forked from prometheus/prometheus
/
main.go
397 lines (305 loc) · 11.9 KB
/
main.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
// Copyright 2013 Prometheus Team
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package main
import (
"flag"
"os"
"os/signal"
"sync"
"syscall"
"time"
"github.com/golang/glog"
"github.com/prometheus/client_golang/extraction"
clientmodel "github.com/prometheus/client_golang/model"
"github.com/prometheus/prometheus/config"
"github.com/prometheus/prometheus/notification"
"github.com/prometheus/prometheus/retrieval"
"github.com/prometheus/prometheus/rules"
"github.com/prometheus/prometheus/storage/metric/tiered"
"github.com/prometheus/prometheus/storage/remote"
"github.com/prometheus/prometheus/storage/remote/opentsdb"
"github.com/prometheus/prometheus/web"
"github.com/prometheus/prometheus/web/api"
)
const deletionBatchSize = 100
// Commandline flags.
var (
configFile = flag.String("configFile", "prometheus.conf", "Prometheus configuration file name.")
metricsStoragePath = flag.String("metricsStoragePath", "/tmp/metrics", "Base path for metrics storage.")
alertmanagerUrl = flag.String("alertmanager.url", "", "The URL of the alert manager to send notifications to.")
remoteTSDBUrl = flag.String("storage.remote.url", "", "The URL of the OpenTSDB instance to send samples to.")
remoteTSDBTimeout = flag.Duration("storage.remote.timeout", 30*time.Second, "The timeout to use when sending samples to OpenTSDB.")
samplesQueueCapacity = flag.Int("storage.queue.samplesCapacity", 4096, "The size of the unwritten samples queue.")
diskAppendQueueCapacity = flag.Int("storage.queue.diskAppendCapacity", 1000000, "The size of the queue for items that are pending writing to disk.")
memoryAppendQueueCapacity = flag.Int("storage.queue.memoryAppendCapacity", 10000, "The size of the queue for items that are pending writing to memory.")
compactInterval = flag.Duration("compact.interval", 3*time.Hour, "The amount of time between compactions.")
compactGroupSize = flag.Int("compact.groupSize", 500, "The minimum group size for compacted samples.")
compactAgeInclusiveness = flag.Duration("compact.ageInclusiveness", 5*time.Minute, "The age beyond which samples should be compacted.")
deleteInterval = flag.Duration("delete.interval", 11*time.Hour, "The amount of time between deletion of old values.")
deleteAge = flag.Duration("delete.ageMaximum", 15*24*time.Hour, "The relative maximum age for values before they are deleted.")
arenaFlushInterval = flag.Duration("arena.flushInterval", 15*time.Minute, "The period at which the in-memory arena is flushed to disk.")
arenaTTL = flag.Duration("arena.ttl", 10*time.Minute, "The relative age of values to purge to disk from memory.")
notificationQueueCapacity = flag.Int("alertmanager.notificationQueueCapacity", 100, "The size of the queue for pending alert manager notifications.")
concurrentRetrievalAllowance = flag.Int("concurrentRetrievalAllowance", 15, "The number of concurrent metrics retrieval requests allowed.")
printVersion = flag.Bool("version", false, "print version information")
shutdownTimeout = flag.Duration("shutdownGracePeriod", 0*time.Second, "The amount of time Prometheus gives background services to finish running when shutdown is requested.")
)
type prometheus struct {
compactionTimer *time.Ticker
deletionTimer *time.Ticker
curationSema chan struct{}
stopBackgroundOperations chan struct{}
unwrittenSamples chan *extraction.Result
ruleManager rules.RuleManager
targetManager retrieval.TargetManager
notifications chan notification.NotificationReqs
storage *tiered.TieredStorage
remoteTSDBQueue *remote.TSDBQueueManager
curationState tiered.CurationStateUpdater
closeOnce sync.Once
}
func (p *prometheus) interruptHandler() {
notifier := make(chan os.Signal)
signal.Notify(notifier, os.Interrupt, syscall.SIGTERM)
<-notifier
glog.Warning("Received SIGINT/SIGTERM; Exiting gracefully...")
