kubernetes/pkg/kubelet/runonce.go
Clayton Coleman 6b9a381185
kubelet: Force deleted pods can fail to move out of terminating
If a CRI error occurs during the terminating phase after a pod is
force deleted (API or static) then the housekeeping loop will not
deliver updates to the pod worker which prevents the pod's state
machine from progressing. The pod will remain in the terminating
phase but no further attempts to terminate or cleanup will occur
until the kubelet is restarted.

The pod worker now maintains a store of the pods state that it is
attempting to reconcile and uses that to resync unknown pods when
SyncKnownPods() is invoked, so that failures in sync methods for
unknown pods no longer hang forever.

The pod worker's store tracks desired updates and the last update
applied on podSyncStatuses. Each goroutine now synchronizes to
acquire the next work item, context, and whether the pod can start.
This synchronization moves the pending update to the stored last
update, which will ensure third parties accessing pod worker state
don't see updates before the pod worker begins synchronizing them.

As a consequence, the update channel becomes a simple notifier
(struct{}) so that SyncKnownPods can coordinate with the pod worker
to create a synthetic pending update for unknown pods (i.e. no one
besides the pod worker has data about those pods). Otherwise the
pending update info would be hidden inside the channel.

In order to properly track pending updates, we have to be very
careful not to mix RunningPods (which are calculated from the
container runtime and are missing all spec info) and config-
sourced pods. Update the pod worker to avoid using ToAPIPod()
and instead require the pod worker to directly use
update.Options.Pod or update.Options.RunningPod for the
correct methods. Add a new SyncTerminatingRuntimePod to prevent
accidental invocations of runtime only pod data.

Finally, fix SyncKnownPods to replay the last valid update for
undesired pods which drives the pod state machine towards
termination, and alter HandlePodCleanups to:

- terminate runtime pods that aren't known to the pod worker
- launch admitted pods that aren't known to the pod worker

Any started pods receive a replay until they reach the finished
state, and then are removed from the pod worker. When a desired
pod is detected as not being in the worker, the usual cause is
that the pod was deleted and recreated with the same UID (almost
always a static pod since API UID reuse is statistically
unlikely). This simplifies the previous restartable pod support.
We are careful to filter for active pods (those not already
terminal or those which have been previously rejected by
admission). We also force a refresh of the runtime cache to
ensure we don't see an older version of the state.

Future changes will allow other components that need to view the
pod worker's actual state (not the desired state the podManager
represents) to retrieve that info from the pod worker.

Several bugs in pod lifecycle have been undetectable at runtime
because the kubelet does not clearly describe the number of pods
in use. To better report, add the following metrics:

  kubelet_desired_pods: Pods the pod manager sees
  kubelet_active_pods: "Admitted" pods that gate new pods
  kubelet_mirror_pods: Mirror pods the kubelet is tracking
  kubelet_working_pods: Breakdown of pods from the last sync in
    each phase, orphaned state, and static or not
  kubelet_restarted_pods_total: A counter for pods that saw a
    CREATE before the previous pod with the same UID was finished
  kubelet_orphaned_runtime_pods_total: A counter for pods detected
    at runtime that were not known to the kubelet. Will be
    populated at Kubelet startup and should never be incremented
    after.

Add a metric check to our e2e tests that verifies the values are
captured correctly during a serial test, and then verify them in
detail in unit tests.

