forked from openshift/velero
/
schedule_controller.go
408 lines (339 loc) · 12.7 KB
/
schedule_controller.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
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
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
/*
Copyright 2017 the Heptio Ark contributors.
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 controller
import (
"context"
"encoding/json"
"fmt"
"sync"
"time"
jsonpatch "github.com/evanphx/json-patch"
"github.com/pkg/errors"
"github.com/robfig/cron"
"github.com/sirupsen/logrus"
apierrors "k8s.io/apimachinery/pkg/api/errors"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/apimachinery/pkg/labels"
"k8s.io/apimachinery/pkg/types"
"k8s.io/apimachinery/pkg/util/clock"
"k8s.io/apimachinery/pkg/util/wait"
"k8s.io/client-go/tools/cache"
"k8s.io/client-go/util/workqueue"
api "github.com/heptio/ark/pkg/apis/ark/v1"
arkv1client "github.com/heptio/ark/pkg/generated/clientset/versioned/typed/ark/v1"
informers "github.com/heptio/ark/pkg/generated/informers/externalversions/ark/v1"
listers "github.com/heptio/ark/pkg/generated/listers/ark/v1"
"github.com/heptio/ark/pkg/metrics"
kubeutil "github.com/heptio/ark/pkg/util/kube"
)
type scheduleController struct {
namespace string
schedulesClient arkv1client.SchedulesGetter
backupsClient arkv1client.BackupsGetter
schedulesLister listers.ScheduleLister
schedulesListerSynced cache.InformerSynced
syncHandler func(scheduleName string) error
queue workqueue.RateLimitingInterface
syncPeriod time.Duration
clock clock.Clock
logger logrus.FieldLogger
metrics *metrics.ServerMetrics
}
func NewScheduleController(
namespace string,
schedulesClient arkv1client.SchedulesGetter,
backupsClient arkv1client.BackupsGetter,
schedulesInformer informers.ScheduleInformer,
syncPeriod time.Duration,
logger logrus.FieldLogger,
metrics *metrics.ServerMetrics,
) *scheduleController {
if syncPeriod < time.Minute {
logger.WithField("syncPeriod", syncPeriod).Info("Provided schedule sync period is too short. Setting to 1 minute")
syncPeriod = time.Minute
}
c := &scheduleController{
namespace: namespace,
schedulesClient: schedulesClient,
backupsClient: backupsClient,
schedulesLister: schedulesInformer.Lister(),
schedulesListerSynced: schedulesInformer.Informer().HasSynced,
queue: workqueue.NewNamedRateLimitingQueue(workqueue.DefaultControllerRateLimiter(), "schedule"),
syncPeriod: syncPeriod,
clock: clock.RealClock{},
logger: logger,
metrics: metrics,
}
c.syncHandler = c.processSchedule
schedulesInformer.Informer().AddEventHandler(
cache.ResourceEventHandlerFuncs{
AddFunc: func(obj interface{}) {
schedule := obj.(*api.Schedule)
switch schedule.Status.Phase {
case "", api.SchedulePhaseNew, api.SchedulePhaseEnabled:
// add to work queue
default:
c.logger.WithFields(logrus.Fields{
"schedule": kubeutil.NamespaceAndName(schedule),
"phase": schedule.Status.Phase,
}).Debug("Schedule is not new, skipping")
return
}
key, err := cache.MetaNamespaceKeyFunc(schedule)
if err != nil {
c.logger.WithError(errors.WithStack(err)).WithField("schedule", schedule).Error("Error creating queue key, item not added to queue")
return
}
c.queue.Add(key)
scheduleName := schedule.GetName()
c.logger.Info("Creating schedule ", scheduleName)
//Init Prometheus metrics to 0 to have them flowing up
metrics.InitSchedule(scheduleName)
},
},
)
return c
}
// Run is a blocking function that runs the specified number of worker goroutines
// to process items in the work queue. It will return when it receives on the
// ctx.Done() channel.
func (controller *scheduleController) Run(ctx context.Context, numWorkers int) error {
var wg sync.WaitGroup
defer func() {
controller.logger.Info("Waiting for workers to finish their work")
controller.queue.ShutDown()
// We have to wait here in the deferred function instead of at the bottom of the function body
// because we have to shut down the queue in order for the workers to shut down gracefully, and
// we want to shut down the queue via defer and not at the end of the body.
