-
Notifications
You must be signed in to change notification settings - Fork 4.7k
/
executor.go
212 lines (177 loc) · 4.89 KB
/
executor.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
/*
Copyright 2019 The Kubernetes Authors.
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 fi
import (
"fmt"
"strings"
"sync"
"time"
"k8s.io/klog/v2"
)
type executor[T SubContext] struct {
context *Context[T]
options RunTasksOptions
}
type taskState[T SubContext] struct {
done bool
key string
task Task[T]
deadline time.Time
lastError error
dependencies []*taskState[T]
}
type RunTasksOptions struct {
MaxTaskDuration time.Duration
WaitAfterAllTasksFailed time.Duration
}
func (o *RunTasksOptions) InitDefaults() {
o.MaxTaskDuration = 10 * time.Minute
o.WaitAfterAllTasksFailed = 10 * time.Second
}
// RunTasks executes all the tasks, considering their dependencies
// It will perform some re-execution on error, retrying as long as progress is still being made
func (e *executor[T]) RunTasks(taskMap map[string]Task[T]) error {
dependencies := FindTaskDependencies(taskMap)
for _, task := range taskMap {
if taskPreRun, ok := task.(TaskPreRun[T]); ok {
if err := taskPreRun.PreRun(e.context); err != nil {
return err
}
}
}
taskStates := make(map[string]*taskState[T])
for k, task := range taskMap {
ts := &taskState[T]{
key: k,
task: task,
}
taskStates[k] = ts
}
for k, ts := range taskStates {
for _, dep := range dependencies[k] {
d := taskStates[dep]
if d == nil {
klog.Fatalf("did not find task state for dependency: %q", k)
}
ts.dependencies = append(ts.dependencies, d)
}
}
for {
var canRun []*taskState[T]
doneCount := 0
for _, ts := range taskStates {
if ts.done {
doneCount++
continue
}
ready := true
for _, dep := range ts.dependencies {
if !dep.done {
ready = false
break
}
}
if ready {
if ts.deadline.IsZero() {
ts.deadline = time.Now().Add(e.options.MaxTaskDuration)
} else if time.Now().After(ts.deadline) {
return fmt.Errorf("deadline exceeded executing task %v. Example error: %v", ts.key, ts.lastError)
}
canRun = append(canRun, ts)
}
}
klog.Infof("Tasks: %d done / %d total; %d can run", doneCount, len(taskStates), len(canRun))
if len(canRun) == 0 {
break
}
progress := false
var tasks []*taskState[T]
tasks = append(tasks, canRun...)
taskErrors := e.forkJoin(tasks)
var errors []error
for i, err := range taskErrors {
ts := tasks[i]
if err != nil {
// print warning message and continue like the task succeeded
if _, ok := err.(*ExistsAndWarnIfChangesError); ok {
klog.Warningf(err.Error())
ts.done = true
ts.lastError = nil
progress = true
continue
}
remaining := time.Second * time.Duration(int(time.Until(ts.deadline).Seconds()))
if _, ok := err.(*TryAgainLaterError); ok {
klog.V(2).Infof("Task %q not ready: %v", ts.key, err)
} else {
klog.Warningf("error running task %q (%v remaining to succeed): %v", ts.key, remaining, err)
}
errors = append(errors, err)
ts.lastError = err
} else {
ts.done = true
ts.lastError = nil
progress = true
}
}
if !progress {
if len(errors) == 0 {
// Logic error!
panic("did not make progress executing tasks; but no errors reported")
}
klog.Infof("No progress made, sleeping before retrying %d task(s)", len(errors))
time.Sleep(e.options.WaitAfterAllTasksFailed)
}
}
// Raise error if not all tasks done - this means they depended on each other
var notDone []string
for _, ts := range taskStates {
if !ts.done {
notDone = append(notDone, ts.key)
}
}
if len(notDone) != 0 {
return fmt.Errorf("Unable to execute tasks (circular dependency): %s", strings.Join(notDone, ", "))
}
return nil
}
func (e *executor[T]) forkJoin(tasks []*taskState[T]) []error {
if len(tasks) == 0 {
return nil
}
results := make([]error, len(tasks))
var resultsMutex sync.Mutex
var wg sync.WaitGroup
for i := 0; i < len(tasks); i++ {
wg.Add(1)
go func(ts *taskState[T], index int) {
defer wg.Done()
resultsMutex.Lock()
results[index] = fmt.Errorf("function panic")
resultsMutex.Unlock()
klog.V(2).Infof("Executing task %q: %v\n", ts.key, ts.task)
if taskNormalize, ok := ts.task.(TaskNormalize[T]); ok {
if err := taskNormalize.Normalize(e.context); err != nil {
results[index] = err
return
}
}
result := ts.task.Run(e.context)
resultsMutex.Lock()
results[index] = result
resultsMutex.Unlock()
}(tasks[i], i)
}
wg.Wait()
return results
}