p.Close()
os.Exit(0)
}
func (p *prometheus) compact(olderThan time.Duration, groupSize int) error {
select {
case s, ok := <-p.curationSema:
if !ok {
glog.Warning("Prometheus is shutting down; no more curation runs are allowed.")
return nil
}
defer func() {
p.curationSema <- s
}()
default:
glog.Warningf("Deferred compaction for %s and %s due to existing operation.", olderThan, groupSize)
return nil
}
processor := tiered.NewCompactionProcessor(&tiered.CompactionProcessorOptions{
MaximumMutationPoolBatch: groupSize * 3,
MinimumGroupSize: groupSize,
})
defer processor.Close()
curator := tiered.NewCurator(&tiered.CuratorOptions{
Stop: p.stopBackgroundOperations,
ViewQueue: p.storage.ViewQueue,
})
defer curator.Close()
return curator.Run(olderThan, clientmodel.Now(), processor, p.storage.DiskStorage.CurationRemarks, p.storage.DiskStorage.MetricSamples, p.storage.DiskStorage.MetricHighWatermarks, p.curationState)
}
func (p *prometheus) delete(olderThan time.Duration, batchSize int) error {
select {
case s, ok := <-p.curationSema:
if !ok {
glog.Warning("Prometheus is shutting down; no more curation runs are allowed.")
return nil
}
defer func() {
p.curationSema <- s
}()
default:
glog.Warningf("Deferred deletion for %s due to existing operation.", olderThan)
return nil
}
processor := tiered.NewDeletionProcessor(&tiered.DeletionProcessorOptions{
MaximumMutationPoolBatch: batchSize,
})
defer processor.Close()
curator := tiered.NewCurator(&tiered.CuratorOptions{
Stop: p.stopBackgroundOperations,
ViewQueue: p.storage.ViewQueue,
})
defer curator.Close()
return curator.Run(olderThan, clientmodel.Now(), processor, p.storage.DiskStorage.CurationRemarks, p.storage.DiskStorage.MetricSamples, p.storage.DiskStorage.MetricHighWatermarks, p.curationState)
}
func (p *prometheus) Close() {
p.closeOnce.Do(p.close)
}
func (p *prometheus) close() {
// The "Done" remarks are a misnomer for some subsystems due to lack of
// blocking and synchronization.
glog.Info("Shutdown has been requested; subsytems are closing:")
p.targetManager.Stop()
glog.Info("Remote Target Manager: Done")
p.ruleManager.Stop()
glog.Info("Rule Executor: Done")
// Stop any currently active curation (deletion or compaction).
close(p.stopBackgroundOperations)
glog.Info("Current Curation Workers: Requested")
// Disallow further curation work.
close(p.curationSema)
// Stop curation timers.
if p.compactionTimer != nil {
p.compactionTimer.Stop()
}
if p.deletionTimer != nil {
p.deletionTimer.Stop()
}
glog.Info("Future Curation Workers: Done")
glog.Infof("Waiting %s for background systems to exit and flush before finalizing (DO NOT INTERRUPT THE PROCESS) ...", *shutdownTimeout)
// Wart: We should have a concrete form of synchronization for this, not a
// hokey sleep statement.
time.Sleep(*shutdownTimeout)
close(p.unwrittenSamples)
p.storage.Close()
glog.Info("Local Storage: Done")
if p.remoteTSDBQueue != nil {
p.remoteTSDBQueue.Close()
glog.Info("Remote Storage: Done")
}
close(p.notifications)
glog.Info("Sundry Queues: Done")
glog.Info("See you next time!")
}
func main() {
// TODO(all): Future additions to main should be, where applicable, glumped
// into the prometheus struct above---at least where the scoping of the entire
// server is concerned.
