forked from jedisct1/heka
-
Notifications
You must be signed in to change notification settings - Fork 0
/
multidecoder.go
463 lines (406 loc) · 12.6 KB
/
multidecoder.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
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
/***** BEGIN LICENSE BLOCK *****
# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this file,
# You can obtain one at http://mozilla.org/MPL/2.0/.
#
# The Initial Developer of the Original Code is the Mozilla Foundation.
# Portions created by the Initial Developer are Copyright (C) 2012-2015
# the Initial Developer. All Rights Reserved.
#
# Contributor(s):
# Victor Ng ([email protected])
# Rob Miller ([email protected])
# Justin Judd ([email protected])
#
# ***** END LICENSE BLOCK *****/
package pipeline
import (
"errors"
"fmt"
"github.com/mozilla-services/heka/message"
"math/rand"
"sort"
"sync"
"sync/atomic"
"time"
)
// multiDecoderNode is used for making sure MultiDecoder dependencies are addressed
type multiDecoderNode struct {
name string
subs []string
dependencyCount int
}
func newMultiDecoderNode(name string, subs []string) multiDecoderNode {
return multiDecoderNode{name, subs, -1}
}
// multiDecoderNodeList fulfills sort.Interface
type multiDecoderNodeList []multiDecoderNode
func (m multiDecoderNodeList) Len() int {
return len(m)
}
func (m multiDecoderNodeList) Swap(i, j int) {
m[i], m[j] = m[j], m[i]
}
func (m multiDecoderNodeList) Less(i, j int) bool {
a := m[i]
b := m[j]
return a.dependencyCount < b.dependencyCount
}
// utility function to make a copy of a map
func copyMap(m map[string]bool) map[string]bool {
d := map[string]bool{}
for k, v := range m {
d[k] = v
}
return d
}
// orderDependencies determines MultiDecoder hierarchy and orders them accordingly
func orderDependencies(decoders []multiDecoderNode) ([]multiDecoderNode, error) {
nodeMap := make(map[string]multiDecoderNode, 0)
for _, node := range decoders {
nodeMap[node.name] = node
}
// rankFunc is a recursive function to traverse graphs from nodes to the root
var rankFunc func(name string, seen map[string]bool) (int, error)
rankFunc = func(name string, seen map[string]bool) (int, error) {
if seen[name] { // If this node has been seen in this path, it means that there is a circular dependency
return -1, errors.New("circular dependency detected")
}
node := nodeMap[name]
if node.dependencyCount >= 0 { // If dependency count is not -1, than we have already calculated it
return node.dependencyCount, nil
}
dependencyCount := len(node.subs)
seen[name] = true
for _, sub := range node.subs {
c, err := rankFunc(sub, copyMap(seen))
if err != nil {
return -1, err
}
dependencyCount += c
}
node.dependencyCount = dependencyCount
nodeMap[name] = node
return dependencyCount, nil
}
// Make sure that rankFunc is called for each multiDecoderNode
for _, node := range decoders {
_, err := rankFunc(node.name, map[string]bool{})
if err != nil {
return nil, err
}
}
// Compile all nodes with dependency count into slice
finishedNodes := make(multiDecoderNodeList, 0)
for _, node := range nodeMap {
finishedNodes = append(finishedNodes, node)
}
// sort nodes so that those with no dependencies are earlier in the list
sort.Sort(finishedNodes)
return finishedNodes, nil
}
// DecoderRunner wrapper that the MultiDecoder will hand to any subs that ask
// for one. Shadows some data and methods, but doesn't spin up any goroutines.
type mDRunner struct {
*dRunner
decoder Decoder
name string
subName string
}
func (mdr *mDRunner) Name() string {
return mdr.name
}
func (mdr *mDRunner) SetName(name string) {
mdr.name = name
}
func (mdr *mDRunner) Plugin() Plugin {
return mdr.decoder.(Plugin)
}
func (mdr *mDRunner) Decoder() Decoder {
return mdr.decoder
}
func (mdr *mDRunner) LogError(err error) {
LogError.Printf("SubDecoder '%s' error: %s", mdr.name, err)
}
func (mdr *mDRunner) LogMessage(msg string) {
LogInfo.Printf("SubDecoder '%s': %s", mdr.name, msg)
}
type MultiDecoder struct {
processMessageCount []int64
processMessageFailures []int64
processMessageSamples []int64
processMessageDuration []int64
totalMessageFailures int64
totalMessageSamples int64
totalMessageDuration int64
idx uint8
sampleDenominator int
sample bool
reportLock sync.RWMutex
pConfig *PipelineConfig
Config *MultiDecoderConfig
Name string
Decoders []Decoder
dRunner DecoderRunner
CascStrat int
neverTrustEncodes bool
}
type MultiDecoderConfig struct {
Subs []string
LogSubErrors bool `toml:"log_sub_errors"`
CascadeStrategy string `toml:"cascade_strategy"`
}
const (
CASC_FIRST_WINS = iota
CASC_ALL
)
var mdStrategies = map[string]int{"first-wins": CASC_FIRST_WINS, "all": CASC_ALL}
func (md *MultiDecoder) ConfigStruct() interface{} {
subs := make([]string, 0)
return &MultiDecoderConfig{subs, false, "first-wins"}
}
// Heka will call this before calling Init() to set the name of the
// MultiDecoder based on the section name in the TOML config.
