forked from asnowfox/vflow
/
sflow.go
287 lines (240 loc) · 5.7 KB
/
sflow.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
//: ----------------------------------------------------------------------------
//: Copyright (C) 2017 Verizon. All Rights Reserved.
//: All Rights Reserved
//:
//: file: sflow.go
//: details: sflow decoding raw data samples
//: author: Mehrdad Arshad Rad
//: date: 02/01/2017
//:
//: 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 (
"bytes"
"encoding/json"
"net"
"strconv"
"sync"
"sync/atomic"
"time"
"github.com/VerizonDigital/vflow/producer"
"github.com/VerizonDigital/vflow/sflow"
)
// SFUDPMsg represents sFlow UDP message
type SFUDPMsg struct {
raddr *net.UDPAddr
body []byte
}
// SFlow represents sFlow collector
type SFlow struct {
port int
addr string
workers int
stop bool
stats SFlowStats
pool chan chan struct{}
}
// SFlowStats represents sflow stats
type SFlowStats struct {
UDPQueue int
MessageQueue int
UDPCount uint64
DecodedCount uint64
MQErrorCount uint64
Workers int32
}
var (
sFlowUDPCh = make(chan SFUDPMsg, 1000)
sFlowMQCh = make(chan []byte, 1000)
// sflow udp payload pool
sFlowBuffer = &sync.Pool{
New: func() interface{} {
return make([]byte, opts.SFlowUDPSize)
},
}
)
// NewSFlow constructs sFlow collector
func NewSFlow() *SFlow {
logger = opts.Logger
return &SFlow{
port: opts.SFlowPort,
workers: opts.SFlowWorkers,
pool: make(chan chan struct{}, maxWorkers),
}
}
func (s *SFlow) run() {
// exit if the sflow is disabled
if !opts.SFlowEnabled {
logger.Println("sflow has been disabled")
return
}
hostPort := net.JoinHostPort(s.addr, strconv.Itoa(s.port))
udpAddr, _ := net.ResolveUDPAddr("udp", hostPort)
conn, err := net.ListenUDP("udp", udpAddr)
if err != nil {
logger.Fatal(err)
}
atomic.AddInt32(&s.stats.Workers, int32(s.workers))
for i := 0; i < s.workers; i++ {
go func() {
wQuit := make(chan struct{})
s.pool <- wQuit
s.sFlowWorker(wQuit)
}()
}
logger.Printf("sFlow is running (UDP: listening on [::]:%d workers#: %d)", s.port, s.workers)
go func() {
p := producer.NewProducer(opts.MQName)
p.MQConfigFile = opts.MQConfigFile
p.MQErrorCount = &s.stats.MQErrorCount
p.Logger = logger
p.Chan = sFlowMQCh
p.Topic = opts.SFlowTopic
if err := p.Run(); err != nil {
logger.Fatal(err)
}
}()
go func() {
if !opts.DynWorkers {
logger.Println("sFlow dynamic worker disabled")
return
}
s.dynWorkers()
}()
for !s.stop {
b := sFlowBuffer.Get().([]byte)
conn.SetReadDeadline(time.Now().Add(1e9))
n, raddr, err := conn.ReadFromUDP(b)
if err != nil {
continue
}
atomic.AddUint64(&s.stats.UDPCount, 1)
sFlowUDPCh <- SFUDPMsg{raddr, b[:n]}
}
}
func (s *SFlow) shutdown() {
s.stop = true
logger.Println("stopping sflow service gracefully ...")
time.Sleep(1 * time.Second)
logger.Println("vFlow has been shutdown")
close(sFlowUDPCh)
}
func (s *SFlow) sFlowWorker(wQuit chan struct{}) {
var (
reader *bytes.Reader
msg SFUDPMsg
ok bool
b []byte
)
LOOP:
for {
select {
case <-wQuit:
break LOOP
case msg, ok = <-sFlowUDPCh:
if !ok {
break LOOP
}
}
if opts.Verbose {
logger.Printf("rcvd sflow data from: %s, size: %d bytes",
msg.raddr, len(msg.body))
}
reader = bytes.NewReader(msg.body)
d := sflow.NewSFDecoder(reader, opts.SFlowTypeFilter)
datagram, err := d.SFDecode()
if err != nil || len(datagram.Samples) < 1 {
sFlowBuffer.Put(msg.body[:opts.SFlowUDPSize])
continue
}
b, err = json.Marshal(datagram)
if err != nil {
sFlowBuffer.Put(msg.body[:opts.SFlowUDPSize])
logger.Println(err)
continue
}
atomic.AddUint64(&s.stats.DecodedCount, 1)
if opts.Verbose {
logger.Println(string(b))
}
select {
case sFlowMQCh <- append([]byte{}, b...):
default:
}
sFlowBuffer.Put(msg.body[:opts.SFlowUDPSize])
}
}
func (s *SFlow) status() *SFlowStats {
return &SFlowStats{
UDPQueue: len(sFlowUDPCh),
MessageQueue: len(sFlowMQCh),
UDPCount: atomic.LoadUint64(&s.stats.UDPCount),
DecodedCount: atomic.LoadUint64(&s.stats.DecodedCount),
MQErrorCount: atomic.LoadUint64(&s.stats.MQErrorCount),
Workers: atomic.LoadInt32(&s.stats.Workers),
}
}
func (s *SFlow) dynWorkers() {
var load, nSeq, newWorkers, workers, n int
tick := time.Tick(120 * time.Second)
for {
<-tick
load = 0
for n = 0; n < 30; n++ {
time.Sleep(1 * time.Second)
load += len(sFlowUDPCh)
}
if load > 15 {
switch {
case load > 300:
newWorkers = 100
case load > 200:
newWorkers = 60
case load > 100:
newWorkers = 40
default:
newWorkers = 30
}
workers = int(atomic.LoadInt32(&s.stats.Workers))
if workers+newWorkers > maxWorkers {
logger.Println("sflow :: max out workers")
continue
}
for n = 0; n < newWorkers; n++ {
go func() {
atomic.AddInt32(&s.stats.Workers, 1)
wQuit := make(chan struct{})
s.pool <- wQuit
s.sFlowWorker(wQuit)
}()
}
}
if load == 0 {
nSeq++
} else {
nSeq = 0
continue
}
if nSeq > 15 {
for n = 0; n < 10; n++ {
if len(s.pool) > s.workers {
atomic.AddInt32(&s.stats.Workers, -1)
wQuit := <-s.pool
close(wQuit)
}
}
nSeq = 0
}
}
}