[6983] | 1 | #region License Information
|
---|
| 2 | /* HeuristicLab
|
---|
[16565] | 3 | * Copyright (C) 2002-2019 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
|
---|
[6983] | 4 | *
|
---|
| 5 | * This file is part of HeuristicLab.
|
---|
| 6 | *
|
---|
| 7 | * HeuristicLab is free software: you can redistribute it and/or modify
|
---|
| 8 | * it under the terms of the GNU General Public License as published by
|
---|
| 9 | * the Free Software Foundation, either version 3 of the License, or
|
---|
| 10 | * (at your option) any later version.
|
---|
| 11 | *
|
---|
| 12 | * HeuristicLab is distributed in the hope that it will be useful,
|
---|
| 13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
| 14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
| 15 | * GNU General Public License for more details.
|
---|
| 16 | *
|
---|
| 17 | * You should have received a copy of the GNU General Public License
|
---|
| 18 | * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
|
---|
| 19 | */
|
---|
| 20 | #endregion
|
---|
| 21 |
|
---|
| 22 | using System;
|
---|
| 23 | using System.Threading;
|
---|
| 24 | using HeuristicLab.Clients.Hive.SlaveCore.Properties;
|
---|
| 25 | using HeuristicLab.Common;
|
---|
| 26 | using HeuristicLab.Core;
|
---|
| 27 | using HeuristicLab.Hive;
|
---|
| 28 |
|
---|
| 29 | namespace HeuristicLab.Clients.Hive.SlaveCore {
|
---|
| 30 | /// <summary>
|
---|
| 31 | /// The executor runs in the appdomain and handles the execution of an Hive task.
|
---|
| 32 | /// </summary>
|
---|
| 33 | public class Executor : MarshalByRefObject, IDisposable {
|
---|
| 34 | private bool wasTaskAborted = false;
|
---|
| 35 | private AutoResetEvent pauseStopSem = new AutoResetEvent(false);
|
---|
| 36 | private AutoResetEvent startTaskSem = new AutoResetEvent(false); // block start method call
|
---|
| 37 | private AutoResetEvent taskStartedSem = new AutoResetEvent(false); // make pause or stop wait until start is finished
|
---|
| 38 | private ExecutorQueue executorQueue;
|
---|
| 39 | private bool taskDataInvalid = false; // if true, the jobdata is not sent when the task is failed
|
---|
| 40 | private ITask task;
|
---|
| 41 | private DateTime creationTime;
|
---|
| 42 |
|
---|
| 43 | public Guid TaskId { get; set; }
|
---|
| 44 | public int CoresNeeded { get; set; }
|
---|
| 45 | public int MemoryNeeded { get; set; }
|
---|
| 46 | public bool IsStopping { get; set; }
|
---|
| 47 | public bool IsPausing { get; set; }
|
---|
| 48 |
|
---|
| 49 | public Exception CurrentException;
|
---|
| 50 | public String CurrentExceptionStr {
|
---|
| 51 | get {
|
---|
| 52 | if (CurrentException != null) {
|
---|
| 53 | return CurrentException.ToString();
|
---|
| 54 | } else {
|
---|
| 55 | return string.Empty;
|
---|
| 56 | }
|
---|
| 57 | }
|
---|
| 58 | }
|
---|
| 59 |
|
---|
| 60 | public ExecutorQueue ExecutorCommandQueue {
|
---|
| 61 | get { return executorQueue; }
|
---|
| 62 | }
|
---|
| 63 |
|
---|
| 64 | private ExecutionState ExecutionState {
|
---|
| 65 | get { return task != null ? task.ExecutionState : HeuristicLab.Core.ExecutionState.Stopped; }
|
---|
| 66 | }
|
---|
