Free cookie consent management tool by TermsFeed Policy Generator

source: trunk/sources/HeuristicLab.Services.Hive/3.3/Manager/EventManager.cs @ 7178

Last change on this file since 7178 was 7178, checked in by ascheibe, 12 years ago

#1672 disabled checking if there are parent tasks which should to be calculated because this case doesn't exist at the moment

File size: 4.9 KB
Line 
1#region License Information
2/* HeuristicLab
3 * Copyright (C) 2002-2011 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
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
22using System;
23using System.Collections.Generic;
24using System.Linq;
25using HeuristicLab.Services.Hive.DataAccess;
26using DT = HeuristicLab.Services.Hive.DataTransfer;
27
28
29namespace HeuristicLab.Services.Hive {
30  /// <summary>
31  /// This class offers methods for cleaning up offline slaves and task
32  /// </summary>
33  public class EventManager : IEventManager {
34    private IHiveDao dao {
35      get { return ServiceLocator.Instance.HiveDao; }
36    }
37    private IAuthorizationManager auth {
38      get { return ServiceLocator.Instance.AuthorizationManager; }
39    }
40    private ILogger log {
41      get { return LogFactory.GetLogger(this.GetType().Namespace); }
42    }
43
44    public void Cleanup() {
45      SetTimeoutSlavesOffline();
46      SetTimeoutTasksWaiting();
47      FinishParentTasks();
48      UpdateStatistics();
49    }
50
51    private void UpdateStatistics() {
52      var slaves = dao.GetSlaves(x => x.SlaveState == SlaveState.Calculating || x.SlaveState == SlaveState.Idle);
53
54      var stats = new DataTransfer.Statistics();
55      stats.TimeStamp = DateTime.Now;
56      var slaveStats = new List<DT.SlaveStatistics>();
57      foreach (var slave in slaves) {
58        slaveStats.Add(new DT.SlaveStatistics() {
59          SlaveId = slave.Id,
60          Cores = slave.Cores.HasValue ? slave.Cores.Value : 0,
61          FreeCores = slave.FreeCores.HasValue ? slave.FreeCores.Value : 0,
62          Memory = slave.Memory.HasValue ? slave.Memory.Value : 0,
63          FreeMemory = slave.FreeMemory.HasValue ? slave.FreeMemory.Value : 0,
64          CpuUtilization = slave.CpuUtilization
65        });
66      }
67      stats.SlaveStatistics = slaveStats;
68      //collecting user statistics slows down the db and results in timeouts.
69      //we have to find another way to deal with this. 
70      //until then the next line is commented out...
71      //stats.UserStatistics = dao.GetUserStatistics();
72      dao.AddStatistics(stats);
73    }
74
75    /// <summary>
76    /// Searches for slaves which are timed out, puts them and their task offline
77    /// </summary>
78    private void SetTimeoutSlavesOffline() {
79      var slaves = dao.GetSlaves(x => x.SlaveState != SlaveState.Offline);
80      foreach (DT.Slave slave in slaves) {
81        if (!slave.LastHeartbeat.HasValue || (DateTime.Now - slave.LastHeartbeat.Value) > HeuristicLab.Services.Hive.Properties.Settings.Default.SlaveHeartbeatTimeout) {
82          slave.SlaveState = DT.SlaveState.Offline;
83          SetTasksWaiting(slave.Id);
84          dao.UpdateSlave(slave);
85        }
86      }
87    }
88
89    /// <summary>
90    /// Looks for parent tasks which have FinishWhenChildJobsFinished and set their state to finished
91    /// </summary>
92    private void FinishParentTasks() {
93      var parentTasksToFinish = dao.GetParentTasks(dao.GetResources(x => true).Select(x => x.Id), 0, true);
94      foreach (var task in parentTasksToFinish) {
95        dao.UpdateTaskState(task.Id, TaskState.Finished, null, null, string.Empty);
96      }
97    }
98
99    private void SetTasksWaiting(Guid slaveId) {
100      var tasks = dao.GetTasks(x => x.State == TaskState.Calculating).Where(x => x.StateLog.Last().SlaveId == slaveId);
101      foreach (var j in tasks) {
102        DT.Task task = dao.UpdateTaskState(j.Id, TaskState.Waiting, slaveId, null, "Slave timed out.");
103        task.Command = null;
104        dao.UpdateTask(task);
105      }
106    }
107
108    /// <summary>
109    /// Looks for task which have not sent heartbeats for some time and reschedules them for calculation
110    /// </summary>
111    private void SetTimeoutTasksWaiting() {
112      var tasks = dao.GetTasks(x => (x.State == TaskState.Calculating && (DateTime.Now - x.LastHeartbeat) > HeuristicLab.Services.Hive.Properties.Settings.Default.CalculatingJobHeartbeatTimeout)
113                               || (x.State == TaskState.Transferring && (DateTime.Now - x.LastHeartbeat) > HeuristicLab.Services.Hive.Properties.Settings.Default.TransferringJobHeartbeatTimeout));
114      foreach (var j in tasks) {
115        DT.Task task = dao.UpdateTaskState(j.Id, TaskState.Waiting, null, null, "Slave timed out.");
116        task.Command = null;
117        dao.UpdateTask(task);
118      }
119    }
120  }
121}
Note: See TracBrowser for help on using the repository browser.