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
|
package controllers
import (
"context"
"fmt"
"sort"
"time"
"github.com/robfig/cron"
kbatch "k8s.io/api/batch/v1"
corev1 "k8s.io/api/core/v1"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/apimachinery/pkg/runtime"
ref "k8s.io/client-go/tools/reference"
ctrl "sigs.k8s.io/controller-runtime"
"sigs.k8s.io/controller-runtime/pkg/client"
"sigs.k8s.io/controller-runtime/pkg/log"
batchv1 "tutorial.kubebuilder.io/project/api/v1"
)
// CronJobReconciler reconciles a CronJob object
type CronJobReconciler struct {
client.Client
Scheme *runtime.Scheme
Clock
}
type realClock struct{}
func (_ realClock) Now() time.Time { return time.Now() }
// clock knows how to get the current time.
// It can be used to fake out timing for testing.
type Clock interface {
Now() time.Time
}
//+kubebuilder:rbac:groups=batch.tutorial.kubebuilder.io,resources=cronjobs,verbs=get;list;watch;create;update;patch;delete
//+kubebuilder:rbac:groups=batch.tutorial.kubebuilder.io,resources=cronjobs/status,verbs=get;update;patch
//+kubebuilder:rbac:groups=batch.tutorial.kubebuilder.io,resources=cronjobs/finalizers,verbs=update
//+kubebuilder:rbac:groups=batch,resources=jobs,verbs=get;list;watch;create;update;patch;delete
//+kubebuilder:rbac:groups=batch,resources=jobs/status,verbs=get
var (
scheduledTimeAnnotation = "batch.tutorial.kubebuilder.io/scheduled-at"
)
// Reconcile is part of the main kubernetes reconciliation loop which aims to
// move the current state of the cluster closer to the desired state.
// TODO(user): Modify the Reconcile function to compare the state specified by
// the CronJob object against the actual cluster state, and then
// perform operations to make the cluster state reflect the state specified by
// the user.
//
// For more details, check Reconcile and its Result here:
// - https://pkg.go.dev/sigs.k8s.io/controller-runtime@v0.12.1/pkg/reconcile
func (r *CronJobReconciler) Reconcile(ctx context.Context, req ctrl.Request) (ctrl.Result, error) {
log := log.FromContext(ctx)
var cronJob batchv1.CronJob
if err := r.Get(ctx, req.NamespacedName, &cronJob); err != nil {
log.Error(err, "unable to fetch CronJob")
// we'll ignore not-found errors, since they can't be fixed by an immediate
// requeue (we'll need to wait for a new notification), and we can get them
// on deleted requests.
return ctrl.Result{}, client.IgnoreNotFound(err)
}
var childJobs kbatch.JobList
if err := r.List(ctx, &childJobs, client.InNamespace(req.Namespace), client.MatchingFields{jobOwnerKey: req.Name}); err != nil {
log.Error(err, "unable to list child Jobs")
return ctrl.Result{}, err
}
// find the active list of jobs
var activeJobs []*kbatch.Job
var successfulJobs []*kbatch.Job
var failedJobs []*kbatch.Job
var mostRecentTime *time.Time // find the last run so we can update the status
isJobFinished := func(job *kbatch.Job) (bool, kbatch.JobConditionType) {
for _, c := range job.Status.Conditions {
if (c.Type == kbatch.JobComplete || c.Type == kbatch.JobFailed) && c.Status == corev1.ConditionTrue {
return true, c.Type
}
}
return false, ""
}
getScheduledTimeForJob := func(job *kbatch.Job) (*time.Time, error) {
timeRaw := job.Annotations[scheduledTimeAnnotation]
if len(timeRaw) == 0 {
return nil, nil
}
timeParsed, err := time.Parse(time.RFC3339, timeRaw)
if err != nil {
return nil, err
}
return &timeParsed, nil
}
for i, job := range childJobs.Items {
_, finishedType := isJobFinished(&job)
switch finishedType {
case "": // ongoing
activeJobs = append(activeJobs, &childJobs.Items[i])
case kbatch.JobFailed:
failedJobs = append(failedJobs, &childJobs.Items[i])
case kbatch.JobComplete:
successfulJobs = append(successfulJobs, &childJobs.Items[i])
}
// We'll store the launch time in an annotation, so we'll reconstitute that from
// the active jobs themselves.
