-
Notifications
You must be signed in to change notification settings - Fork 2
/
controller.go
481 lines (420 loc) · 12.7 KB
/
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
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
package kubebench
import (
"context"
"errors"
"fmt"
"hash/fnv"
"hash/maphash"
"io"
"reflect"
"sync"
"time"
castaipb "github.com/castai/kvisor/api/v1/runtime"
"github.com/castai/kvisor/cmd/controller/kube"
"github.com/castai/kvisor/cmd/controller/state/kubebench/spec"
"github.com/castai/kvisor/pkg/logging"
"github.com/cenkalti/backoff/v4"
"github.com/google/uuid"
lru "github.com/hashicorp/golang-lru"
jsoniter "github.com/json-iterator/go"
"github.com/samber/lo"
"google.golang.org/grpc"
"google.golang.org/grpc/encoding/gzip"
batchv1 "k8s.io/api/batch/v1"
corev1 "k8s.io/api/core/v1"
k8serrors "k8s.io/apimachinery/pkg/api/errors"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/apimachinery/pkg/labels"
"k8s.io/client-go/kubernetes"
)
type castaiClient interface {
KubeBenchReportIngest(ctx context.Context, in *castaipb.KubeBenchReport, opts ...grpc.CallOption) (*castaipb.KubeBenchReportIngestResponse, error)
}
const (
nodeScanTimeout = 5 * time.Minute
labelJobName = "job-name"
maxConcurrentJobs = 1
)
type kubeController interface {
GetKvisorAgentImageDetails() (kube.ImageDetails, bool)
}
type Config struct {
Enabled bool
Force bool
ScanInterval time.Duration `validate:"required"`
JobImagePullPolicy string
CloudProvider string
JobNamespace string
}
func NewController(
log *logging.Logger,
client kubernetes.Interface,
cfg Config,
castClient castaiClient,
logsReader kube.PodLogProvider,
kubeController kubeController,
scannedNodes []string,
) *Controller {
nodeCache, _ := lru.New(1000)
for _, node := range scannedNodes {
nodeCache.Add(node, struct{}{})
}
return &Controller{
log: log.WithField("component", "kubebench"),
client: client,
cfg: cfg,
delta: newDeltaState(),
castClient: castClient,
logsProvider: logsReader,
kubeController: kubeController,
scannedNodes: nodeCache,
finishedJobDeleteWaitDuration: 10 * time.Second,
kubeBenchReportsCache: map[uint64]*castaipb.KubeBenchReport{},
}
}
type Controller struct {
log *logging.Logger
client kubernetes.Interface
cfg Config
castClient castaiClient
delta *nodeDeltaState
logsProvider kube.PodLogProvider
kubeController kubeController
finishedJobDeleteWaitDuration time.Duration
scannedNodes *lru.Cache
// After job finishes with store report in memory grouped by similar nodes.
// This allows to reduce number of jobs since we near identical reports.
kubeBenchReportsCache map[uint64]*castaipb.KubeBenchReport
kubeBenchReportsCacheMu sync.Mutex
}
func (c *Controller) RequiredTypes() []reflect.Type {
return []reflect.Type{
reflect.TypeOf(&corev1.Node{}),
}
}
func (c *Controller) OnAdd(obj kube.Object) {
node, ok := obj.(*corev1.Node)
if ok {
_, scanned := c.scannedNodes.Get(string(node.GetUID()))
if isNodeReady(node) && !scanned {
c.delta.upsert(node)
}
}
}
func (c *Controller) OnUpdate(obj kube.Object) {
c.OnAdd(obj)
}
func (c *Controller) OnDelete(obj kube.Object) {
node, ok := obj.(*corev1.Node)
if ok {
c.delta.delete(node)
c.scannedNodes.Remove(string(obj.GetUID()))
}
}
func (c *Controller) Run(ctx context.Context) error {
c.log.Info("running")
defer c.log.Infof("stopping")
ticker := time.NewTicker(c.cfg.ScanInterval)
defer ticker.Stop()
for {
select {
case <-ctx.Done():
return ctx.Err()
case <-ticker.C:
err := c.process(ctx)
if err != nil && !errors.Is(err, context.Canceled) {
c.log.Errorf("error linting nodes: %v", err)
}
}
}
}
func (c *Controller) process(ctx context.Context) (rerr error) {
nodes := c.findNodesForScan()
if len(nodes) == 0 {
return nil
}
c.log.Infof("processing kube-bench")
defer c.log.Info("processing kube-bench done")
var wg sync.WaitGroup
for _, n := range nodes {
job := n
wg.Add(1)
go func() {
defer wg.Done()
ctx, cancel := context.WithTimeout(ctx, nodeScanTimeout)
defer cancel()
err := c.lintNode(ctx, job.node)
if err != nil {
c.log.WithField("node", job.node.Name).Errorf("kube-bench: %v", err)
job.setFailed()
return
}
c.delta.delete(job.node)
}()
}
wg.Wait()
return nil
}
func (c *Controller) findNodesForScan() []*nodeJob {
nodes := c.delta.peek()
var res []*nodeJob
for _, nodeJob := range nodes {
if nodeJob.ready() && len(nodeJob.node.Spec.Taints) == 0 {
res = append(res, nodeJob)
if len(res) == maxConcurrentJobs {
break
}
}
}
return res
}
func (c *Controller) RequiredInformers() []reflect.Type {
return []reflect.Type{reflect.TypeOf(&corev1.Node{})}
}
func (c *Controller) lintNode(ctx context.Context, node *corev1.Node) (rerr error) {
// Check if node is in node group for already scanned jobs.
