink& tsink, |
58 | | QueryContext* query_ctx); |
59 | | |
60 | | Status execute(bool* eos); |
61 | | |
62 | | // if the pipeline create a bunch of pipeline task |
63 | | // must be call after all pipeline task is finish to release resource |
64 | | Status close(Status exec_status); |
65 | | |
66 | 0 | PipelineFragmentContext* fragment_context() { return _fragment_context; } |
67 | | |
68 | | QueryContext* query_context(); |
69 | | |
70 | 0 | int get_previous_core_id() const { |
71 | 0 | return _previous_schedule_id != -1 ? _previous_schedule_id |
72 | 0 | : _pipeline->_previous_schedule_id; |
73 | 0 | } |
74 | | |
75 | 0 | void set_previous_core_id(int id) { |
76 | 0 | if (id != _previous_schedule_id) { |
77 | 0 | if (_previous_schedule_id != -1) { |
78 | 0 | COUNTER_UPDATE(_core_change_times, 1); |
79 | 0 | } |
80 | 0 | _previous_schedule_id = id; |
81 | 0 | } |
82 | 0 | } |
83 | | |
84 | | void finalize(); |
85 | | |
86 | | std::string debug_string(); |
87 | | |
88 | 0 | bool is_pending_finish() { |
89 | 0 | for (auto* fin_dep : _finish_dependencies) { |
90 | 0 | _blocked_dep = fin_dep->is_blocked_by(this); |
91 | 0 | if (_blocked_dep != nullptr) { |
92 | 0 | _blocked_dep->start_watcher(); |
93 | 0 | return true; |
94 | 0 | } |
95 | 0 | } |
96 | 0 | return false; |
97 | 0 | } |
98 | | |
99 | 0 | std::shared_ptr<BasicSharedState> get_source_shared_state() { |
100 | 0 | return _op_shared_states.contains(_source->operator_id()) |
101 | 0 | ? _op_shared_states[_source->operator_id()] |
102 | 0 | : nullptr; |
103 | 0 | } |
104 | | |
105 | 0 | void inject_shared_state(std::shared_ptr<BasicSharedState> shared_state) { |
106 | 0 | if (!shared_state) { |
107 | 0 | return; |
108 | 0 | } |
109 | | // Shared state is created by upstream task's sink operator and shared by source operator of this task. |
110 | 0 | for (auto& op : _operators) { |
111 | 0 | if (shared_state->related_op_ids.contains(op->operator_id())) { |
112 | 0 | _op_shared_states.insert({op->operator_id(), shared_state}); |
113 | 0 | return; |
114 | 0 | } |
115 | 0 | } |
116 | 0 | if (shared_state->related_op_ids.contains(_sink->dests_id().front())) { |
117 | 0 | DCHECK(_sink_shared_state == nullptr); |
118 | 0 | _sink_shared_state = shared_state; |
119 | 0 | } |
120 | 0 | } |
121 | | |
122 | 0 | std::shared_ptr<BasicSharedState> get_sink_shared_state() { return _sink_shared_state; } |
123 | | |
124 | 0 | BasicSharedState* get_op_shared_state(int id) { |
125 | 0 | if (!_op_shared_states.contains(id)) { |
126 | 0 | return nullptr; |
127 | 0 | } |
128 | 0 | return _op_shared_states[id].get(); |
129 | 0 | } |
130 | | |
131 | | void wake_up(); |
132 | | |
133 | 0 | DataSinkOperatorPtr sink() const { return _sink; } |
134 | | |
135 | 0 | int task_id() const { return _index; }; |
136 | 0 | bool is_finalized() const { return _finalized; } |
137 | | |
138 | 0 | void clear_blocking_state(bool wake_up_by_downstream = false) { |
139 | 0 | _state->get_query_ctx()->get_execution_dependency()->set_always_ready(); |
140 | | // We use a lock to assure all dependencies are not deconstructed here. |
141 | 0 | std::unique_lock<std::mutex> lc(_dependency_lock); |
142 | 0 | _wake_up_by_downstream = _wake_up_by_downstream || wake_up_by_downstream; |
143 | 0 | if (!_finalized) { |
144 | 0 | _execution_dep->set_always_ready(); |
145 | 0 | for (auto* dep : _filter_dependencies) { |
146 | 0 | dep->set_always_ready(); |
147 | 0 | } |
148 | 0 | for (auto& deps : _read_dependencies) { |
149 | 0 | for (auto* dep : deps) { |
150 | 0 | dep->set_always_ready(); |
151 | 0 | } |
152 | 0 | } |
153 | 0 | for (auto* dep : _write_dependencies) { |
154 | 0 | dep->set_always_ready(); |
155 | 0 | } |
