mirror of
https://github.com/k3s-io/kubernetes.git
synced 2025-07-31 23:37:01 +00:00
3eadd1a9ea
A number of race conditions exist when pods are terminated early in
their lifecycle because components in the kubelet need to know "no
running containers" or "containers can't be started from now on" but
were relying on outdated state.
Only the pod worker knows whether containers are being started for
a given pod, which is required to know when a pod is "terminated"
(no running containers, none coming). Move that responsibility and
podKiller function into the pod workers, and have everything that
was killing the pod go into the UpdatePod loop. Split syncPod into
three phases - setup, terminate containers, and cleanup pod - and
have transitions between those methods be visible to other
components. After this change, to kill a pod you tell the pod worker
to UpdatePod({UpdateType: SyncPodKill, Pod: pod}).
Several places in the kubelet were incorrect about whether they
were handling terminating (should stop running, might have
containers) or terminated (no running containers) pods. The pod worker
exposes methods that allow other loops to know when to set up or tear
down resources based on the state of the pod - these methods remove
the possibility of race conditions by ensuring a single component is
responsible for knowing each pod's allowed state and other components
simply delegate to checking whether they are in the window by UID.
Removing containers now no longer blocks final pod deletion in the
API server and are handled as background cleanup. Node shutdown
no longer marks pods as failed as they can be restarted in the
next step.
See https://docs.google.com/document/d/1Pic5TPntdJnYfIpBeZndDelM-AbS4FN9H2GTLFhoJ04/edit# for details
110 lines
3.3 KiB
Go
110 lines
3.3 KiB
Go
/*
|
|
Copyright 2019 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 volumemanager
|
|
|
|
import (
|
|
v1 "k8s.io/api/core/v1"
|
|
"k8s.io/kubernetes/pkg/kubelet/config"
|
|
"k8s.io/kubernetes/pkg/kubelet/container"
|
|
"k8s.io/kubernetes/pkg/volume/util/types"
|
|
)
|
|
|
|
// FakeVolumeManager is a test implementation that just tracks calls
|
|
type FakeVolumeManager struct {
|
|
volumes map[v1.UniqueVolumeName]bool
|
|
reportedInUse map[v1.UniqueVolumeName]bool
|
|
}
|
|
|
|
// NewFakeVolumeManager creates a new VolumeManager test instance
|
|
func NewFakeVolumeManager(initialVolumes []v1.UniqueVolumeName) *FakeVolumeManager {
|
|
volumes := map[v1.UniqueVolumeName]bool{}
|
|
for _, v := range initialVolumes {
|
|
volumes[v] = true
|
|
}
|
|
return &FakeVolumeManager{
|
|
volumes: volumes,
|
|
reportedInUse: map[v1.UniqueVolumeName]bool{},
|
|
}
|
|
}
|
|
|
|
// Run is not implemented
|
|
func (f *FakeVolumeManager) Run(sourcesReady config.SourcesReady, stopCh <-chan struct{}) {
|
|
}
|
|
|
|
// WaitForAttachAndMount is not implemented
|
|
func (f *FakeVolumeManager) WaitForAttachAndMount(pod *v1.Pod) error {
|
|
return nil
|
|
}
|
|
|
|
// WaitForUnmount is not implemented
|
|
func (f *FakeVolumeManager) WaitForUnmount(pod *v1.Pod) error {
|
|
return nil
|
|
}
|
|
|
|
// GetMountedVolumesForPod is not implemented
|
|
func (f *FakeVolumeManager) GetMountedVolumesForPod(podName types.UniquePodName) container.VolumeMap {
|
|
return nil
|
|
}
|
|
|
|
// GetPossiblyMountedVolumesForPod is not implemented
|
|
func (f *FakeVolumeManager) GetPossiblyMountedVolumesForPod(podName types.UniquePodName) container.VolumeMap {
|
|
return nil
|
|
}
|
|
|
|
// GetExtraSupplementalGroupsForPod is not implemented
|
|
func (f *FakeVolumeManager) GetExtraSupplementalGroupsForPod(pod *v1.Pod) []int64 {
|
|
return nil
|
|
}
|
|
|
|
// GetVolumesInUse returns a list of the initial volumes
|
|
func (f *FakeVolumeManager) GetVolumesInUse() []v1.UniqueVolumeName {
|
|
inuse := []v1.UniqueVolumeName{}
|
|
for v := range f.volumes {
|
|
inuse = append(inuse, v)
|
|
}
|
|
return inuse
|
|
}
|
|
|
|
// ReconcilerStatesHasBeenSynced is not implemented
|
|
func (f *FakeVolumeManager) ReconcilerStatesHasBeenSynced() bool {
|
|
return true
|
|
}
|
|
|
|
// VolumeIsAttached is not implemented
|
|
func (f *FakeVolumeManager) VolumeIsAttached(volumeName v1.UniqueVolumeName) bool {
|
|
return false
|
|
}
|
|
|
|
// MarkVolumesAsReportedInUse adds the given volumes to the reportedInUse map
|
|
func (f *FakeVolumeManager) MarkVolumesAsReportedInUse(volumesReportedAsInUse []v1.UniqueVolumeName) {
|
|
for _, reportedVolume := range volumesReportedAsInUse {
|
|
if _, ok := f.volumes[reportedVolume]; ok {
|
|
f.reportedInUse[reportedVolume] = true
|
|
}
|
|
}
|
|
}
|
|
|
|
// GetVolumesReportedInUse is a test function only that returns a list of volumes
|
|
// from the reportedInUse map
|
|
func (f *FakeVolumeManager) GetVolumesReportedInUse() []v1.UniqueVolumeName {
|
|
inuse := []v1.UniqueVolumeName{}
|
|
for reportedVolume := range f.reportedInUse {
|
|
inuse = append(inuse, reportedVolume)
|
|
}
|
|
return inuse
|
|
}
|