| 67 |
|
---|
| 68 | public TimeSpan ExecutionTime {
|
---|
| 69 | get { return task != null ? task.ExecutionTime : new TimeSpan(0, 0, 0); }
|
---|
| 70 | }
|
---|
| 71 |
|
---|
| 72 | public Executor() {
|
---|
| 73 | IsStopping = false;
|
---|
| 74 | IsPausing = false;
|
---|
| 75 | executorQueue = new ExecutorQueue();
|
---|
| 76 | }
|
---|
| 77 |
|
---|
| 78 | public void Start(byte[] serializedJob) {
|
---|
| 79 | try {
|
---|
| 80 | creationTime = DateTime.Now;
|
---|
| 81 | task = PersistenceUtil.Deserialize<ITask>(serializedJob);
|
---|
| 82 |
|
---|
| 83 | RegisterJobEvents();
|
---|
| 84 |
|
---|
| 85 | task.Start();
|
---|
| 86 | if (!startTaskSem.WaitOne(Settings.Default.ExecutorSemTimeouts)) {
|
---|
| 87 | throw new TimeoutException("Timeout when starting the task. TaskStarted event was not fired.");
|
---|
| 88 | }
|
---|
| 89 | }
|
---|
| 90 | catch (Exception e) {
|
---|
[11082] | 91 | HandleStartStopPauseError(e);
|
---|
| 92 | }
|
---|
| 93 | finally {
|
---|
[6983] | 94 | taskStartedSem.Set();
|
---|
| 95 | }
|
---|
| 96 | }
|
---|
| 97 |
|
---|
| 98 | public void Pause() {
|
---|
| 99 | IsPausing = true;
|
---|
| 100 | // wait until task is started. if this does not happen, the Task is null an we give up
|
---|
| 101 | taskStartedSem.WaitOne(Settings.Default.ExecutorSemTimeouts);
|
---|
| 102 | if (task == null) {
|
---|
[11082] | 103 | HandleStartStopPauseError(new Exception("Pausing task " + this.TaskId + ": Task is null"));
|
---|
[6983] | 104 | return;
|
---|
| 105 | }
|
---|
| 106 |
|
---|
| 107 | if (task.ExecutionState == ExecutionState.Started) {
|
---|
| 108 | try {
|
---|
| 109 | task.Pause();
|
---|
| 110 | //we need to block the pause...
|
---|
[11082] | 111 | if (!pauseStopSem.WaitOne(Settings.Default.ExecutorSemTimeouts)) {
|
---|
| 112 | throw new Exception("Pausing task " + this.TaskId + " timed out.");
|
---|
| 113 | }
|
---|
[6983] | 114 | }
|
---|
| 115 | catch (Exception ex) {
|
---|
[11082] | 116 | HandleStartStopPauseError(ex);
|
---|
[6983] | 117 | }
|
---|
| 118 | }
|
---|
| 119 | }
|
---|
| 120 |
|
---|
| 121 | public void Stop() {
|
---|
| 122 | IsStopping = true;
|
---|
| 123 | // wait until task is started. if this does not happen, the Task is null an we give up
|
---|
| 124 | taskStartedSem.WaitOne(Settings.Default.ExecutorSemTimeouts);
|
---|
[11082] | 125 | wasTaskAborted = true;
|
---|
| 126 |
|
---|
[6983] | 127 | if (task == null) {
|
---|
[11082] | 128 | HandleStartStopPauseError(new Exception("Stopping task " + this.TaskId + ": Task is null"));
|
---|
| 129 | return;
|
---|
[6983] | 130 | }
|
---|
| 131 |
|
---|
| 132 | if ((ExecutionState == ExecutionState.Started) || (ExecutionState == ExecutionState.Paused)) {
|
---|
| 133 | try {
|
---|
| 134 | task.Stop();
|
---|
[11082] | 135 | if (!pauseStopSem.WaitOne(Settings.Default.ExecutorSemTimeouts)) {
|
---|
| 136 | throw new Exception("Stopping task " + this.TaskId + " timed out.");
|
---|
| 137 | }
|
---|
[6983] | 138 | }
|
---|
| 139 | catch (Exception ex) {
|
---|
[11082] | 140 | HandleStartStopPauseError(ex);
|
---|
[6983] | 141 | }
|
---|
| 142 | }
|
---|
| 143 | }
|
---|
| 144 |
|
---|
