/
rte.go
442 lines (371 loc) · 17.4 KB
/
rte.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
/*
Copyright 2020 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.
*/
/*
* resource-topology-exporter specific tests
*/
package rte
import (
"context"
"strings"
"time"
"github.com/onsi/ginkgo/v2"
"github.com/onsi/gomega"
corev1 "k8s.io/api/core/v1"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/apimachinery/pkg/fields"
clientset "k8s.io/client-go/kubernetes"
"k8s.io/klog/v2"
"sigs.k8s.io/controller-runtime/pkg/client"
"github.com/k8stopologyawareschedwg/noderesourcetopology-api/pkg/apis/topology/v1alpha2"
topologyclientset "github.com/k8stopologyawareschedwg/noderesourcetopology-api/pkg/generated/clientset/versioned"
"github.com/k8stopologyawareschedwg/podfingerprint"
"github.com/k8stopologyawareschedwg/resource-topology-exporter/pkg/k8sannotations"
"github.com/k8stopologyawareschedwg/resource-topology-exporter/pkg/nrtupdater"
"github.com/k8stopologyawareschedwg/resource-topology-exporter/test/e2e/utils/fixture"
e2enodes "github.com/k8stopologyawareschedwg/resource-topology-exporter/test/e2e/utils/nodes"
e2enodetopology "github.com/k8stopologyawareschedwg/resource-topology-exporter/test/e2e/utils/nodetopology"
e2epods "github.com/k8stopologyawareschedwg/resource-topology-exporter/test/e2e/utils/pods"
e2ertepod "github.com/k8stopologyawareschedwg/resource-topology-exporter/test/e2e/utils/pods/rtepod"
"github.com/k8stopologyawareschedwg/resource-topology-exporter/test/e2e/utils/remoteexec"
e2econsts "github.com/k8stopologyawareschedwg/resource-topology-exporter/test/e2e/utils/testconsts"
e2etestenv "github.com/k8stopologyawareschedwg/resource-topology-exporter/test/e2e/utils/testenv"
)
const (
updateIntervalExtraSafety = 10 * time.Second
)
var _ = ginkgo.Describe("[RTE][InfraConsuming] Resource topology exporter", func() {
var (
initialized bool
topologyUpdaterNode *corev1.Node
workerNodes []corev1.Node
)
f := fixture.New()
ginkgo.BeforeEach(func() {
var err error
nsCleanup, err := f.CreateNamespace("rte")
gomega.Expect(err).ToNot(gomega.HaveOccurred())
ginkgo.DeferCleanup(nsCleanup)
if !initialized {
workerNodes, err = e2enodes.GetWorkerNodes(f.K8SCli)
gomega.Expect(err).ToNot(gomega.HaveOccurred())
gomega.Expect(workerNodes).ToNot(gomega.BeEmpty())
// pick any worker node. The (implicit, TODO: make explicit) assumption is
// the daemonset runs on CI on all the worker nodes.
var hasLabel bool
topologyUpdaterNode, hasLabel = e2enodes.PickTargetNode(workerNodes)
gomega.Expect(topologyUpdaterNode).ToNot(gomega.BeNil())
if !hasLabel {
// during the e2e tests we expect changes on the node topology.
// but in an environment with multiple worker nodes, we might be looking at the wrong node.