// In found we can skip job scheduling.
if cachedReport, found := c.findScannedReport(node); found {
report := &castaipb.KubeBenchReport{
Controls: cachedReport.Controls,
Node: &castaipb.KubeBenchNode{
NodeName: node.Name,
ResourceUid: string(node.UID),
},
}
_, err := c.castClient.KubeBenchReportIngest(ctx, report)
if err != nil {
return err
}
c.scannedNodes.Add(string(node.UID), struct{}{})
return nil
}
c.log.Debugf("starting kube-bench lint for node=%s", node.Name)
jobName := generateName(node.GetName())
err := c.deleteJob(ctx, jobName)
if err != nil && !k8serrors.IsNotFound(err) {
c.log.Errorf("can not delete job %q: %v", jobName, err)
return err
}
if err == nil {
err = c.waitJobDeleted(ctx, jobName)
if err != nil {
return err
}
}
kubeBenchPod, err := c.createKubebenchJob(ctx, node, jobName)
if err != nil {
return err
}
report, err := c.getReportFromLogs(ctx, node, kubeBenchPod.Name)
if err != nil {
return fmt.Errorf("reading kube-bench report from pod logs: %w", err)
}
c.addReportToCache(node, report)
_, err = c.castClient.KubeBenchReportIngest(ctx, report, grpc.UseCompressor(gzip.Name))
if err != nil {
return err
}
c.scannedNodes.Add(string(node.UID), struct{}{})
if c.finishedJobDeleteWaitDuration != 0 {
go func() {
// Wait some time before deleting job. This is useful for observability and e2e tests.
select {
case <-ctx.Done():
return
case <-time.After(c.finishedJobDeleteWaitDuration):
}
err = c.deleteJob(ctx, jobName)
if err != nil {
c.log.Errorf("failed deleting job %s: %v", jobName, err)
}
}()
} else {
err = c.deleteJob(ctx, jobName)
if err != nil {
c.log.Errorf("failed deleting job %s: %v", jobName, err)
}
}
return nil
}
func (c *Controller) findScannedReport(n *corev1.Node) (*castaipb.KubeBenchReport, bool) {
c.kubeBenchReportsCacheMu.Lock()
defer c.kubeBenchReportsCacheMu.Unlock()
key := getNodeGroupKey(n)
report, found := c.kubeBenchReportsCache[key]
return report, found
}
func (c *Controller) addReportToCache(n *corev1.Node, report *castaipb.KubeBenchReport) {
c.kubeBenchReportsCacheMu.Lock()
defer c.kubeBenchReportsCacheMu.Unlock()
key := getNodeGroupKey(n)
c.kubeBenchReportsCache[key] = report
}
// We are interested in kube-bench pod succeeding and not the Job
func (c *Controller) createKubebenchJob(ctx context.Context, node *corev1.Node, jobName string) (*corev1.Pod, error) {
specFn := resolveSpec(c.cfg.CloudProvider, node)
jobSpec := specFn(node.GetName(), jobName)
// Set image.
imageDetails, found := c.kubeController.GetKvisorAgentImageDetails()
if !found {
return nil, errors.New("kvisor image details not found")
}
cont := jobSpec.Spec.Template.Spec.Containers[0]
cont.Image = imageDetails.ScannerImageName
cont.ImagePullPolicy = corev1.PullPolicy(c.cfg.JobImagePullPolicy)
jobSpec.Spec.Template.Spec.Containers[0] = cont
jobSpec.Spec.Template.Spec.ImagePullSecrets = imageDetails.ImagePullSecrets
job, err := c.client.BatchV1().
Jobs(c.cfg.JobNamespace).
