1 | using System;
|
---|
2 | using System.Linq;
|
---|
3 | using System.Transactions;
|
---|
4 | using HeuristicLab.Services.Access;
|
---|
5 | using HeuristicLab.Services.Hive.DataAccess;
|
---|
6 |
|
---|
7 | namespace HeuristicLab.Services.Hive {
|
---|
8 | public class HiveStatisticsGenerator : IStatisticsGenerator {
|
---|
9 |
|
---|
10 | private IUserManager userManager { get { return ServiceLocator.Instance.UserManager; } }
|
---|
11 |
|
---|
12 | private static readonly TimeSpan SmallestTimeSpan = new TimeSpan(0, 5, 0);
|
---|
13 |
|
---|
14 | public void GenerateStatistics() {
|
---|
15 | using (var db = new HiveDataContext(Settings.Default.HeuristicLab_Hive_LinqConnectionString))
|
---|
16 | using (var transaction = new TransactionScope(TransactionScopeOption.Required, new TransactionOptions { IsolationLevel = IsolationLevel.ReadCommitted })) {
|
---|
17 |
|
---|
18 | var newTime = UpdateDimensionTables(db);
|
---|
19 | db.SubmitChanges();
|
---|
20 |
|
---|
21 | if (newTime != null) {
|
---|
22 | UpdateFactTables(newTime, db);
|
---|
23 | db.SubmitChanges();
|
---|
24 | }
|
---|
25 |
|
---|
26 | transaction.Complete();
|
---|
27 | }
|
---|
28 | }
|
---|
29 |
|
---|
30 | private DimTime UpdateDimensionTables(HiveDataContext db) {
|
---|
31 | var newTime = UpdateTime(db);
|
---|
32 | // Update other tables out of sync with time dimension?
|
---|
33 | UpdateUsers(db);
|
---|
34 | UpdateJobs(db);
|
---|
35 | UpdateClients(db);
|
---|
36 |
|
---|
37 | return newTime;
|
---|
38 | }
|
---|
39 |
|
---|
40 | private DimTime UpdateTime(HiveDataContext db) {
|
---|
41 | var lastUpdateTime =
|
---|
42 | (from t in db.DimTimes
|
---|
43 | orderby t.Time descending
|
---|
44 | select t.Time)
|
---|
45 | .FirstOrDefault();
|
---|
46 |
|
---|
47 | var now = DateTime.Now;
|
---|
48 | DimTime newTime = null;
|
---|
49 |
|
---|
50 | if (lastUpdateTime == default(DateTime) || lastUpdateTime + SmallestTimeSpan < now) {
|
---|
51 | newTime = new DimTime {
|
---|
52 | Time = new DateTime(now.Year, now.Month, now.Day, now.Hour, now.Minute - now.Minute % SmallestTimeSpan.Minutes, 0),
|
---|
53 | Hour = new DateTime(now.Year, now.Month, now.Day, now.Hour, 0, 0),
|
---|
54 | Day = new DateTime(now.Year, now.Month, now.Day, 0, 0, 0),
|
---|
55 | Month = new DateTime(now.Year, now.Month, 1, 0, 0, 0),
|
---|
56 | Year = new DateTime(now.Year, 1, 1, 0, 0, 0)
|
---|
57 | };
|
---|
58 | db.DimTimes.InsertOnSubmit(newTime);
|
---|
59 | }
|
---|
60 |
|
---|
61 | return newTime;
|
---|
62 | }
|
---|
63 |
|
---|
64 | private void UpdateUsers(HiveDataContext db) {
|
---|
65 | var newUsers =
|
---|
66 | from u in db.Resources.Where(x => x.OwnerUserId != null).Select(x => x.OwnerUserId.Value).Union(db.Jobs.Select(x => x.OwnerUserId))
|
---|
67 | where !db.DimUsers.Select(x => x.UserId).Contains(u)
|
---|
68 | select u;
|
---|
69 |
|
---|
70 | var newDimUsers =
|
---|
71 | from u in newUsers.ToList()
|
---|
72 | select new DimUser {
|
---|
73 | UserId = u,
|
---|