// thus, we assign a unique label to the picked worker node
// and making sure to deploy the pod on it during the test using nodeSelector
err = e2enodes.LabelNode(f.K8SCli, topologyUpdaterNode, map[string]string{e2econsts.TestNodeLabel: ""})
gomega.Expect(err).ToNot(gomega.HaveOccurred())
}
initialized = true
}
})
ginkgo.Context("with cluster configured", func() {
ginkgo.It("[NotificationFile] it should react to pod changes using the smart poller with notification file", func() {
initialNodeTopo := e2enodetopology.GetNodeTopology(f.TopoCli, topologyUpdaterNode.Name)
stopChan := make(chan struct{})
doneChan := make(chan struct{})
started := false
go func() {
defer ginkgo.GinkgoRecover()
<-stopChan
rtePod, err := e2epods.GetPodOnNode(f.K8SCli, topologyUpdaterNode.Name, e2etestenv.GetNamespaceName(), e2etestenv.RTELabelName)
gomega.Expect(err).ToNot(gomega.HaveOccurred())
rteContainerName, err := e2ertepod.FindRTEContainerName(rtePod)
gomega.Expect(err).ToNot(gomega.HaveOccurred())
rteNotifyFilePath, err := e2ertepod.FindNotificationFilePath(rtePod)
gomega.Expect(err).ToNot(gomega.HaveOccurred())
cmd := []string{"/bin/touch", rteNotifyFilePath}
_, _, err = remoteexec.CommandOnPod(f.Ctx, f.K8SCli, rtePod, rteContainerName, cmd...)
gomega.Expect(err).ToNot(gomega.HaveOccurred(), "failed exec command %v on pod %q", cmd, client.ObjectKeyFromObject(rtePod).String())
klog.Infof("notification triggered, exiting")
doneChan <- struct{}{}
}()
ginkgo.By("getting the updated topology")
var err error
var finalNodeTopo *v1alpha2.NodeResourceTopology
gomega.Eventually(func() bool {
if !started {
stopChan <- struct{}{}
started = true
}
finalNodeTopo, err = f.TopoCli.TopologyV1alpha2().NodeResourceTopologies().Get(context.TODO(), topologyUpdaterNode.Name, metav1.GetOptions{})
if err != nil {
klog.Infof("failed to get the node topology resource: %v", err)
return false
}
if finalNodeTopo.ObjectMeta.ResourceVersion == initialNodeTopo.ObjectMeta.ResourceVersion {
klog.Infof("resource %s not yet updated - resource version not bumped", topologyUpdaterNode.Name)
return false
}
klog.Infof("resource %s updated! - resource version bumped (old %v new %v)", topologyUpdaterNode.Name, initialNodeTopo.ObjectMeta.ResourceVersion, finalNodeTopo.ObjectMeta.ResourceVersion)
reason, ok := finalNodeTopo.Annotations[k8sannotations.RTEUpdate]
if !ok {
klog.Infof("resource %s missing annotation!", topologyUpdaterNode.Name)
return false
}
klog.Infof("resource %s reason %v expected %v", topologyUpdaterNode.Name, reason, nrtupdater.RTEUpdateReactive)
return reason == nrtupdater.RTEUpdateReactive
}).WithTimeout(31*time.Second).WithPolling(1*time.Second).Should(gomega.BeTrue(), "didn't get updated node topology info")
ginkgo.By("checking the topology was updated for the right reason")
<-doneChan
gomega.Expect(finalNodeTopo.Annotations).ToNot(gomega.BeNil(), "missing annotations entirely")
reason := finalNodeTopo.Annotations[k8sannotations.RTEUpdate]
gomega.Expect(reason).To(gomega.Equal(nrtupdater.RTEUpdateReactive), "update reason error: expected %q got %q", nrtupdater.RTEUpdateReactive, reason)
})
})
ginkgo.Context("with pod fingerprinting enabled", func() {
ginkgo.It("[PodFingerprint] it should report the computation method in the attributes", func() {
nrt := e2enodetopology.GetNodeTopology(f.TopoCli, topologyUpdaterNode.Name)
klog.Infof("Initial NRT: %q generation=%v resourceVersion=%v", nrt.Name, nrt.Generation, nrt.ResourceVersion)