wg.Wait()
controller.logger.Info("All workers have finished")
}()
controller.logger.Info("Starting ScheduleController")
defer controller.logger.Info("Shutting down ScheduleController")
controller.logger.Info("Waiting for caches to sync")
if !cache.WaitForCacheSync(ctx.Done(), controller.schedulesListerSynced) {
return errors.New("timed out waiting for caches to sync")
}
controller.logger.Info("Caches are synced")
wg.Add(numWorkers)
for i := 0; i < numWorkers; i++ {
go func() {
wait.Until(controller.runWorker, time.Second, ctx.Done())
wg.Done()
}()
}
go wait.Until(controller.enqueueAllEnabledSchedules, controller.syncPeriod, ctx.Done())
<-ctx.Done()
return nil
}
func (controller *scheduleController) enqueueAllEnabledSchedules() {
schedules, err := controller.schedulesLister.Schedules(controller.namespace).List(labels.NewSelector())
if err != nil {
controller.logger.WithError(errors.WithStack(err)).Error("Error listing Schedules")
return
}
for _, schedule := range schedules {
if schedule.Status.Phase != api.SchedulePhaseEnabled {
continue
}
key, err := cache.MetaNamespaceKeyFunc(schedule)
if err != nil {
controller.logger.WithError(errors.WithStack(err)).WithField("schedule", schedule).Error("Error creating queue key, item not added to queue")
continue
}
controller.queue.Add(key)
}
}
func (controller *scheduleController) runWorker() {
// continually take items off the queue (waits if it's
// empty) until we get a shutdown signal from the queue
for controller.processNextWorkItem() {
}
}
func (controller *scheduleController) processNextWorkItem() bool {
key, quit := controller.queue.Get()
if quit {
return false
}
// always call done on this item, since if it fails we'll add
// it back with rate-limiting below
defer controller.queue.Done(key)
err := controller.syncHandler(key.(string))
if err == nil {
// If you had no error, tell the queue to stop tracking history for your key. This will reset
// things like failure counts for per-item rate limiting.
controller.queue.Forget(key)
return true
}
controller.logger.WithError(err).WithField("key", key).Error("Error in syncHandler, re-adding item to queue")
// we had an error processing the item so add it back
// into the queue for re-processing with rate-limiting
controller.queue.AddRateLimited(key)
return true
}
func (controller *scheduleController) processSchedule(key string) error {
logContext := controller.logger.WithField("key", key)
logContext.Debug("Running processSchedule")
ns, name, err := cache.SplitMetaNamespaceKey(key)
if err != nil {
return errors.Wrap(err, "error splitting queue key")
}
logContext.Debug("Getting Schedule")
schedule, err := controller.schedulesLister.Schedules(ns).Get(name)
if err != nil {
// schedule no longer exists
if apierrors.IsNotFound(err) {
logContext.WithError(err).Debug("Schedule not found")
return nil
}
return errors.Wrap(err, "error getting Schedule")
}
switch schedule.Status.Phase {
case "", api.SchedulePhaseNew, api.SchedulePhaseEnabled:
// valid phase for processing
default:
return nil
}
logContext.Debug("Cloning schedule")
// store ref to original for creating patch
original := schedule
// don't modify items in the cache
schedule = schedule.DeepCopy()
// validation - even if the item is Enabled, we can't trust it
// so re-validate
currentPhase := schedule.Status.Phase
cronSchedule, errs := parseCronSchedule(schedule, controller.logger)
if len(errs) > 0 {
schedule.Status.Phase = api.SchedulePhaseFailedValidation
schedule.Status.ValidationErrors = errs
} else {
schedule.Status.Phase = api.SchedulePhaseEnabled
}
// update status if it's changed
if currentPhase != schedule.Status.Phase {
updatedSchedule, err := patchSchedule(original, schedule, controller.schedulesClient)
if err != nil {
return errors.Wrapf(err, "error updating Schedule phase to %s", schedule.Status.Phase)
}
schedule = updatedSchedule
}
if schedule.Status.Phase != api.SchedulePhaseEnabled {
return nil
}
// check for the schedule being due to run, and submit a Backup if so
if err := controller.submitBackupIfDue(schedule, cronSchedule); err != nil {
return err
}
return nil
}
func parseCronSchedule(itm *api.Schedule, logger logrus.FieldLogger) (cron.Schedule, []string) {
var validationErrors []string
var schedule cron.Schedule
// cron.Parse panics if schedule is empty
if len(itm.Spec.Schedule) == 0 {
validationErrors = append(validationErrors, "Schedule must be a non-empty valid Cron expression")
return nil, validationErrors
}
logContext := logger.WithField("schedule", kubeutil.NamespaceAndName(itm))
// adding a recover() around cron.Parse because it panics on empty string and is possible
// that it panics under other scenarios as well.
