[12853] | 1 | #region License Information
|
---|
| 2 | /* HeuristicLab
|
---|
[14185] | 3 | * Copyright (C) 2002-2016 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
|
---|
[12853] | 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;
|
---|
[12789] | 23 | using System.Linq;
|
---|
| 24 | using HeuristicLab.Services.Hive.DataAccess;
|
---|
| 25 | using HeuristicLab.Services.Hive.DataAccess.Interfaces;
|
---|
| 26 |
|
---|
| 27 | namespace HeuristicLab.Services.Hive.Manager {
|
---|
[12861] | 28 | public class EventManager : IEventManager {
|
---|
[12789] | 29 | private const string SlaveTimeout = "Slave timed out.";
|
---|
| 30 | private IPersistenceManager PersistenceManager {
|
---|
| 31 | get { return ServiceLocator.Instance.PersistenceManager; }
|
---|
| 32 | }
|
---|
| 33 |
|
---|
| 34 | public void Cleanup() {
|
---|
[12858] | 35 | var pm = PersistenceManager;
|
---|
| 36 | pm.UseTransaction(() => {
|
---|
| 37 | SetTimeoutSlavesOffline(pm);
|
---|
| 38 | SetTimeoutTasksWaiting(pm);
|
---|
| 39 | DeleteObsoleteSlaves(pm);
|
---|
| 40 | pm.SubmitChanges();
|
---|
| 41 | });
|
---|
[12789] | 42 |
|
---|
[12858] | 43 | pm.UseTransaction(() => {
|
---|
| 44 | FinishParentTasks(pm);
|
---|
| 45 | pm.SubmitChanges();
|
---|
| 46 | });
|
---|
[12789] | 47 | }
|
---|
| 48 |
|
---|
| 49 | /// <summary>
|
---|
| 50 | /// Searches for slaves which are timed out, puts them and their task offline
|
---|
| 51 | /// </summary>
|
---|
| 52 | private void SetTimeoutSlavesOffline(IPersistenceManager pm) {
|
---|
| 53 | var slaveDao = pm.SlaveDao;
|
---|
| 54 | var slaves = slaveDao.GetOnlineSlaves();
|
---|
| 55 | foreach (var slave in slaves) {
|
---|
| 56 | if (!slave.LastHeartbeat.HasValue ||
|
---|
| 57 | (DateTime.Now - slave.LastHeartbeat.Value) >
|
---|
[12866] | 58 | Properties.Settings.Default.SlaveHeartbeatTimeout) {
|
---|
[12789] | 59 | slave.SlaveState = SlaveState.Offline;
|
---|
| 60 | }
|
---|
| 61 | }
|
---|
| 62 | }
|
---|
| 63 |
|
---|
| 64 | /// <summary>
|
---|
| 65 | /// Looks for parent tasks which have FinishWhenChildJobsFinished and set their state to finished
|
---|
| 66 | /// </summary>
|
---|
| 67 | private void FinishParentTasks(IPersistenceManager pm) {
|
---|
| 68 | var resourceDao = pm.ResourceDao;
|
---|
| 69 | var taskDao = pm.TaskDao;
|
---|
| 70 | var resourceIds = resourceDao.GetAll().Select(x => x.ResourceId).ToList();
|
---|
| 71 | var parentTasksToFinish = taskDao.GetParentTasks(resourceIds, 0, true);
|
---|
| 72 | foreach (var task in parentTasksToFinish) {
|
---|
| 73 | task.State = TaskState.Finished;
|
---|
| 74 | task.StateLogs.Add(new StateLog {
|
---|
| 75 | State = task.State,
|
---|
| 76 | SlaveId = null,
|
---|
| 77 | UserId = null,
|
---|
| 78 | Exception = string.Empty,
|
---|
| 79 | DateTime = DateTime.Now
|
---|
| 80 | });
|
---|
| 81 | }
|
---|
| 82 | }
|
---|
| 83 |
|
---|
| 84 | /// <summary>
|
---|
| 85 | /// Looks for task which have not sent heartbeats for some time and reschedules them for calculation
|
---|
| 86 | /// </summary>
|
---|
| 87 | private void SetTimeoutTasksWaiting(IPersistenceManager pm) {
|
---|
| 88 | var taskDao = pm.TaskDao;
|
---|
[12866] | 89 | var tasks = taskDao.GetAll().Where(x => (x.State == TaskState.Calculating && (DateTime.Now - x.LastHeartbeat) > Properties.Settings.Default.CalculatingJobHeartbeatTimeout)
|
---|
| 90 | || (x.State == TaskState.Transferring && (DateTime.Now - x.LastHeartbeat) > Properties.Settings.Default.TransferringJobHeartbeatTimeout));
|
---|
[12789] | 91 | foreach (var task in tasks) {
|
---|
| 92 | task.State = TaskState.Waiting;
|
---|
| 93 | task.StateLogs.Add(new StateLog {
|
---|
| 94 | State = task.State,
|
---|
| 95 | SlaveId = null,
|
---|
| 96 | UserId = null,
|
---|
| 97 | Exception = SlaveTimeout,
|
---|
| 98 | DateTime = DateTime.Now
|
---|
| 99 | });
|
---|
| 100 | task.Command = null;
|
---|
| 101 | }
|
---|
| 102 | }
|
---|
| 103 |
|
---|
| 104 | /// <summary>
|
---|
| 105 | /// Searches for slaves that are disposable and deletes them if they were offline for too long
|
---|
| 106 | /// </summary>
|
---|
| 107 | private void DeleteObsoleteSlaves(IPersistenceManager pm) {
|
---|
| 108 | var slaveDao = pm.SlaveDao;
|
---|
| 109 | var downtimeDao = pm.DowntimeDao;
|
---|
| 110 | var slaveIds = slaveDao.GetAll()
|
---|
| 111 | .Where(x => x.IsDisposable.GetValueOrDefault()
|
---|
| 112 | && x.SlaveState == SlaveState.Offline
|
---|
[12866] | 113 | && (DateTime.Now - x.LastHeartbeat) > Properties.Settings.Default.SweepInterval)
|
---|
[12789] | 114 | .Select(x => x.ResourceId)
|
---|
| 115 | .ToList();
|
---|
| 116 | foreach (var id in slaveIds) {
|
---|
| 117 | bool downtimesAvailable = downtimeDao.GetByResourceId(id).Any();
|
---|
| 118 | if (!downtimesAvailable) {
|
---|
| 119 | slaveDao.Delete(id);
|
---|
| 120 | }
|
---|
| 121 | }
|
---|
| 122 | }
|
---|
| 123 | }
|
---|
| 124 | }
|
---|