if _, ok := findAttribute(nrt.Attributes, podfingerprint.Attribute); !ok {
ginkgo.Skip("pod fingerprinting attribute not found - assuming disabled")
}
meth, ok := findAttribute(nrt.Attributes, podfingerprint.AttributeMethod)
gomega.Expect(ok).To(gomega.BeTrue(), "attribute %q missing, but PFP reported", podfingerprint.AttributeMethod)
// note this is a subset of all the available methods declared in the podfingerprint package
validMethods := []string{
podfingerprint.MethodAll,
podfingerprint.MethodWithExclusiveResources,
}
gomega.Expect(validMethods).Should(gomega.ContainElement(meth), "unsupported PFP computation method %q", meth)
})
ginkgo.It("[PodFingerprint] it should report stable value if the pods do not change", func() {
prevNrt := e2enodetopology.GetNodeTopology(f.TopoCli, topologyUpdaterNode.Name)
klog.Infof("Initial NRT: %q generation=%v resourceVersion=%v", prevNrt.Name, prevNrt.Generation, prevNrt.ResourceVersion)
if _, ok := prevNrt.Annotations[podfingerprint.Annotation]; !ok {
ginkgo.Skip("pod fingerprinting annotation not found - assuming disabled")
}
if _, ok := findAttribute(prevNrt.Attributes, podfingerprint.Attribute); !ok {
ginkgo.Skip("pod fingerprinting attribute not found - assuming disabled")
}
dumpPods(f.K8SCli, topologyUpdaterNode.Name, "reference pods")
updateInterval, method, err := estimateUpdateInterval(*prevNrt)
gomega.Expect(err).ToNot(gomega.HaveOccurred())
klog.Infof("%s update interval: %s", method, updateInterval)
// 3 timess is "long enough" - decided after quick tuning and try/error
// if the object does not change, neither resourceVersion will. So we can just sleep.
maxSteps := 3
for step := 0; step < maxSteps; step++ {
klog.Infof("waiting for %s: %d/%d", updateInterval, step+1, maxSteps)
time.Sleep(updateInterval)
}
currNrt := e2enodetopology.GetNodeTopology(f.TopoCli, topologyUpdaterNode.Name)
klog.Infof("Control NRT: %q generation=%v resourceVersion=%v", prevNrt.Name, prevNrt.Generation, prevNrt.ResourceVersion)
// note we don't test no pods have been added/deleted. This is because the suite is supposed to own the cluster while it runs
// IOW, if we don't create/delete pods explicitely, noone else is supposed to do
pfpStable := expectPodFingerprint(*prevNrt, "==", *currNrt)
if !pfpStable {
dumpPods(f.K8SCli, topologyUpdaterNode.Name, "after PFP mismatch")
// ignore errors and carry on. We don't want to fail the test because of missing debug info.
dumpRTELogs(f.K8SCli, topologyUpdaterNode.Name)
}
gomega.Expect(pfpStable).To(gomega.BeTrue(), "PFP changed unexpectedly")
})
ginkgo.It("[release][PodFingerprint] it should report updated value if the set of running pods changes", func() {
nodes, err := e2enodes.FilterNodesWithEnoughCores(workerNodes, "1000m")
gomega.Expect(err).ToNot(gomega.HaveOccurred())
if len(nodes) < 1 {
ginkgo.Skip("not enough allocatable cores for this test")
}
var currNrt *v1alpha2.NodeResourceTopology
prevNrt := e2enodetopology.GetNodeTopology(f.TopoCli, topologyUpdaterNode.Name)
if _, ok := prevNrt.Annotations[podfingerprint.Annotation]; !ok {
ginkgo.Skip("pod fingerprinting not found - assuming disabled")
}
if _, ok := findAttribute(prevNrt.Attributes, podfingerprint.Attribute); !ok {
ginkgo.Skip("pod fingerprinting attribute not found - assuming disabled")
}
dumpPods(f.K8SCli, topologyUpdaterNode.Name, "reference pods")
updateInterval, method, err := estimateUpdateInterval(*prevNrt)
