-
Notifications
You must be signed in to change notification settings - Fork 197
/
runnable_job.go
171 lines (147 loc) · 5.15 KB
/
runnable_job.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
package kube
import (
"context"
"fmt"
"time"
"github.com/aquasecurity/starboard/pkg/runner"
batchv1 "k8s.io/api/batch/v1"
corev1 "k8s.io/api/core/v1"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/apimachinery/pkg/runtime"
"k8s.io/apimachinery/pkg/util/wait"
"k8s.io/client-go/informers"
"k8s.io/client-go/kubernetes"
"k8s.io/client-go/tools/cache"
"k8s.io/klog/v2"
"k8s.io/utils/pointer"
"sigs.k8s.io/controller-runtime/pkg/controller/controllerutil"
)
var defaultResyncDuration = 30 * time.Minute
type runnableJob struct {
scheme *runtime.Scheme
clientset kubernetes.Interface
logsReader LogsReader
job *batchv1.Job // job to be run
secrets []*corev1.Secret // secrets that the job references
}
// NewRunnableJob constructs a new Runnable task defined as Kubernetes
// job configuration and secrets that it references.
func NewRunnableJob(
scheme *runtime.Scheme,
clientset kubernetes.Interface,
job *batchv1.Job,
secrets ...*corev1.Secret,
) runner.Runnable {
return &runnableJob{
scheme: scheme,
clientset: clientset,
logsReader: NewLogsReader(clientset),
job: job,
secrets: secrets,
}
}
// Run runs synchronously the task as Kubernetes job.
// It creates Kubernetes job and secrets provided as constructor parameters.
// This method blocks and waits for the job completion or failure.
// For each secret it also sets the owner reference that points to the job
// so when the job is deleted secrets are garbage collected.
func (r *runnableJob) Run(ctx context.Context) error {
informerFactory := informers.NewSharedInformerFactoryWithOptions(
r.clientset,
defaultResyncDuration,
informers.WithNamespace(r.job.Namespace),
)
jobsInformer := informerFactory.Batch().V1().Jobs()
eventsInformer := informerFactory.Core().V1().Events()
var err error
for i, secret := range r.secrets {
klog.V(3).Infof("Creating secret %q", r.job.Namespace+"/"+secret.Name)
r.secrets[i], err = r.clientset.CoreV1().Secrets(r.job.Namespace).Create(ctx, secret, metav1.CreateOptions{})
if err != nil {
return fmt.Errorf("creating secret: %w", err)
}
}
klog.V(3).Infof("Creating job %q", r.job.Namespace+"/"+r.job.Name)
r.job, err = r.clientset.BatchV1().Jobs(r.job.Namespace).Create(ctx, r.job, metav1.CreateOptions{})
if err != nil {
return fmt.Errorf("creating job: %w", err)
}
for i, secret := range r.secrets {
klog.V(3).Infof("Setting owner reference secret %q -> job %q", r.job.Namespace+"/"+secret.Name, r.job.Namespace+"/"+r.job.Name)
err = controllerutil.SetOwnerReference(r.job, secret, r.scheme)
if err != nil {
return fmt.Errorf("setting owner reference: %w", err)
}
klog.V(3).Infof("Updating secret %q", r.job.Namespace+"/"+secret.Name)
r.secrets[i], err = r.clientset.CoreV1().Secrets(r.job.Namespace).Update(ctx, secret, metav1.UpdateOptions{})
if err != nil {
return fmt.Errorf("updating secret: %w", err)
}
}
complete := make(chan error)
jobsInformer.Informer().AddEventHandler(cache.ResourceEventHandlerFuncs{ // nolint: errcheck
UpdateFunc: func(_, newObj interface{}) {
newJob, ok := newObj.(*batchv1.Job)
if !ok {
return
}
if r.job.UID != newJob.UID {
return
}
if len(newJob.Status.Conditions) == 0 {
return
}
switch condition := newJob.Status.Conditions[0]; condition.Type {
case batchv1.JobComplete:
klog.V(3).Infof("Stopping runnable job on task completion with status: %s", batchv1.JobComplete)
complete <- nil
case batchv1.JobFailed:
klog.V(3).Infof("Stopping runnable job on task failure with status: %s", batchv1.JobFailed)
complete <- fmt.Errorf("job failed: %s: %s", condition.Reason, condition.Message)
}
},
})
eventsInformer.Informer().AddEventHandler(cache.ResourceEventHandlerFuncs{ // nolint: errcheck
AddFunc: func(obj interface{}) {
event := obj.(*corev1.Event)
// TODO We might want to look into events associate with the pod controlled by the current scan job.
// For example, when a pod cannot be scheduled due to insufficient resource requests,
// the event will be attached to the pod, not the job.
if event.InvolvedObject.UID != r.job.UID {
return
}
if event.Type == corev1.EventTypeNormal {
klog.V(3).Infof("Event: %s (%s)", event.Message, event.Reason)
}
if event.Type == corev1.EventTypeWarning {
complete <- fmt.Errorf("warning event received: %s (%s)", event.Message, event.Reason)
return
}
},
})
informerFactory.Start(wait.NeverStop)
informerFactory.WaitForCacheSync(wait.NeverStop)
err = <-complete
if err != nil {
r.logTerminatedContainersErrors(ctx)
}
return err
}
func (r *runnableJob) logTerminatedContainersErrors(ctx context.Context) {
statuses, err := r.logsReader.GetTerminatedContainersStatusesByJob(ctx, r.job)
if err != nil {
klog.Errorf("Error while getting terminated containers statuses for job %q", r.job.Namespace+"/"+r.job.Name)
}
for container, status := range statuses {
if status.ExitCode == 0 {
continue
}
klog.Errorf("Container %s terminated with %s: %s", container, status.Reason, status.Message)
}
}
func GetActiveDeadlineSeconds(d time.Duration) *int64 {
if d > 0 {
return pointer.Int64Ptr(int64(d.Seconds()))
}
return nil
}