156 | 0 | for (auto* dep : _finish_dependencies) { |
157 | 0 | dep->set_always_ready(); |
158 | 0 | } |
159 | 0 | } |
160 | 0 | } |
161 | | |
162 | 0 | void set_task_queue(MultiCoreTaskQueue* task_queue) { _task_queue = task_queue; } |
163 | 0 | MultiCoreTaskQueue* get_task_queue() { return _task_queue; } |
164 | | |
165 | | static constexpr auto THREAD_TIME_SLICE = 100'000'000ULL; |
166 | | |
167 | | // 1 used for update priority queue |
168 | | // note(wb) an ugly implementation, need refactor later |
169 | | // 1.1 pipeline task |
170 | 0 | void inc_runtime_ns(uint64_t delta_time) { this->_runtime += delta_time; } |
171 | 0 | uint64_t get_runtime_ns() const { return this->_runtime; } |
172 | | |
173 | | // 1.2 priority queue's queue level |
174 | 0 | void update_queue_level(int queue_level) { this->_queue_level = queue_level; } |
175 | 0 | int get_queue_level() const { return this->_queue_level; } |
176 | | |
177 | | // 1.3 priority queue's core id |
178 | 0 | void set_core_id(int core_id) { this->_core_id = core_id; } |
179 | 0 | int get_core_id() const { return this->_core_id; } |
180 | | |
181 | | /** |
182 | | * Return true if: |
183 | | * 1. `enable_force_spill` is true which forces this task to spill data. |
184 | | * 2. Or memory consumption reaches the high water mark of current workload group (80% of memory limitation by default) and revocable_mem_bytes is bigger than min_revocable_mem_bytes. |
185 | | * 3. Or memory consumption is higher than the low water mark of current workload group (50% of memory limitation by default) and `query_weighted_consumption >= query_weighted_limit` and revocable memory is big enough. |
186 | | */ |
187 | | static bool should_revoke_memory(RuntimeState* state, int64_t revocable_mem_bytes); |
188 | | |
189 | 0 | void put_in_runnable_queue() { |
190 | 0 | _schedule_time++; |
191 | 0 | _wait_worker_watcher.start(); |
192 | 0 | } |
193 | | |
194 | 0 | void pop_out_runnable_queue() { _wait_worker_watcher.stop(); } |
195 | | |
196 | 0 | bool is_running() { return _running.load(); } |
197 | 0 | void set_running(bool running) { _running = running; } |
198 | | |
199 | 0 | bool is_exceed_debug_timeout() { |
200 | 0 | if (_has_exceed_timeout) { |
201 | 0 | return true; |
202 | 0 | } |
203 | | // If enable_debug_log_timeout_secs <= 0, then disable the log |
204 | 0 | if (_pipeline_task_watcher.elapsed_time() > |
205 | 0 | config::enable_debug_log_timeout_secs * 1000L * 1000L * 1000L) { |
206 | 0 | _has_exceed_timeout = true; |
207 | 0 | return true; |
208 | 0 | } |
209 | 0 | return false; |
210 | 0 | } |
211 | | |
212 | 0 | void log_detail_if_need() { |
213 | 0 | if (config::enable_debug_log_timeout_secs < 1) { |
214 | 0 | return; |
215 | 0 | } |
216 | 0 | if (is_exceed_debug_timeout()) { |
217 | 0 | LOG(INFO) << "query id|instanceid " << print_id(_state->query_id()) << "|" |
218 | 0 | << print_id(_state->fragment_instance_id()) |
219 | 0 | << " current pipeline exceed run time " |
220 | 0 | << config::enable_debug_log_timeout_secs << " seconds. " |
221 | 0 | << "/n task detail:" << debug_string(); |
222 | 0 | } |
223 | 0 | } |
224 | | |
225 | 0 | RuntimeState* runtime_state() const { return _state; } |
226 | | |
227 | 0 | RuntimeProfile* get_task_profile() const { return _task_profile.get(); } |
228 | | |
229 | 0 | std::string task_name() const { return fmt::format("task{}({})", _index, _pipeline->_name); } |
230 | | |
231 | 0 | void stop_if_finished() { |
232 | 0 | if (_sink->is_finished(_state)) { |
233 | 0 | clear_blocking_state(); |
234 | 0 | } |
235 | 0 | } |
236 | | |
237 | 0 | PipelineId pipeline_id() const { return _pipeline->id(); } |
238 | | |
239 | 0 | bool wake_up_by_downstream() const { return _wake_up_by_downstream; } |