gomega.Expect(err).ToNot(gomega.HaveOccurred())
klog.Infof("%s update interval: %s", method, updateInterval)
sleeperPod := e2epods.MakeGuaranteedSleeperPod("1000m")
pod, err := e2epods.CreateSync(f, sleeperPod)
gomega.Expect(err).ToNot(gomega.HaveOccurred())
// (try to) delete the pod twice is no bother
podNamespace, podName := pod.Namespace, pod.Name
ginkgo.DeferCleanup(e2epods.DeletePodSyncByName, f, podNamespace, podName)
currNrt = getUpdatedNRT(f.TopoCli, topologyUpdaterNode.Name, *prevNrt, updateInterval)
pfpChanged := expectPodFingerprint(*prevNrt, "!=", *currNrt)
errMessage := "PFP did not change after pod creation"
if !pfpChanged {
dumpPods(f.K8SCli, topologyUpdaterNode.Name, errMessage)
}
gomega.Expect(pfpChanged).To(gomega.BeTrue(), errMessage)
// since we need to delete the pod anyway, let's use this to run another check
prevNrt = currNrt
err = e2epods.DeletePodSyncByName(f, podNamespace, podName)
gomega.Expect(err).ToNot(gomega.HaveOccurred())
currNrt = getUpdatedNRT(f.TopoCli, topologyUpdaterNode.Name, *prevNrt, updateInterval)
pfpChanged = expectPodFingerprint(*prevNrt, "!=", *currNrt)
errMessage = "PFP did not change after pod deletion"
if !pfpChanged {
dumpPods(f.K8SCli, topologyUpdaterNode.Name, errMessage)
}
gomega.Expect(pfpChanged).To(gomega.BeTrue(), errMessage)
})
})
ginkgo.Context("with refresh-node-resources enabled", func() {
ginkgo.It("[NodeRefresh] should be able to detect devices", func() {
gomega.Eventually(func() bool {
nrt := e2enodetopology.GetNodeTopology(f.TopoCli, topologyUpdaterNode.Name)
devName := e2etestenv.GetDeviceName()
for _, zone := range nrt.Zones {
for _, res := range zone.Resources {
if res.Name == devName {
return true
}
}
}
return false
}).WithTimeout(30*time.Second).WithPolling(10*time.Second).Should(gomega.BeTrue(), "device: %q was not found in NRT: %q", e2etestenv.GetDeviceName(), topologyUpdaterNode.Name)
})
ginkgo.It("[NodeRefresh] should log the refresh message", func() {
rtePod, err := e2epods.GetPodOnNode(f.K8SCli, topologyUpdaterNode.Name, e2etestenv.GetNamespaceName(), e2etestenv.RTELabelName)
gomega.Expect(err).ToNot(gomega.HaveOccurred())
rteContainerName, err := e2ertepod.FindRTEContainerName(rtePod)
gomega.Expect(err).ToNot(gomega.HaveOccurred())
gomega.Eventually(func() bool {
logs, err := e2epods.GetLogsForPod(f.K8SCli, rtePod.Namespace, rtePod.Name, rteContainerName)
gomega.Expect(err).ToNot(gomega.HaveOccurred())
return strings.Contains(logs, "tracking node resources") || strings.Contains(logs, "update node resources")
}).WithTimeout(42*time.Second).WithPolling(5*time.Second).Should(gomega.BeTrue(), "container: %q in pod: %q doesn't contains the refresh log message", rteContainerName, rtePod.Name)
})
})
})
func getUpdatedNRT(topologyClient *topologyclientset.Clientset, nodeName string, prevNrt v1alpha2.NodeResourceTopology, timeout time.Duration) *v1alpha2.NodeResourceTopology {
var err error
var currNrt *v1alpha2.NodeResourceTopology
gomega.EventuallyWithOffset(1, func() bool {
currNrt, err = topologyClient.TopologyV1alpha2().NodeResourceTopologies().Get(context.TODO(), nodeName, metav1.GetOptions{})
if err != nil {
klog.Infof("failed to get the node topology resource: %v", err)
return false
}
if currNrt.ObjectMeta.ResourceVersion == prevNrt.ObjectMeta.ResourceVersion {
klog.Infof("resource %s not yet updated - resource version not bumped", nodeName)
return false
}
return true