flag.Parse()
versionInfoTmpl.Execute(os.Stdout, BuildInfo)
if *printVersion {
os.Exit(0)
}
conf, err := config.LoadFromFile(*configFile)
if err != nil {
glog.Fatalf("Error loading configuration from %s: %v", *configFile, err)
}
ts, err := tiered.NewTieredStorage(uint(*diskAppendQueueCapacity), 100, *arenaFlushInterval, *arenaTTL, *metricsStoragePath)
if err != nil {
glog.Fatal("Error opening storage: ", err)
}
var remoteTSDBQueue *remote.TSDBQueueManager = nil
if *remoteTSDBUrl == "" {
glog.Warningf("No TSDB URL provided; not sending any samples to long-term storage")
} else {
openTSDB := opentsdb.NewClient(*remoteTSDBUrl, *remoteTSDBTimeout)
remoteTSDBQueue = remote.NewTSDBQueueManager(openTSDB, 512)
go remoteTSDBQueue.Run()
}
unwrittenSamples := make(chan *extraction.Result, *samplesQueueCapacity)
ingester := &retrieval.MergeLabelsIngester{
Labels: conf.GlobalLabels(),
CollisionPrefix: clientmodel.ExporterLabelPrefix,
Ingester: retrieval.ChannelIngester(unwrittenSamples),
}
compactionTimer := time.NewTicker(*compactInterval)
deletionTimer := time.NewTicker(*deleteInterval)
// Queue depth will need to be exposed
targetManager := retrieval.NewTargetManager(ingester, *concurrentRetrievalAllowance)
targetManager.AddTargetsFromConfig(conf)
notifications := make(chan notification.NotificationReqs, *notificationQueueCapacity)
// Queue depth will need to be exposed
ruleManager := rules.NewRuleManager(&rules.RuleManagerOptions{
Results: unwrittenSamples,
Notifications: notifications,
EvaluationInterval: conf.EvaluationInterval(),
Storage: ts,
PrometheusUrl: web.MustBuildServerUrl(),
})
if err := ruleManager.AddRulesFromConfig(conf); err != nil {
glog.Fatal("Error loading rule files: ", err)
}
go ruleManager.Run()
notificationHandler := notification.NewNotificationHandler(*alertmanagerUrl, notifications)
go notificationHandler.Run()
flags := map[string]string{}
flag.VisitAll(func(f *flag.Flag) {
flags[f.Name] = f.Value.String()
})
prometheusStatus := &web.PrometheusStatusHandler{
BuildInfo: BuildInfo,
Config: conf.String(),
RuleManager: ruleManager,
TargetPools: targetManager.Pools(),
Flags: flags,
Birth: time.Now(),
}
alertsHandler := &web.AlertsHandler{
RuleManager: ruleManager,
}
databasesHandler := &web.DatabasesHandler{
Provider: ts.DiskStorage,
RefreshInterval: 5 * time.Minute,
}
metricsService := &api.MetricsService{
Config: &conf,
TargetManager: targetManager,
Storage: ts,
}
prometheus := &prometheus{
compactionTimer: compactionTimer,
deletionTimer: deletionTimer,
curationState: prometheusStatus,
curationSema: make(chan struct{}, 1),
unwrittenSamples: unwrittenSamples,
stopBackgroundOperations: make(chan struct{}),
ruleManager: ruleManager,
targetManager: targetManager,
notifications: notifications,
storage: ts,
remoteTSDBQueue: remoteTSDBQueue,
}
defer prometheus.Close()
webService := &web.WebService{
StatusHandler: prometheusStatus,
MetricsHandler: metricsService,
DatabasesHandler: databasesHandler,
AlertsHandler: alertsHandler,
QuitDelegate: prometheus.Close,
}
prometheus.curationSema <- struct{}{}
storageStarted := make(chan bool)
go ts.Serve(storageStarted)
<-storageStarted
go prometheus.interruptHandler()
go func() {
for _ = range prometheus.compactionTimer.C {
glog.Info("Starting compaction...")
err := prometheus.compact(*compactAgeInclusiveness, *compactGroupSize)
if err != nil {
glog.Error("could not compact: ", err)
}
glog.Info("Done")
}
}()
go func() {
for _ = range prometheus.deletionTimer.C {
glog.Info("Starting deletion of stale values...")
err := prometheus.delete(*deleteAge, deletionBatchSize)
if err != nil {
glog.Error("could not delete: ", err)
}
glog.Info("Done")
}
}()
go func() {
err := webService.ServeForever()
if err != nil {
glog.Fatal(err)
}
}()
// TODO(all): Migrate this into prometheus.serve().
for block := range unwrittenSamples {
if block.Err == nil && len(block.Samples) > 0 {
ts.AppendSamples(block.Samples)
if remoteTSDBQueue != nil {
remoteTSDBQueue.Queue(block.Samples)
}
}
}
}