func() {
defer func() {
if r := recover(); r != nil {
logContext.WithFields(logrus.Fields{
"schedule": itm.Spec.Schedule,
"recover": r,
}).Debug("Panic parsing schedule")
validationErrors = append(validationErrors, fmt.Sprintf("invalid schedule: %v", r))
}
}()
if res, err := cron.ParseStandard(itm.Spec.Schedule); err != nil {
logContext.WithError(errors.WithStack(err)).WithField("schedule", itm.Spec.Schedule).Debug("Error parsing schedule")
validationErrors = append(validationErrors, fmt.Sprintf("invalid schedule: %v", err))
} else {
schedule = res
}
}()
if len(validationErrors) > 0 {
return nil, validationErrors
}
return schedule, nil
}
func (controller *scheduleController) submitBackupIfDue(item *api.Schedule, cronSchedule cron.Schedule) error {
var (
now = controller.clock.Now()
isDue, nextRunTime = getNextRunTime(item, cronSchedule, now)
logContext = controller.logger.WithField("schedule", kubeutil.NamespaceAndName(item))
)
if !isDue {
logContext.WithField("nextRunTime", nextRunTime).Info("Schedule is not due, skipping")
return nil
}
// Don't attempt to "catch up" if there are any missed or failed runs - simply
// trigger a Backup if it's time.
//
// It might also make sense in the future to explicitly check for currently-running
// backups so that we don't overlap runs (for disk snapshots in particular, this can
// lead to performance issues).
logContext.WithField("nextRunTime", nextRunTime).Info("Schedule is due, submitting Backup")
backup := getBackup(item, now)
if _, err := controller.backupsClient.Backups(backup.Namespace).Create(backup); err != nil {
return errors.Wrap(err, "error creating Backup")
}
original := item
schedule := item.DeepCopy()
schedule.Status.LastBackup = metav1.NewTime(now)
if _, err := patchSchedule(original, schedule, controller.schedulesClient); err != nil {
return errors.Wrapf(err, "error updating Schedule's LastBackup time to %v", schedule.Status.LastBackup)
}
return nil
}
func getNextRunTime(schedule *api.Schedule, cronSchedule cron.Schedule, asOf time.Time) (bool, time.Time) {
// get the latest run time (if the schedule hasn't run yet, this will be the zero value which will trigger
// an immediate backup)
lastBackupTime := schedule.Status.LastBackup.Time
nextRunTime := cronSchedule.Next(lastBackupTime)
return asOf.After(nextRunTime), nextRunTime
}
func getBackup(item *api.Schedule, timestamp time.Time) *api.Backup {
backup := &api.Backup{
Spec: item.Spec.Template,
ObjectMeta: metav1.ObjectMeta{
Namespace: item.Namespace,
Name: fmt.Sprintf("%s-%s", item.Name, timestamp.Format("20060102150405")),
Labels: map[string]string{
"ark-schedule": item.Name,
},
},
}
return backup
}
func patchSchedule(original, updated *api.Schedule, client arkv1client.SchedulesGetter) (*api.Schedule, error) {
origBytes, err := json.Marshal(original)
if err != nil {
return nil, errors.Wrap(err, "error marshalling original schedule")
}
updatedBytes, err := json.Marshal(updated)
if err != nil {
return nil, errors.Wrap(err, "error marshalling updated schedule")
}
patchBytes, err := jsonpatch.CreateMergePatch(origBytes, updatedBytes)
if err != nil {
return nil, errors.Wrap(err, "error creating json merge patch for schedule")
}
res, err := client.Schedules(original.Namespace).Patch(original.Name, types.MergePatchType, patchBytes)
if err != nil {
return nil, errors.Wrap(err, "error patching schedule")
}
return res, nil
}