}).WithTimeout(timeout+updateIntervalExtraSafety).WithPolling(1*time.Second).Should(gomega.BeTrue(), "didn't get updated node topology info")
return currNrt
}
func dumpPods(cs clientset.Interface, nodeName, message string) {
nodeSelector := fields.Set{
"spec.nodeName": nodeName,
}.AsSelector().String()
pods, err := cs.CoreV1().Pods(e2etestenv.GetNamespaceName()).List(context.TODO(), metav1.ListOptions{FieldSelector: nodeSelector})
gomega.Expect(err).ToNot(gomega.HaveOccurred())
klog.Infof("BEGIN pods running on %q: %s", nodeName, message)
for _, pod := range pods.Items {
klog.Infof("%s %s/%s status=%s (%s %s)", nodeName, pod.Namespace, pod.Name, pod.Status.Phase, pod.Status.Message, pod.Status.Reason)
}
klog.Infof("END pods running on %q: %s", nodeName, message)
}
func expectPodFingerprint(nrt1 v1alpha2.NodeResourceTopology, mode string, nrt2 v1alpha2.NodeResourceTopology) bool {
pfp1, ok1 := extractPFP(nrt1)
if !ok1 {
return false
}
pfp2, ok2 := extractPFP(nrt2)
if !ok2 {
return false
}
switch mode {
case "==":
return expectEqualPFPs(nrt1.Name, pfp1, nrt2.Name, pfp2)
case "!=":
return expectDifferentPFPs(nrt1.Name, pfp1, nrt2.Name, pfp2)
default:
klog.Infof("unsupported comparison mode %q", mode)
return false
}
}
func extractPFP(nrt v1alpha2.NodeResourceTopology) (string, bool) {
pfpAnn, okAnn := nrt.Annotations[podfingerprint.Annotation]
if !okAnn {
klog.Infof("cannot find pod fingerprint annotation in NRT %q", nrt.Name)
return "", false
}
pfpAttr, okAttr := findAttribute(nrt.Attributes, podfingerprint.Attribute)
if !okAttr {
klog.Infof("cannot find pod fingerprint attribute in NRT %q", nrt.Name)
return "", false
}
if pfpAnn != pfpAttr {
klog.Infof("PFP mismatch in %q annotation=%q attribute=%q", nrt.Name, pfpAnn, pfpAttr)
return "", false
}
return pfpAttr, true
}
func expectEqualPFPs(name1, pfp1, name2, pfp2 string) bool {
if pfp1 != pfp2 {
klog.Infof("fingerprint mismatch NRT %q PFP %q vs NRT %q PFP %q", name1, pfp1, name2, pfp2)
return false
}
return true
}
func expectDifferentPFPs(name1, pfp1, name2, pfp2 string) bool {
if pfp1 == pfp2 {
klog.Infof("fingerprint equality NRT %q PFP %q vs NRT %q PFP %q", name2, pfp1, name2, pfp2)
return false
}
return true
}
func estimateUpdateInterval(nrt v1alpha2.NodeResourceTopology) (time.Duration, string, error) {
fallbackInterval, err := time.ParseDuration(e2etestenv.GetPollInterval())
if err != nil {
return fallbackInterval, "estimated", err
}
klog.Infof("Annotations for %q: %#v", nrt.Name, nrt.Annotations)
updateIntervalAnn, ok := nrt.Annotations[k8sannotations.UpdateInterval]
if !ok {
// no annotation, we need to guess
return fallbackInterval, "estimated", nil
}
updateInterval, err := time.ParseDuration(updateIntervalAnn)
if err != nil {
return fallbackInterval, "estimated", err
}
return updateInterval, "computed", nil
}
func dumpRTELogs(cs clientset.Interface, nodeName string) error {
rtePod, err := e2epods.GetPodOnNode(cs, nodeName, e2etestenv.GetNamespaceName(), e2etestenv.RTELabelName)
if err != nil {
return err
}
rteContainerName, err := e2ertepod.FindRTEContainerName(rtePod)
if err != nil {
return err
}
logs, err := e2epods.GetLogsForPod(cs, rtePod.Namespace, rtePod.Name, rteContainerName)
if err != nil {
return err
}
klog.Infof("RTE logs:\n%s", logs)
return nil
}
func findAttribute(attrs v1alpha2.AttributeList, name string) (string, bool) {
for _, attr := range attrs {
if attr.Name == name {
return attr.Value, true
}
}
return "", false
}