func (md *MultiDecoder) SetName(name string) {
md.Name = name
}
// Heka will call this before calling any other methods to give us access to
// the pipeline configuration.
func (md *MultiDecoder) SetPipelineConfig(pConfig *PipelineConfig) {
md.pConfig = pConfig
}
func (md *MultiDecoder) Init(config interface{}) (err error) {
md.Config = config.(*MultiDecoderConfig)
numSubs := len(md.Config.Subs)
if numSubs == 0 {
return errors.New("At least one subdecoder must be specified.")
}
md.Decoders = make([]Decoder, len(md.Config.Subs))
var (
ok bool
decoder Decoder
)
if md.CascStrat, ok = mdStrategies[md.Config.CascadeStrategy]; !ok {
return fmt.Errorf("Unrecognized cascade strategy: %s", md.Config.CascadeStrategy)
}
for i, name := range md.Config.Subs {
if decoder, ok = md.pConfig.Decoder(name); !ok {
return fmt.Errorf("Non-existent subdecoder: %s", name)
}
md.Decoders[i] = decoder
}
// We can trust the embedded decoders to leave the pack.MsgBytes and
// pack.TrustMsgBytes values in the right state in all cases except when
// cascade_strategy == "all", an earlier decoder sets the encoding, but
// the last one in the list does not. We check for this case and, if so,
// explicitly set pack.TrustMsgBytes to false for all packs on every
// successful decode.
if md.CascStrat == CASC_ALL {
lastDecoder := md.Decoders[len(md.Decoders)-1]
_, ok = lastDecoder.(EncodesMsgBytes)
if !ok {
for _, d := range md.Decoders {
if _, ok = d.(EncodesMsgBytes); ok {
md.neverTrustEncodes = true
break
}
}
}
}
md.processMessageCount = make([]int64, numSubs)
md.processMessageFailures = make([]int64, numSubs)
md.processMessageSamples = make([]int64, numSubs)
md.processMessageDuration = make([]int64, numSubs)
md.sampleDenominator = md.pConfig.Globals.SampleDenominator
return nil
}
// Heka will call this to give us access to the runner. We'll store it for
// ourselves, but also have to pass on a wrapped version to any subdecoders
// that might need it.
func (md *MultiDecoder) SetDecoderRunner(dr DecoderRunner) {
md.dRunner = dr
for i, decoder := range md.Decoders {
subName := md.Config.Subs[i]
if wanter, ok := decoder.(WantsDecoderRunner); ok {
// It wants a DecoderRunner, have to create one. But first we need
// to get our hands on a *dRunner.
var realDRunner *dRunner
if realDRunner, ok = dr.(*dRunner); !ok {
// It's not a *dRunner, maybe it's an *mDRunner?
var mdr *mDRunner
if mdr, ok = dr.(*mDRunner); ok {
// Bingo, we can grab its *dRunner.
realDRunner = mdr.dRunner
}
}
if realDRunner == nil {
// Couldn't get a *dRunner. Just log an error and pass the
// outer DecoderRunner through.
dr.LogError(fmt.Errorf("Can't create nested DecoderRunner for '%s'",
subName))
wanter.SetDecoderRunner(dr)
continue
}
// We have a *dRunner, use it to create an *mDRunner.
subRunner := &mDRunner{
realDRunner,
decoder,
fmt.Sprintf("%s-%s", realDRunner.name, subName),
subName,
}
wanter.SetDecoderRunner(subRunner)
}
}
}
// Heka will call this at DecoderRunner shutdown time, we might need to pass
// this along to subdecoders.
func (md *MultiDecoder) Shutdown() {
for _, decoder := range md.Decoders {
if wanter, ok := decoder.(WantsDecoderRunnerShutdown); ok {
wanter.Shutdown()
}
}
}
// Recurses through a decoder chain, decoding the original pack and returning
// it and any generated extra packs.
func (md *MultiDecoder) getDecodedPacks(chain []Decoder, inPacks []*PipelinePack) (
packs []*PipelinePack, anyMatch bool) {
var startTime time.Time
decoder := chain[0]
for _, p := range inPacks {
atomic.AddInt64(&md.processMessageCount[md.idx], 1)
if md.sample {
startTime = time.Now()
}
ps, err := decoder.Decode(p)
if md.sample {
duration := time.Since(startTime).Nanoseconds()
md.reportLock.Lock()
md.processMessageDuration[md.idx] += duration
md.processMessageSamples[md.idx]++
md.reportLock.Unlock()
}
if ps != nil {
anyMatch = true
packs = append(packs, ps...)