240 | | |
241 | | private: |
242 | | friend class RuntimeFilterDependency; |
243 | | bool _is_blocked(); |
244 | | bool _wait_to_start(); |
245 | | |
246 | | Status _extract_dependencies(); |
247 | | void _init_profile(); |
248 | | void _fresh_profile_counter(); |
249 | | Status _open(); |
250 | | |
251 | | uint32_t _index; |
252 | | PipelinePtr _pipeline; |
253 | | bool _has_exceed_timeout = false; |
254 | | bool _opened; |
255 | | RuntimeState* _state = nullptr; |
256 | | int _previous_schedule_id = -1; |
257 | | uint32_t _schedule_time = 0; |
258 | | std::unique_ptr<doris::vectorized::Block> _block; |
259 | | PipelineFragmentContext* _fragment_context = nullptr; |
260 | | MultiCoreTaskQueue* _task_queue = nullptr; |
261 | | |
262 | | // used for priority queue |
263 | | // it may be visited by different thread but there is no race condition |
264 | | // so no need to add lock |
265 | | uint64_t _runtime = 0; |
266 | | // it's visited in one thread, so no need to thread synchronization |
267 | | // 1 get task, (set _queue_level/_core_id) |
268 | | // 2 exe task |
269 | | // 3 update task statistics(update _queue_level/_core_id) |
270 | | int _queue_level = 0; |
271 | | int _core_id = 0; |
272 | | |
273 | | RuntimeProfile* _parent_profile = nullptr; |
274 | | std::unique_ptr<RuntimeProfile> _task_profile; |
275 | | RuntimeProfile::Counter* _task_cpu_timer = nullptr; |
276 | | RuntimeProfile::Counter* _prepare_timer = nullptr; |
277 | | RuntimeProfile::Counter* _open_timer = nullptr; |
278 | | RuntimeProfile::Counter* _exec_timer = nullptr; |
279 | | RuntimeProfile::Counter* _get_block_timer = nullptr; |
280 | | RuntimeProfile::Counter* _get_block_counter = nullptr; |
281 | | RuntimeProfile::Counter* _sink_timer = nullptr; |
282 | | RuntimeProfile::Counter* _close_timer = nullptr; |
283 | | RuntimeProfile::Counter* _schedule_counts = nullptr; |
284 | | MonotonicStopWatch _wait_worker_watcher; |
285 | | RuntimeProfile::Counter* _wait_worker_timer = nullptr; |
286 | | // TODO we should calculate the time between when really runnable and runnable |
287 | | RuntimeProfile::Counter* _yield_counts = nullptr; |
288 | | RuntimeProfile::Counter* _core_change_times = nullptr; |
289 | | |
290 | | MonotonicStopWatch _pipeline_task_watcher; |
291 | | |
292 | | Operators _operators; // left is _source, right is _root |
293 | | OperatorXBase* _source; |
294 | | OperatorXBase* _root; |
295 | | DataSinkOperatorPtr _sink; |
296 | | |
297 | | // `_read_dependencies` is stored as same order as `_operators` |
298 | | std::vector<std::vector<Dependency*>> _read_dependencies; |
299 | | std::vector<Dependency*> _write_dependencies; |
300 | | std::vector<Dependency*> _finish_dependencies; |
301 | | std::vector<Dependency*> _filter_dependencies; |
302 | | |
303 | | // All shared states of this pipeline task. |
304 | | std::map<int, std::shared_ptr<BasicSharedState>> _op_shared_states; |
305 | | std::shared_ptr<BasicSharedState> _sink_shared_state; |
306 | | std::vector<TScanRangeParams> _scan_ranges; |
307 | | std::map<int, std::pair<std::shared_ptr<LocalExchangeSharedState>, std::shared_ptr<Dependency>>> |
308 | | _le_state_map; |
309 | | int _task_idx; |
310 | | bool _dry_run = false; |
311 | | |
312 | | Dependency* _blocked_dep = nullptr; |
313 | | |
314 | | Dependency* _execution_dep = nullptr; |
315 | | |
316 | | std::atomic<bool> _finalized = false; |
317 | | std::mutex _dependency_lock; |
318 | | |
319 | | std::atomic<bool> _running = false; |
320 | | std::atomic<bool> _eos = false; |
321 | | std::atomic<bool> _wake_up_by_downstream = false; |
322 | | }; |
323 | | |
324 | | } // namespace doris::pipeline |