scheduledTimeForJob, err := getScheduledTimeForJob(&job)
if err != nil {
log.Error(err, "unable to parse schedule time for child job", "job", &job)
continue
}
if scheduledTimeForJob != nil {
if mostRecentTime == nil {
mostRecentTime = scheduledTimeForJob
} else if mostRecentTime.Before(*scheduledTimeForJob) {
mostRecentTime = scheduledTimeForJob
}
}
}
if mostRecentTime != nil {
cronJob.Status.LastScheduleTime = &metav1.Time{Time: *mostRecentTime}
} else {
cronJob.Status.LastScheduleTime = nil
}
cronJob.Status.Active = nil
for _, activeJob := range activeJobs {
jobRef, err := ref.GetReference(r.Scheme, activeJob)
if err != nil {
log.Error(err, "unable to make reference to active job", "job", activeJob)
continue
}
cronJob.Status.Active = append(cronJob.Status.Active, *jobRef)
}
log.V(1).Info("job count", "active jobs", len(activeJobs), "successful jobs", len(successfulJobs), "failed jobs", len(failedJobs))
if err := r.Status().Update(ctx, &cronJob); err != nil {
log.Error(err, "unable to update CronJob status")
return ctrl.Result{}, err
}
// NB: deleting these is "best effort" -- if we fail on a particular one,
// we won't requeue just to finish the deleting.
if cronJob.Spec.FailedJobsHistoryLimit != nil {
sort.Slice(failedJobs, func(i, j int) bool {
if failedJobs[i].Status.StartTime == nil {
return failedJobs[j].Status.StartTime != nil
}
return failedJobs[i].Status.StartTime.Before(failedJobs[j].Status.StartTime)
})
for i, job := range failedJobs {
if int32(i) >= int32(len(failedJobs))-*cronJob.Spec.FailedJobsHistoryLimit {
break
}
if err := r.Delete(ctx, job, client.PropagationPolicy(metav1.DeletePropagationBackground)); client.IgnoreNotFound(err) != nil {
log.Error(err, "unable to delete old failed job", "job", job)
} else {
log.V(0).Info("deleted old failed job", "job", job)
}
}
}
if cronJob.Spec.SuccessfulJobsHistoryLimit != nil {
sort.Slice(successfulJobs, func(i, j int) bool {
if successfulJobs[i].Status.StartTime == nil {
return successfulJobs[j].Status.StartTime != nil
}
return successfulJobs[i].Status.StartTime.Before(successfulJobs[j].Status.StartTime)
})
for i, job := range successfulJobs {
if int32(i) >= int32(len(successfulJobs))-*cronJob.Spec.SuccessfulJobsHistoryLimit {
break
}
if err := r.Delete(ctx, job, client.PropagationPolicy(metav1.DeletePropagationBackground)); (err) != nil {
log.Error(err, "unable to delete old successful job", "job", job)
} else {
log.V(0).Info("deleted old successful job", "job", job)
}
}
}
if cronJob.Spec.Suspend != nil && *cronJob.Spec.Suspend {
log.V(1).Info("cronjob suspended, skipping")
return ctrl.Result{}, nil
}
getNextSchedule := func(cronJob *batchv1.CronJob, now time.Time) (lastMissed time.Time, next time.Time, err error) {
sched, err := cron.ParseStandard(cronJob.Spec.Schedule)
if err != nil {
return time.Time{}, time.Time{}, fmt.Errorf("Unparseable schedule %q: %v", cronJob.Spec.Schedule, err)
}
// for optimization purposes, cheat a bit and start from our last observed run time
// we could reconstitute this here, but there's not much point, since we've
// just updated it.
var earliestTime time.Time
if cronJob.Status.LastScheduleTime != nil {
earliestTime = cronJob.Status.LastScheduleTime.Time
} else {
earliestTime = cronJob.ObjectMeta.CreationTimestamp.Time
}
if cronJob.Spec.StartingDeadlineSeconds != nil {
// controller is not going to schedule anything below this point
schedulingDeadline := now.Add(-time.Second * time.Duration(*cronJob.Spec.StartingDeadlineSeconds))
if schedulingDeadline.After(earliestTime) {
earliestTime = schedulingDeadline
}
}
if earliestTime.After(now) {
return time.Time{}, sched.Next(now), nil
}
starts := 0
for t := sched.Next(earliestTime); !t.After(now); t = sched.Next(t) {
lastMissed = t
// An object might miss several starts. For example, if
// controller gets wedged on Friday at 5:01pm when everyone has
// gone home, and someone comes in on Tuesday AM and discovers
// the problem and restarts the controller, then all the hourly
// jobs, more than 80 of them for one hourly scheduledJob, should
// all start running with no further intervention (if the scheduledJob
// allows concurrency and late starts).
//
// However, if there is a bug somewhere, or incorrect clock
// on controller's server or apiservers (for setting creationTimestamp)
// then there could be so many missed start times (it could be off
// by decades or more), that it would eat up all the CPU and memory
// of this controller. In that case, we want to not try to list
// all the missed start times.
starts++
if starts > 100 {
// We can't get the most recent times so just return an empty slice
return time.Time{}, time.Time{}, fmt.Errorf("Too many missed start times (> 100). Set or decrease .spec.startingDeadlineSeconds or check clock skew.")