Create(ctx, jobSpec, metav1.CreateOptions{})
if err != nil {
c.log.Errorf("can not create kube-bench scan job: %v", err)
return nil, err
}
selector := labels.Set{labelJobName: job.Name}
var kubeBenchPod *corev1.Pod
podCtx, cancel := context.WithTimeout(ctx, 1*time.Minute)
defer cancel()
err = backoff.Retry(
func() error {
pods, err := c.client.CoreV1().Pods(c.cfg.JobNamespace).List(ctx, metav1.ListOptions{
LabelSelector: selector.String(),
})
if err != nil {
return err
}
if len(pods.Items) < 1 {
return fmt.Errorf("pod not found")
}
kubeBenchPod = &pods.Items[0]
if kubeBenchPod.Status.Phase == corev1.PodFailed {
return backoff.Permanent(fmt.Errorf("kube-bench failed: %s", kubeBenchPod.Status.Message))
}
if kubeBenchPod.Status.Phase == corev1.PodSucceeded {
return nil
}
return fmt.Errorf("unknown err")
}, backoff.WithContext(backoff.NewConstantBackOff(10*time.Second), podCtx))
if err != nil {
return nil, err
}
return kubeBenchPod, nil
}
func (c *Controller) deleteJob(ctx context.Context, jobName string) error {
return c.client.BatchV1().Jobs(c.cfg.JobNamespace).Delete(ctx, jobName, metav1.DeleteOptions{
GracePeriodSeconds: lo.ToPtr(int64(0)),
PropagationPolicy: lo.ToPtr(metav1.DeletePropagationBackground),
})
}
func (c *Controller) waitJobDeleted(ctx context.Context, jobName string) error {
deleteCtx, cancel := context.WithTimeout(ctx, 1*time.Minute)
defer cancel()
return backoff.Retry(
func() error {
_, err := c.client.BatchV1().Jobs(c.cfg.JobNamespace).Get(deleteCtx, jobName, metav1.GetOptions{})
if err != nil {
if k8serrors.IsNotFound(err) {
return nil
}
return backoff.Permanent(err)
}
return fmt.Errorf("job not yet deleted")
}, backoff.WithContext(backoff.NewConstantBackOff(10*time.Second), deleteCtx))
}
func (c *Controller) getReportFromLogs(ctx context.Context, node *corev1.Node, kubeBenchPodName string) (*castaipb.KubeBenchReport, error) {
logReader, err := c.logsProvider.GetLogReader(ctx, c.cfg.JobNamespace, kubeBenchPodName)
if err != nil {
return nil, err
}
defer logReader.Close()
report, err := io.ReadAll(logReader)
if err != nil {
return nil, err
}
var customReport KubeBenchReport
err = jsoniter.Unmarshal(report, &customReport)
if err != nil {
return nil, err
}
if len(customReport.Controls) == 0 {
c.log.Infof("no checks found for node: %s", node.Name)
return nil, nil
}
nodeID, err := uuid.Parse(string(node.UID))
if err != nil {
return nil, fmt.Errorf("can't parse node UID: %w", err)
}
res := &castaipb.KubeBenchReport{
Controls: lo.Map(customReport.Controls, func(item *Controls, index int) *castaipb.KubeBenchControls {
return &castaipb.KubeBenchControls{
Version: item.Version,
Groups: lo.Map(item.Groups, func(item *Group, index int) *castaipb.KubeBenchGroup {
return &castaipb.KubeBenchGroup{
Checks: lo.Map(item.Checks, func(item *Check, index int) *castaipb.KubeBenchCheck {
return &castaipb.KubeBenchCheck{
Id: item.ID,
Text: item.Text,
TestInfo: item.TestInfo,
State: string(item.State),
}
}),
}
}),
}
}),
Node: &castaipb.KubeBenchNode{
NodeName: node.Name,
ResourceUid: nodeID.String(),
},
}
return res, nil
}
func generateName(nodeName string) string {
h := fnv.New32a()
h.Write([]byte(nodeName))
return fmt.Sprintf("castai-kube-bench-%d", h.Sum32())
}
func resolveSpec(provider string, node *corev1.Node) func(nodeName, jobname string) *batchv1.Job {
switch provider {
case "gke":
return spec.GKE
case "aks":
return spec.AKS
case "eks":
return spec.EKS
default:
labels := node.GetLabels()
if _, ok := labels["node-role.kubernetes.io/control-plane"]; ok {
return spec.Master
}
return spec.Node
}
}
func isNodeReady(n *corev1.Node) bool {
if len(n.Status.Conditions) == 0 {
return false
}
lastCondition := n.Status.Conditions[len(n.Status.Conditions)-1]
return lastCondition.Type == corev1.NodeReady && lastCondition.Status == corev1.ConditionTrue
}
var nodeGroupsHashSeed maphash.Seed
func init() {
nodeGroupsHashSeed = maphash.MakeSeed()
}
// getNodeGroupKey creates hash for nodes group.
// Thread safe.
func getNodeGroupKey(n *corev1.Node) uint64 {
var hash maphash.Hash
hash.SetSeed(nodeGroupsHashSeed)
if v, found := n.Labels["provisioner.cast.ai/node-configuration-name"]; found {
_, _ = hash.WriteString(v)
}
if v, found := n.Labels["provisioner.cast.ai/node-configuration-version"]; found {
_, _ = hash.WriteString(v)
}
_, _ = hash.WriteString(n.Status.NodeInfo.Architecture)
_, _ = hash.WriteString(n.Status.NodeInfo.ContainerRuntimeVersion)
_, _ = hash.WriteString(n.Status.NodeInfo.KubeletVersion)
_, _ = hash.WriteString(n.Status.NodeInfo.OSImage)
return hash.Sum64()
}