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
|
package roadrunner
import (
"context"
"errors"
"fmt"
"time"
)
const MB = 1024 * 1024
type Supervisor interface {
Attach(pool Pool)
StartWatching() error
StopWatching()
Detach()
}
type staticPoolSupervisor struct {
// maxWorkerMemory in MB
maxWorkerMemory uint64
// maxPoolMemory in MB
maxPoolMemory uint64
// maxWorkerTTL in seconds
maxWorkerTTL uint64
// maxWorkerIdle in seconds
maxWorkerIdle uint64
// watchTimeout in seconds
watchTimeout uint64
stopCh chan struct{}
pool Pool
}
/*
The arguments are:
maxWorkerMemory - maximum memory allowed for a single worker
maxPoolMemory - maximum pool memory allowed for a pool of a workers
maxTtl - maximum ttl for the worker after which it will be killed and replaced
maxIdle - maximum time to live for the worker in Ready state
watchTimeout - time between watching for the workers/pool status
*/
// TODO might be just wrap the pool and return ControlledPool with included Pool interface
func NewStaticPoolSupervisor(maxWorkerMemory, maxPoolMemory, maxTtl, maxIdle, watchTimeout uint64) Supervisor {
if maxWorkerMemory == 0 {
// just set to a big number, 5GB
maxPoolMemory = 5000 * MB
}
if watchTimeout == 0 {
watchTimeout = 60
}
return &staticPoolSupervisor{
maxWorkerMemory: maxWorkerMemory,
maxPoolMemory: maxPoolMemory,
maxWorkerTTL: maxTtl,
maxWorkerIdle: maxIdle,
stopCh: make(chan struct{}),
}
}
func (sps *staticPoolSupervisor) Attach(pool Pool) {
sps.pool = pool
}
func (sps *staticPoolSupervisor) StartWatching() error {
go func() {
watchTout := time.NewTicker(time.Second * time.Duration(sps.watchTimeout))
for {
select {
case <-sps.stopCh:
watchTout.Stop()
return
// stop here
case <-watchTout.C:
err := sps.control()
if err != nil {
sps.pool.Events() <- PoolEvent{Payload: err}
}
}
}
}()
return nil
}
func (sps *staticPoolSupervisor) StopWatching() {
sps.stopCh <- struct{}{}
}
func (sps *staticPoolSupervisor) Detach() {
}
func (sps *staticPoolSupervisor) control() error {
if sps.pool == nil {
return errors.New("pool should be attached")
}
now := time.Now()
ctx := context.TODO()
// THIS IS A COPY OF WORKERS
workers := sps.pool.Workers()
var totalUsedMemory uint64
for i := 0; i < len(workers); i++ {
if workers[i].State().Value() == StateInvalid {
continue
}
s, err := WorkerProcessState(workers[i])
if err != nil {
panic(err)
// push to pool events??
}
if sps.maxWorkerTTL != 0 && now.Sub(workers[i].Created()).Seconds() >= float64(sps.maxWorkerTTL) {
err = sps.pool.RemoveWorker(ctx, workers[i])
if err != nil {
return err
}
// after remove worker we should exclude it from further analysis
workers = append(workers[:i], workers[i+1:]...)
}
if sps.maxWorkerMemory != 0 && s.MemoryUsage >= sps.maxWorkerMemory*MB {
// TODO events
sps.pool.Events() <- PoolEvent{Payload: fmt.Errorf("max allowed memory reached (%vMB)", sps.maxWorkerMemory)}
err = sps.pool.RemoveWorker(ctx, workers[i])
if err != nil {
return err
}
workers = append(workers[:i], workers[i+1:]...)
continue
}
// firs we check maxWorker idle
if sps.maxWorkerIdle != 0 {
// then check for the worker state
if workers[i].State().Value() != StateReady {
continue
}
/*
Calculate idle time
If worker in the StateReady, we read it LastUsed timestamp as UnixNano uint64
2. For example maxWorkerIdle is equal to 5sec, then, if (time.Now - LastUsed) > maxWorkerIdle
we are guessing that worker overlap idle time and has to be killed
*/
// get last used unix nano
lu := workers[i].State().LastUsed()
// convert last used to unixNano and sub time.now
res := int64(lu) - now.UnixNano()
// maxWorkerIdle more than diff between now and last used
if int64(sps.maxWorkerIdle)-res <= 0 {
sps.pool.Events() <- PoolEvent{Payload: fmt.Errorf("max allowed worker idle time elapsed. actual idle time: %v, max idle time: %v", sps.maxWorkerIdle, res)}
err = sps.pool.RemoveWorker(ctx, workers[i])
if err != nil {
return err
}
workers = append(workers[:i], workers[i+1:]...)
}
}
// the very last step is to calculate pool memory usage (except excluded workers)
totalUsedMemory += s.MemoryUsage
}
// if current usage more than max allowed pool memory usage
if totalUsedMemory > sps.maxPoolMemory {
// destroy pool
totalUsedMemory = 0
sps.pool.Destroy(ctx)
}
return nil
}
|