74 | Name = userManager.GetUserById(u).UserName
|
---|
75 | };
|
---|
76 |
|
---|
77 | db.DimUsers.InsertAllOnSubmit(newDimUsers);
|
---|
78 |
|
---|
79 | // insert NULL-User
|
---|
80 | if (!db.DimUsers.Any(x => x.UserId == Guid.Empty)) {
|
---|
81 | db.DimUsers.InsertOnSubmit(new DimUser { UserId = Guid.Empty, Name = "NULL" });
|
---|
82 | }
|
---|
83 | }
|
---|
84 |
|
---|
85 | private void UpdateJobs(HiveDataContext db) {
|
---|
86 | var newJobs =
|
---|
87 | from j in db.Jobs
|
---|
88 | where !db.DimJobs.Select(x => x.JobId).Contains(j.JobId)
|
---|
89 | select j;
|
---|
90 |
|
---|
91 | var newDimJobs =
|
---|
92 | from j in newJobs.ToList()
|
---|
93 | select new DimJob {
|
---|
94 | JobId = j.JobId,
|
---|
95 | JobName = j.Name,
|
---|
96 | UserId = j.OwnerUserId,
|
---|
97 | UserName = userManager.GetUserById(j.OwnerUserId).UserName
|
---|
98 | };
|
---|
99 |
|
---|
100 | db.DimJobs.InsertAllOnSubmit(newDimJobs);
|
---|
101 |
|
---|
102 | // insert NULL-Job
|
---|
103 | if (!db.DimJobs.Any(x => x.JobId == Guid.Empty)) {
|
---|
104 | db.DimJobs.InsertOnSubmit(new DimJob { JobId = Guid.Empty, JobName = "NULL", UserId = Guid.Empty, UserName = "NULL" });
|
---|
105 | }
|
---|
106 | }
|
---|
107 |
|
---|
108 | private void UpdateClients(HiveDataContext db) {
|
---|
109 | var removedClients =
|
---|
110 | from c in db.DimClients
|
---|
111 | where c.ExpirationTime == null &&
|
---|
112 | !db.Resources.OfType<Slave>().Select(x => x.ResourceId).Contains(c.ResourceId)
|
---|
113 | select c;
|
---|
114 |
|
---|
115 | var modifiedClients =
|
---|
116 | from s in db.Resources.OfType<Slave>()
|
---|
117 | join c in db.DimClients on s.ResourceId equals c.ResourceId
|
---|
118 | where c.ExpirationTime == null
|
---|
119 | && (s.Name != c.Name || s.ParentResourceId != c.ResourceGroupId ||
|
---|
120 | s.ParentResource.ParentResourceId != c.ResourceGroup2Id)
|
---|
121 | select new { Slave = s, Client = c };
|
---|
122 |
|
---|
123 | foreach (var client in removedClients.Union(modifiedClients.Select(x => x.Client))) {
|
---|
124 | client.ExpirationTime = DateTime.Now;
|
---|
125 | }
|
---|
126 |
|
---|
127 | var newClients =
|
---|
128 | from s in db.Resources.OfType<Slave>()
|
---|
129 | where !db.DimClients.Select(x => x.ResourceId).Contains(s.ResourceId)
|
---|
130 | || modifiedClients.Select(x => x.Slave.ResourceId).Contains(s.ResourceId)
|
---|
131 | select new {
|
---|
132 | Slave = s,
|
---|
133 | Group = s.ParentResourceId,
|
---|
134 | Group2 = s.ParentResource.ParentResourceId
|
---|
135 | };
|
---|
136 |
|
---|
137 | var newDimClients =
|
---|
138 | from s in newClients.ToList()
|
---|
139 | select new DimClient {
|
---|
140 | ResourceId = s.Slave.ResourceId,
|
---|
141 | Name = s.Slave.Name,
|
---|
142 | ExpirationTime = null,
|
---|
143 | ResourceGroupId = s.Group,
|
---|
144 | ResourceGroup2Id = s.Group2
|
---|
145 | };
|
---|
146 |
|
---|
147 | db.DimClients.InsertAllOnSubmit(newDimClients);
|
---|
148 | }
|
---|
149 |
|
---|