| 145 | private void RegisterJobEvents() {
|
---|
| 146 | task.TaskStopped += new EventHandler(Task_TaskStopped);
|
---|
| 147 | task.TaskFailed += new EventHandler(Task_TaskFailed);
|
---|
| 148 | task.TaskPaused += new EventHandler(Task_TaskPaused);
|
---|
| 149 | task.TaskStarted += new EventHandler(Task_TaskStarted);
|
---|
| 150 | }
|
---|
| 151 |
|
---|
| 152 | private void DeregisterJobEvents() {
|
---|
| 153 | task.TaskStopped -= new EventHandler(Task_TaskStopped);
|
---|
| 154 | task.TaskFailed -= new EventHandler(Task_TaskFailed);
|
---|
| 155 | task.TaskPaused -= new EventHandler(Task_TaskPaused);
|
---|
| 156 | task.TaskStarted -= new EventHandler(Task_TaskStarted);
|
---|
| 157 | }
|
---|
| 158 |
|
---|
| 159 | #region Task Events
|
---|
| 160 | private void Task_TaskFailed(object sender, EventArgs e) {
|
---|
| 161 | IsStopping = true;
|
---|
| 162 | EventArgs<Exception> ex = (EventArgs<Exception>)e;
|
---|
| 163 | CurrentException = ex.Value;
|
---|
| 164 | executorQueue.AddMessage(ExecutorMessageType.TaskFailed);
|
---|
| 165 | }
|
---|
| 166 |
|
---|
| 167 | private void Task_TaskStopped(object sender, EventArgs e) {
|
---|
| 168 | IsStopping = true;
|
---|
| 169 | if (wasTaskAborted) {
|
---|
| 170 | pauseStopSem.Set();
|
---|
| 171 | }
|
---|
| 172 | executorQueue.AddMessage(ExecutorMessageType.TaskStopped);
|
---|
| 173 | }
|
---|
| 174 |
|
---|
| 175 | private void Task_TaskPaused(object sender, EventArgs e) {
|
---|
| 176 | IsPausing = true;
|
---|
| 177 | pauseStopSem.Set();
|
---|
| 178 | executorQueue.AddMessage(ExecutorMessageType.TaskPaused);
|
---|
| 179 | }
|
---|
| 180 |
|
---|
| 181 | private void Task_TaskStarted(object sender, EventArgs e) {
|
---|
| 182 | startTaskSem.Set();
|
---|
| 183 | executorQueue.AddMessage(ExecutorMessageType.TaskStarted);
|
---|
| 184 | }
|
---|
| 185 | #endregion
|
---|
| 186 |
|
---|
| 187 | public TaskData GetTaskData() {
|
---|
| 188 | if (taskDataInvalid) return null;
|
---|
| 189 |
|
---|
[7166] | 190 | if (task != null && task.ExecutionState == ExecutionState.Started) {
|
---|
[6983] | 191 | throw new InvalidStateException("Task is still running");
|
---|
[11082] | 192 | }
|
---|
| 193 |
|
---|
| 194 | TaskData taskData = null;
|
---|
| 195 | if (task == null) {
|
---|
| 196 | if (CurrentException == null) {
|
---|
| 197 | CurrentException = new Exception("Task with id " + this.TaskId + " is null, sending empty task");
|
---|
| 198 | }
|
---|
[6983] | 199 | } else {
|
---|
[11082] | 200 | taskData = new TaskData();
|
---|
| 201 | taskData.Data = PersistenceUtil.Serialize(task);
|
---|
[6983] | 202 | taskData.TaskId = TaskId;
|
---|
| 203 | }
|
---|
[11082] | 204 | return taskData;
|
---|
[6983] | 205 | }
|
---|
| 206 |
|
---|
| 207 | public void Dispose() {
|
---|
| 208 | if (task != null)
|
---|
| 209 | DeregisterJobEvents();
|
---|
| 210 | task = null;
|
---|
| 211 | }
|
---|
[11082] | 212 |
|
---|
| 213 | private void HandleStartStopPauseError(Exception e) {
|
---|
| 214 | taskDataInvalid = true;
|
---|
| 215 | Task_TaskFailed(this, new EventArgs<Exception>(e));
|
---|
| 216 | }
|
---|
[6983] | 217 | }
|
---|
| 218 | }
|
---|