Free cookie consent management tool by TermsFeed Policy Generator

source: branches/HeuristicLab.Hive-3.4/sources/HeuristicLab.Services.Hive/3.3/Manager/EventManager.cs @ 6946

Last change on this file since 6946 was 6946, checked in by ascheibe, 11 years ago

#1233

  • disable logging of the user statistics on the server because of high run time demands
  • also show in the slave UI the timestamps of arrived messages
File size: 5.0 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      log.Log("EventManager.Cleanup()");
46      SetTimeoutSlavesOffline();
47      SetTimeoutTasksWaiting();
48      FinishParentTasks();
49      UpdateStatistics();
50    }
51
52    private void UpdateStatistics() {
53      var slaves = dao.GetSlaves(x => x.SlaveState == SlaveState.Calculating || x.SlaveState == SlaveState.Idle);
54
55      var stats = new DataTransfer.Statistics();
56      stats.TimeStamp = DateTime.Now;
57      var slaveStats = new List<DT.SlaveStatistics>();
58      foreach (var slave in slaves) {
59        slaveStats.Add(new DT.SlaveStatistics() {
60          SlaveId = slave.Id,
61          Cores = slave.Cores.HasValue ? slave.Cores.Value : 0,
62          FreeCores = slave.FreeCores.HasValue ? slave.FreeCores.Value : 0,
63          Memory = slave.Memory.HasValue ? slave.Memory.Value : 0,
64          FreeMemory = slave.FreeMemory.HasValue ? slave.FreeMemory.Value : 0,
65          CpuUtilization = slave.CpuUtilization
66        });
67      }
68      stats.SlaveStatistics = slaveStats;
69      //collecting user statistics slows down the db and results in timeouts.
70      //we have to find another way to deal with this. 
71      //until then the next line is commented out...
72      //stats.UserStatistics = dao.GetUserStatistics();
73      dao.AddStatistics(stats);
74    }
75
76    /// <summary>
77    /// Searches for slaves which are timed out, puts them and their task offline
78    /// </summary>
79    private void SetTimeoutSlavesOffline() {
80      var slaves = dao.GetSlaves(x => x.SlaveState != SlaveState.Offline);
81      foreach (DT.Slave slave in slaves) {
82        if (!slave.LastHeartbeat.HasValue || (DateTime.Now - slave.LastHeartbeat.Value) > HeuristicLab.Services.Hive.Properties.Settings.Default.SlaveHeartbeatTimeout) {
83          slave.SlaveState = DT.SlaveState.Offline;
84          SetTasksWaiting(slave.Id);
85          dao.UpdateSlave(slave);
86        }
87      }
88    }
89
90    /// <summary>
91    /// Looks for parent tasks which have FinishWhenChildJobsFinished and set their state to finished
92    /// </summary>
93    private void FinishParentTasks() {
94      var parentTasksToFinish = dao.GetParentTasks(dao.GetResources(x => true).Select(x => x.Id), 0, true);
95      foreach (var task in parentTasksToFinish) {
96        dao.UpdateTaskState(task.Id, TaskState.Finished, null, null, string.Empty);
97      }
98    }
99
100    private void SetTasksWaiting(Guid slaveId) {
101      var tasks = dao.GetTasks(x => x.State == TaskState.Calculating).Where(x => x.StateLog.Last().SlaveId == slaveId);
102      foreach (var j in tasks) {
103        DT.Task task = dao.UpdateTaskState(j.Id, TaskState.Waiting, slaveId, null, "Slave timed out.");
104        task.Command = null;
105        dao.UpdateTask(task);
106      }
107    }
108
109    /// <summary>
110    /// Looks for task which have not sent heartbeats for some time and reschedules them for calculation
111    /// </summary>
112    private void SetTimeoutTasksWaiting() {
113      var tasks = dao.GetTasks(x => (x.State == TaskState.Calculating && (DateTime.Now - x.LastHeartbeat) > HeuristicLab.Services.Hive.Properties.Settings.Default.CalculatingJobHeartbeatTimeout)
114                               || (x.State == TaskState.Transferring && (DateTime.Now - x.LastHeartbeat) > HeuristicLab.Services.Hive.Properties.Settings.Default.TransferringJobHeartbeatTimeout));
115      foreach (var j in tasks) {
116        DT.Task task = dao.UpdateTaskState(j.Id, TaskState.Waiting, null, null, "Slave timed out.");
117        task.Command = null;
118        dao.UpdateTask(task);
119      }
120    }
121  }
122}
Note: See TracBrowser for help on using the repository browser.