150 | private void UpdateFactTables(DimTime newTime, HiveDataContext db) {
|
---|
151 | UpdateClientInfoFacts(newTime, db);
|
---|
152 | UpdateTaskFacts(newTime, db);
|
---|
153 | }
|
---|
154 |
|
---|
155 | private void UpdateClientInfoFacts(DimTime newTime, HiveDataContext db) {
|
---|
156 | var lastFacts =
|
---|
157 | from cf in db.FactClientInfos
|
---|
158 | join r in db.DimClients on cf.ClientId equals r.Id
|
---|
159 | group cf by r.ResourceId into grpFacts
|
---|
160 | select new {
|
---|
161 | ResourceId = grpFacts.Key,
|
---|
162 | Fact = grpFacts.OrderByDescending(x => x.Time).First(),
|
---|
163 | };
|
---|
164 |
|
---|
165 | var slaves =
|
---|
166 | from s in db.Resources.OfType<Slave>()
|
---|
167 | join c in db.DimClients on s.ResourceId equals c.ResourceId
|
---|
168 | join lcf in lastFacts on c.ResourceId equals lcf.ResourceId into joinCf
|
---|
169 | from cf in joinCf.DefaultIfEmpty()
|
---|
170 | where c.ExpirationTime == null
|
---|
171 | select new {
|
---|
172 | Slave = s,
|
---|
173 | Client = c,
|
---|
174 | LastFact = cf != null ? cf.Fact : null
|
---|
175 | };
|
---|
176 |
|
---|
177 | var clientFacts =
|
---|
178 | from s in slaves.ToList()
|
---|
179 | select new FactClientInfo {
|
---|
180 | DimClient = s.Client,
|
---|
181 | DimTime = newTime,
|
---|
182 | UserId = s.Slave.OwnerUserId ?? Guid.Empty,
|
---|
183 | NumUsedCores =
|
---|
184 | s.Slave.Cores != null && s.Slave.FreeCores != null
|
---|
185 | ? s.Slave.Cores.Value - s.Slave.FreeCores.Value
|
---|
186 | : 0,
|
---|
187 | NumTotalCores = s.Slave.Cores ?? 0,
|
---|
188 | UsedMemory =
|
---|
189 | s.Slave.Memory != null && s.Slave.FreeMemory != null
|
---|
190 | ? s.Slave.Memory.Value - s.Slave.FreeMemory.Value
|
---|
191 | : 0,
|
---|
192 | TotalMemory = s.Slave.Memory ?? 0,
|
---|
193 | CpuUtilization = s.Slave.CpuUtilization,
|
---|
194 | TrafficIn = 0,
|
---|
195 | TrafficOut = 0,
|
---|
196 | TotalTimeIdle = CalcNewTotalTime(s.LastFact, newTime.Time,
|
---|
197 | x => x.TotalTimeIdle,
|
---|
198 | () => s.Slave.SlaveState == SlaveState.Idle && s.Slave.IsAllowedToCalculate),
|
---|
199 | TotalTimeCalculating = CalcNewTotalTime(s.LastFact, newTime.Time,
|
---|
200 | x => x.TotalTimeCalculating,
|
---|
201 | () => s.Slave.SlaveState == SlaveState.Calculating),
|
---|
202 | TotalTimeTransferring = 0.0,
|
---|
203 | TotalTimeUnavailable = CalcNewTotalTime(s.LastFact, newTime.Time,
|
---|
204 | x => x.TotalTimeUnavailable,
|
---|
205 | () => s.Slave.SlaveState == SlaveState.Idle && !s.Slave.IsAllowedToCalculate),
|
---|
206 | TotalTimeOffline = CalcNewTotalTime(s.LastFact, newTime.Time,
|
---|
207 | x => x.TotalTimeOffline,
|
---|
208 | () => s.Slave.SlaveState == SlaveState.Offline)
|
---|
209 | };
|
---|
210 |
|
---|
211 | db.FactClientInfos.InsertAllOnSubmit(clientFacts);
|
---|
212 | }
|
---|
213 |
|
---|
214 | private double CalcNewTotalTime(FactClientInfo lastFact, DateTime newTime, Func<FactClientInfo, double> selector, Func<bool> condition) {
|
---|
