forked from kubernetes/kubernetes
-
Notifications
You must be signed in to change notification settings - Fork 0
/
worker.go
247 lines (205 loc) · 7.27 KB
/
worker.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
/*
Copyright 2015 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.
*/
package prober
import (
"math/rand"
"time"
"github.com/prometheus/client_golang/prometheus"
"k8s.io/api/core/v1"
"k8s.io/apimachinery/pkg/util/runtime"
"k8s.io/klog"
podutil "k8s.io/kubernetes/pkg/api/v1/pod"
kubecontainer "k8s.io/kubernetes/pkg/kubelet/container"
"k8s.io/kubernetes/pkg/kubelet/prober/results"
"k8s.io/kubernetes/pkg/kubelet/util/format"
)
// worker handles the periodic probing of its assigned container. Each worker has a go-routine
// associated with it which runs the probe loop until the container permanently terminates, or the
// stop channel is closed. The worker uses the probe Manager's statusManager to get up-to-date
// container IDs.
type worker struct {
// Channel for stopping the probe.
stopCh chan struct{}
// The pod containing this probe (read-only)
pod *v1.Pod
// The container to probe (read-only)
container v1.Container
// Describes the probe configuration (read-only)
spec *v1.Probe
// The type of the worker.
probeType probeType
// The probe value during the initial delay.
initialValue results.Result
// Where to store this workers results.
resultsManager results.Manager
probeManager *manager
// The last known container ID for this worker.
containerID kubecontainer.ContainerID
// The last probe result for this worker.
lastResult results.Result
// How many times in a row the probe has returned the same result.
resultRun int
// If set, skip probing.
onHold bool
// proberResultsMetricLabels holds the labels attached to this worker
// for the ProberResults metric.
proberResultsMetricLabels prometheus.Labels
}
// Creates and starts a new probe worker.
func newWorker(
m *manager,
probeType probeType,
pod *v1.Pod,
container v1.Container) *worker {
w := &worker{
stopCh: make(chan struct{}, 1), // Buffer so stop() can be non-blocking.
pod: pod,
container: container,
probeType: probeType,
probeManager: m,
}
switch probeType {
case readiness:
w.spec = container.ReadinessProbe
w.resultsManager = m.readinessManager
w.initialValue = results.Failure
case liveness:
w.spec = container.LivenessProbe
w.resultsManager = m.livenessManager
w.initialValue = results.Success
}
w.proberResultsMetricLabels = prometheus.Labels{
"probe_type": w.probeType.String(),
"container_name": w.container.Name,
"pod_name": w.pod.Name,
"namespace": w.pod.Namespace,
"pod_uid": string(w.pod.UID),
}
return w
}
// run periodically probes the container.
func (w *worker) run() {
probeTickerPeriod := time.Duration(w.spec.PeriodSeconds) * time.Second
// If kubelet restarted the probes could be started in rapid succession.
// Let the worker wait for a random portion of tickerPeriod before probing.
time.Sleep(time.Duration(rand.Float64() * float64(probeTickerPeriod)))
probeTicker := time.NewTicker(probeTickerPeriod)
defer func() {
// Clean up.
probeTicker.Stop()
if !w.containerID.IsEmpty() {
w.resultsManager.Remove(w.containerID)
}
w.probeManager.removeWorker(w.pod.UID, w.container.Name, w.probeType)
ProberResults.Delete(w.proberResultsMetricLabels)
}()
probeLoop:
for w.doProbe() {
// Wait for next probe tick.
select {
case <-w.stopCh:
break probeLoop
case <-probeTicker.C:
// continue
}
}
}
// stop stops the probe worker. The worker handles cleanup and removes itself from its manager.
// It is safe to call stop multiple times.
func (w *worker) stop() {
select {
case w.stopCh <- struct{}{}:
default: // Non-blocking.
}
}
// doProbe probes the container once and records the result.
// Returns whether the worker should continue.
func (w *worker) doProbe() (keepGoing bool) {
defer func() { recover() }() // Actually eat panics (HandleCrash takes care of logging)
defer runtime.HandleCrash(func(_ interface{}) { keepGoing = true })
status, ok := w.probeManager.statusManager.GetPodStatus(w.pod.UID)
if !ok {
// Either the pod has not been created yet, or it was already deleted.
klog.V(3).Infof("No status for pod: %v", format.Pod(w.pod))
return true
}
// Worker should terminate if pod is terminated.
if status.Phase == v1.PodFailed || status.Phase == v1.PodSucceeded {
klog.V(3).Infof("Pod %v %v, exiting probe worker",
format.Pod(w.pod), status.Phase)
return false
}
c, ok := podutil.GetContainerStatus(status.ContainerStatuses, w.container.Name)
if !ok || len(c.ContainerID) == 0 {
// Either the container has not been created yet, or it was deleted.
klog.V(3).Infof("Probe target container not found: %v - %v",
format.Pod(w.pod), w.container.Name)
return true // Wait for more information.
}
if w.containerID.String() != c.ContainerID {
if !w.containerID.IsEmpty() {
w.resultsManager.Remove(w.containerID)
}
w.containerID = kubecontainer.ParseContainerID(c.ContainerID)
w.resultsManager.Set(w.containerID, w.initialValue, w.pod)
// We've got a new container; resume probing.
w.onHold = false
}
if w.onHold {
// Worker is on hold until there is a new container.
return true
}
if c.State.Running == nil {
klog.V(3).Infof("Non-running container probed: %v - %v",
format.Pod(w.pod), w.container.Name)
if !w.containerID.IsEmpty() {
w.resultsManager.Set(w.containerID, results.Failure, w.pod)
}
// Abort if the container will not be restarted.
return c.State.Terminated == nil ||
w.pod.Spec.RestartPolicy != v1.RestartPolicyNever
}
if int32(time.Since(c.State.Running.StartedAt.Time).Seconds()) < w.spec.InitialDelaySeconds {
return true
}
// TODO: in order for exec probes to correctly handle downward API env, we must be able to reconstruct
// the full container environment here, OR we must make a call to the CRI in order to get those environment
// values from the running container.
result, err := w.probeManager.prober.probe(w.probeType, w.pod, status, w.container, w.containerID)
if err != nil {
// Prober error, throw away the result.
return true
}
if w.lastResult == result {
w.resultRun++
} else {
w.lastResult = result
w.resultRun = 1
}
if (result == results.Failure && w.resultRun < int(w.spec.FailureThreshold)) ||
(result == results.Success && w.resultRun < int(w.spec.SuccessThreshold)) {
// Success or failure is below threshold - leave the probe state unchanged.
return true
}
w.resultsManager.Set(w.containerID, result, w.pod)
ProberResults.With(w.proberResultsMetricLabels).Set(result.ToPrometheusType())
if w.probeType == liveness && result == results.Failure {
// The container fails a liveness check, it will need to be restarted.
// Stop probing until we see a new container ID. This is to reduce the
// chance of hitting #21751, where running `docker exec` when a
// container is being stopped may lead to corrupted container state.
w.onHold = true
w.resultRun = 0
}
return true
}