Adds 23 series to the kubelet /metrics endpoint.
2023-03-08 22:03:51 -06:00

177 lines
5.9 KiB
Go

/*
Copyright 2014 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 kubelet
import (
"context"
"fmt"
"os"
"time"
v1 "k8s.io/api/core/v1"
"k8s.io/klog/v2"
kubecontainer "k8s.io/kubernetes/pkg/kubelet/container"
kubetypes "k8s.io/kubernetes/pkg/kubelet/types"
"k8s.io/kubernetes/pkg/kubelet/util/format"
)
const (
runOnceManifestDelay = 1 * time.Second
runOnceMaxRetries = 10
runOnceRetryDelay = 1 * time.Second
runOnceRetryDelayBackoff = 2
)
// RunPodResult defines the running results of a Pod.
type RunPodResult struct {
Pod *v1.Pod
Err error
}
// RunOnce polls from one configuration update and run the associated pods.
func (kl *Kubelet) RunOnce(updates <-chan kubetypes.PodUpdate) ([]RunPodResult, error) {
ctx := context.Background()
// Setup filesystem directories.
if err := kl.setupDataDirs(); err != nil {
return nil, err
}
// If the container logs directory does not exist, create it.
if _, err := os.Stat(ContainerLogsDir); err != nil {
if err := kl.os.MkdirAll(ContainerLogsDir, 0755); err != nil {
klog.ErrorS(err, "Failed to create directory", "path", ContainerLogsDir)
}
}
select {
case u := <-updates:
klog.InfoS("Processing manifest with pods", "numPods", len(u.Pods))
result, err := kl.runOnce(ctx, u.Pods, runOnceRetryDelay)
klog.InfoS("Finished processing pods", "numPods", len(u.Pods))
return result, err
case <-time.After(runOnceManifestDelay):
return nil, fmt.Errorf("no pod manifest update after %v", runOnceManifestDelay)
}
}
// runOnce runs a given set of pods and returns their status.
func (kl *Kubelet) runOnce(ctx context.Context, pods []*v1.Pod, retryDelay time.Duration) (results []RunPodResult, err error) {
ch := make(chan RunPodResult)
admitted := []*v1.Pod{}
for _, pod := range pods {
// Check if we can admit the pod.
if ok, reason, message := kl.canAdmitPod(admitted, pod); !ok {
kl.rejectPod(pod, reason, message)
results = append(results, RunPodResult{pod, nil})
continue
}
admitted = append(admitted, pod)
go func(pod *v1.Pod) {
err := kl.runPod(ctx, pod, retryDelay)
ch <- RunPodResult{pod, err}
}(pod)
}
klog.InfoS("Waiting for pods", "numPods", len(admitted))
failedPods := []string{}
for i := 0; i < len(admitted); i++ {
res := <-ch
results = append(results, res)
if res.Err != nil {
failedContainerName, err := kl.getFailedContainers(ctx, res.Pod)
if err != nil {
klog.InfoS("Unable to get failed containers' names for pod", "pod", klog.KObj(res.Pod), "err", err)
} else {
klog.InfoS("Unable to start pod because container failed", "pod", klog.KObj(res.Pod), "containerName", failedContainerName)
}
failedPods = append(failedPods, format.Pod(res.Pod))
} else {
klog.InfoS("Started pod", "pod", klog.KObj(res.Pod))
}
}
if len(failedPods) > 0 {
return results, fmt.Errorf("error running pods: %v", failedPods)
}
klog.InfoS("Pods started", "numPods", len(pods))
return results, err
}
// runPod runs a single pod and wait until all containers are running.
func (kl *Kubelet) runPod(ctx context.Context, pod *v1.Pod, retryDelay time.Duration) error {
var isTerminal bool
delay := retryDelay
retry := 0
for !isTerminal {
status, err := kl.containerRuntime.GetPodStatus(ctx, pod.UID, pod.Name, pod.Namespace)
if err != nil {
return fmt.Errorf("unable to get status for pod %q: %v", format.Pod(pod), err)
}
if kl.isPodRunning(pod, status) {
klog.InfoS("Pod's containers running", "pod", klog.KObj(pod))
return nil
}
klog.InfoS("Pod's containers not running: syncing", "pod", klog.KObj(pod))
klog.InfoS("Creating a mirror pod for static pod", "pod", klog.KObj(pod))
if err := kl.podManager.CreateMirrorPod(pod); err != nil {
klog.ErrorS(err, "Failed creating a mirror pod", "pod", klog.KObj(pod))
}
mirrorPod, _ := kl.podManager.GetMirrorPodByPod(pod)
if isTerminal, err = kl.SyncPod(ctx, kubetypes.SyncPodUpdate, pod, mirrorPod, status); err != nil {
return fmt.Errorf("error syncing pod %q: %v", format.Pod(pod), err)
}
if retry >= runOnceMaxRetries {
return fmt.Errorf("timeout error: pod %q containers not running after %d retries", format.Pod(pod), runOnceMaxRetries)
}
// TODO(proppy): health checking would be better than waiting + checking the state at the next iteration.
klog.InfoS("Pod's containers synced, waiting", "pod", klog.KObj(pod), "duration", delay)
time.Sleep(delay)
retry++
delay *= runOnceRetryDelayBackoff
}
return nil
}
// isPodRunning returns true if all containers of a manifest are running.
func (kl *Kubelet) isPodRunning(pod *v1.Pod, status *kubecontainer.PodStatus) bool {
for _, c := range pod.Spec.Containers {
cs := status.FindContainerStatusByName(c.Name)
if cs == nil || cs.State != kubecontainer.ContainerStateRunning {
klog.InfoS("Container not running", "pod", klog.KObj(pod), "containerName", c.Name)
return false
}
}
return true
}
// getFailedContainer returns failed container name for pod.
func (kl *Kubelet) getFailedContainers(ctx context.Context, pod *v1.Pod) ([]string, error) {
status, err := kl.containerRuntime.GetPodStatus(ctx, pod.UID, pod.Name, pod.Namespace)
if err != nil {
return nil, fmt.Errorf("unable to get status for pod %q: %v", format.Pod(pod), err)
}
var containerNames []string
for _, cs := range status.ContainerStatuses {
if cs.State != kubecontainer.ContainerStateRunning && cs.ExitCode != 0 {
containerNames = append(containerNames, cs.Name)
}
}
return containerNames, nil
}