215 | if (lastFact == null) {
|
---|
216 | return 0.0;
|
---|
217 | }
|
---|
218 | return condition()
|
---|
219 | ? selector(lastFact) + (newTime - lastFact.Time).TotalMinutes
|
---|
220 | : selector(lastFact);
|
---|
221 | }
|
---|
222 |
|
---|
223 | private void UpdateTaskFacts(DimTime newTime, HiveDataContext db) {
|
---|
224 | var lastFacts =
|
---|
225 | from fact in db.FactTasks
|
---|
226 | group fact by fact.TaskId into factsPerTask
|
---|
227 | select factsPerTask.OrderByDescending(x => x.EndTime).First();
|
---|
228 |
|
---|
229 | var lastFactAndStateLogsAndLastClientPerTask =
|
---|
230 | from task in db.Tasks.Where(task => !task.IsParentTask)
|
---|
231 | join lastFact in lastFacts on task.TaskId equals lastFact.TaskId into lastFactsPerTask
|
---|
232 | from lastFact in lastFactsPerTask.DefaultIfEmpty()
|
---|
233 | let newStateLogs = task.StateLogs.Where(stateLog => (lastFact == null || stateLog.DateTime > lastFact.EndTime) && stateLog.DateTime < newTime.Time)
|
---|
234 | let lastSlaveId = newStateLogs.OrderByDescending(stateLog => stateLog.DateTime).First(stateLog => stateLog.SlaveId != null).SlaveId
|
---|
235 | join client in db.DimClients.Where(client => client.ExpirationTime == null) on lastSlaveId equals client.ResourceId into clientPerTask
|
---|
236 | from client in clientPerTask.DefaultIfEmpty()
|
---|
237 | select new {
|
---|
238 | Task = task,
|
---|
239 | TaskDataSize = task.JobData.DataSize,
|
---|
240 | LastFact = lastFact,
|
---|
241 | StateLogs = newStateLogs.OrderBy(stateLog => stateLog.DateTime),
|
---|
242 | LastClientId = client != null ? client.Id : default(Guid?)
|
---|
243 | };
|
---|
244 |
|
---|
245 | var newTaskFacts =
|
---|
246 | from t in lastFactAndStateLogsAndLastClientPerTask.ToList()
|
---|
247 | select new FactTask {
|
---|
248 | TaskId = t.Task.TaskId,
|
---|
249 | JobId = t.Task.JobId,
|
---|
250 | DimTimeStart = t.LastFact != null ? t.LastFact.DimTimeStart : newTime,
|
---|
251 | DimTimeEnd = newTime,
|
---|
252 | LastClientId = t.LastClientId ?? (t.LastFact != null ? t.LastFact.LastClientId : Guid.Empty),
|
---|
253 | Priority = t.Task.Priority,
|
---|
254 | CoresRequired = t.Task.CoresNeeded,
|
---|
255 | MemoryRequired = t.Task.MemoryNeeded,
|
---|
256 | TaskSize = t.LastFact != null ? t.LastFact.TaskSize : t.TaskDataSize,
|
---|
257 | ResultSize = t.Task.State == TaskState.Finished ? t.TaskDataSize : default(long?),
|
---|
258 | NumCalculationRuns = (t.LastFact != null ? t.LastFact.NumCalculationRuns : 0) + t.StateLogs.Count(x => x.State == TaskState.Calculating),
|
---|
259 | NumFails = (t.LastFact != null ? t.LastFact.NumFails : 0) + t.StateLogs.Count(x => x.State == TaskState.Failed || !string.IsNullOrEmpty(x.Exception)),
|
---|
260 | TotalWaitingTime = 0, //ToDo
|
---|
261 | TotalRuntime = 0, //ToDo
|
---|
262 | TotalTransferTime = 0 //ToDo
|
---|
263 | };
|
---|
264 |
|
---|
265 | db.FactTasks.InsertAllOnSubmit(newTaskFacts);
|
---|
266 | }
|
---|
267 | }
|
---|
268 | } |
---|