} else {
atomic.AddInt64(&md.processMessageFailures[md.idx], 1)
if err != nil && md.Config.LogSubErrors {
idx := len(md.Decoders) - len(chain)
err = fmt.Errorf("Subdecoder '%s' decode error: %s",
md.Config.Subs[idx], err)
md.dRunner.LogError(err)
}
packs = append(packs, p)
}
}
if len(chain) > 1 {
md.idx++
var otherMatch bool
packs, otherMatch = md.getDecodedPacks(chain[1:], packs)
anyMatch = anyMatch || otherMatch
}
return
}
// Runs the message payload against each of the decoders.
func (md *MultiDecoder) Decode(pack *PipelinePack) (packs []*PipelinePack, err error) {
md.sample = (rand.Intn(md.sampleDenominator) == 0 ||
atomic.LoadInt64(&md.processMessageCount[0]) == 0)
var startTime time.Time
if md.sample {
startTime = time.Now()
defer func() {
duration := time.Since(startTime).Nanoseconds()
md.reportLock.Lock()
md.totalMessageDuration += duration
md.totalMessageSamples++
md.reportLock.Unlock()
}()
}
if md.CascStrat == CASC_FIRST_WINS {
var subStartTime time.Time
for i, d := range md.Decoders {
count := atomic.AddInt64(&md.processMessageCount[i], 1)
if md.sample || count == 1 {
subStartTime = time.Now()
}
packs, err = d.Decode(pack)
if md.sample || count == 1 {
duration := time.Since(subStartTime).Nanoseconds()
md.reportLock.Lock()
md.processMessageDuration[i] += duration
md.processMessageSamples[i]++
md.reportLock.Unlock()
}
if packs != nil {
return
}
atomic.AddInt64(&md.processMessageFailures[i], 1)
if err != nil && md.Config.LogSubErrors {
err = fmt.Errorf("Subdecoder '%s' decode error: %s", md.Config.Subs[i],
err)
md.dRunner.LogError(err)
}
}
// If we got this far none of the decoders succeeded.
atomic.AddInt64(&md.totalMessageFailures, 1)
err = errors.New("All subdecoders failed.")
packs = nil
} else {
// If we get here we know cascade_strategy == "all".
var anyMatch bool
md.idx = 0
packs, anyMatch = md.getDecodedPacks(md.Decoders, []*PipelinePack{pack})
if !anyMatch {
atomic.AddInt64(&md.totalMessageFailures, 1)
err = errors.New("All subdecoders failed.")
packs = nil
} else if md.neverTrustEncodes {
for _, p := range packs {
p.TrustMsgBytes = false
}
}
}
return
}
func (md *MultiDecoder) EncodesMsgBytes() bool {
return true
}
func (md *MultiDecoder) ReportMsg(msg *message.Message) error {
md.reportLock.RLock()
defer md.reportLock.RUnlock()
var tmp int64
for i, sub := range md.Config.Subs {
message.NewInt64Field(msg,
fmt.Sprintf("ProcessMessageCount-%s", sub),
atomic.LoadInt64(&md.processMessageCount[i]), "count")
message.NewInt64Field(msg,
fmt.Sprintf("ProcessMessageFailures-%s", sub),
atomic.LoadInt64(&md.processMessageFailures[i]), "count")
message.NewInt64Field(msg,
fmt.Sprintf("ProcessMessageSamples-%s", sub),
md.processMessageSamples[i], "count")
tmp = 0
if md.processMessageSamples[i] > 0 {
tmp = md.processMessageDuration[i] / md.processMessageSamples[i]
}
message.NewInt64Field(msg,
fmt.Sprintf("ProcessMessageAvgDuration-%s", sub), tmp, "ns")
}
message.NewInt64Field(msg, "ProcessMessageCount",
atomic.LoadInt64(&md.processMessageCount[0]), "count")
message.NewInt64Field(msg, "ProcessMessageFailures",
atomic.LoadInt64(&md.totalMessageFailures), "count")
message.NewInt64Field(msg, "ProcessMessageSamples", md.totalMessageSamples, "count")
tmp = 0
if md.totalMessageSamples > 0 {
tmp = md.totalMessageDuration / md.totalMessageSamples
}
message.NewInt64Field(msg, "ProcessMessageAvgDuration", tmp, "ns")
return nil
}
func init() {
RegisterPlugin("MultiDecoder", func() interface{} {
return new(MultiDecoder)
})
}