}
}
return lastMissed, sched.Next(now), nil
}
// figure out the next times that we need to create
// jobs at (or anything we missed).
missedRun, nextRun, err := getNextSchedule(&cronJob, r.Now())
if err != nil {
log.Error(err, "unable to figure out CronJob schedule")
// we don't really care about requeuing until we get an update that
// fixes the schedule, so don't return an error
return ctrl.Result{}, nil
}
scheduledResult := ctrl.Result{RequeueAfter: nextRun.Sub(r.Now())} // save this so we can re-use it elsewhere
log = log.WithValues("now", r.Now(), "next run", nextRun)
if missedRun.IsZero() {
log.V(1).Info("no upcoming scheduled times, sleeping until next")
return scheduledResult, nil
}
// make sure we're not too late to start the run
log = log.WithValues("current run", missedRun)
tooLate := false
if cronJob.Spec.StartingDeadlineSeconds != nil {
tooLate = missedRun.Add(time.Duration(*cronJob.Spec.StartingDeadlineSeconds) * time.Second).Before(r.Now())
}
if tooLate {
log.V(1).Info("missed starting deadline for last run, sleeping till next")
// TODO(directxman12): events
return scheduledResult, nil
}
// figure out how to run this job -- concurrency policy might forbid us from running
// multiple at the same time...
if cronJob.Spec.ConcurrencyPolicy == batchv1.ForbidConcurrent && len(activeJobs) > 0 {
log.V(1).Info("concurrency policy blocks concurrent runs, skipping", "num active", len(activeJobs))
return scheduledResult, nil
}
// ...or instruct us to replace existing ones...
if cronJob.Spec.ConcurrencyPolicy == batchv1.ReplaceConcurrent {
for _, activeJob := range activeJobs {
// we don't care if the job was already deleted
if err := r.Delete(ctx, activeJob, client.PropagationPolicy(metav1.DeletePropagationBackground)); client.IgnoreNotFound(err) != nil {
log.Error(err, "unable to delete active job", "job", activeJob)
return ctrl.Result{}, err
}
}
}
constructJobForCronJob := func(cronJob *batchv1.CronJob, scheduledTime time.Time) (*kbatch.Job, error) {
// We want job names for a given nominal start time to have a deterministic name to avoid the same job being created twice
name := fmt.Sprintf("%s-%d", cronJob.Name, scheduledTime.Unix())
job := &kbatch.Job{
ObjectMeta: metav1.ObjectMeta{
Labels: make(map[string]string),
Annotations: make(map[string]string),
Name: name,
Namespace: cronJob.Namespace,
},
Spec: *cronJob.Spec.JobTemplate.Spec.DeepCopy(),
}
for k, v := range cronJob.Spec.JobTemplate.Annotations {
job.Annotations[k] = v
}
job.Annotations[scheduledTimeAnnotation] = scheduledTime.Format(time.RFC3339)
for k, v := range cronJob.Spec.JobTemplate.Labels {
job.Labels[k] = v
}
if err := ctrl.SetControllerReference(cronJob, job, r.Scheme); err != nil {
return nil, err
}
return job, nil
}
// actually make the job...
job, err := constructJobForCronJob(&cronJob, missedRun)
if err != nil {
log.Error(err, "unable to construct job from template")
// don't bother requeuing until we get a change to the spec
return scheduledResult, nil
}
// ...and create it on the cluster
if err := r.Create(ctx, job); err != nil {
log.Error(err, "unable to create Job for CronJob", "job", job)
return ctrl.Result{}, err
}
log.V(1).Info("created Job for CronJob run", "job", job)
// we'll requeue once we see the running job, and update our status
return scheduledResult, nil
}
var (
jobOwnerKey = ".metadata.controller"
apiGVStr = batchv1.GroupVersion.String()
)
func (r *CronJobReconciler) SetupWithManager(mgr ctrl.Manager) error {
// set up a real clock, since we're not in a test
if r.Clock == nil {
r.Clock = realClock{}
}
if err := mgr.GetFieldIndexer().IndexField(context.Background(), &kbatch.Job{}, jobOwnerKey, func(rawObj client.Object) []string {
// grab the job object, extract the owner...
job := rawObj.(*kbatch.Job)
owner := metav1.GetControllerOf(job)
if owner == nil {
return nil
}
// ...make sure it's a CronJob...
if owner.APIVersion != apiGVStr || owner.Kind != "CronJob" {
return nil
}
// ...and if so, return it
return []string{owner.Name}
}); err != nil {
return err
}
return ctrl.NewControllerManagedBy(mgr).
For(&batchv1.CronJob{}).
Owns(&kbatch.Job{}).
Complete(r)
}
|
