github/kubernetes
1
0
Fork 0
mirror of https://github.com/k3s-io/kubernetes.git synced 2025-09-21 01:50:55 +00:00
Code Issues Projects Releases Wiki Activity

kubelet: Handle UID reuse in pod worker

Browse Source 
If a pod is killed (no longer wanted) and then a subsequent create/
add/update event is seen in the pod worker, assume that a pod UID
was reused (as it could be in static pods) and have the next
SyncKnownPods after the pod terminates remove the worker history so
that the config loop can restart the static pod, as well as return
to the caller the fact that this termination was not final.

The housekeeping loop then reconciles the desired state of the Kubelet
(pods in pod manager that are not in a terminal state, i.e. admitted
pods) with the pod worker by resubmitting those pods. This adds a
small amount of latency (2s) when a pod UID is reused and the pod
is terminated and restarted.
This commit is contained in:
Clayton Coleman
2021-09-08 14:38:09 -04:00
parent 47086a6623
commit d5719800bf
4 changed files with 118 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
pkg/kubelet/kubelet.go
View File

@@ -2227,6 +2227,8 @@ func (kl *Kubelet) HandlePodAdditions(pods []*v1.Pod) {
}
mirrorPod, _ := kl.podManager.GetMirrorPodByPod(pod)
kl.dispatchWork(pod, kubetypes.SyncPodCreate, mirrorPod, start)
// TODO: move inside syncPod and make reentrant
// https://github.com/kubernetes/kubernetes/issues/105014
kl.probeManager.AddPod(pod)
}
}
@@ -2261,6 +2263,9 @@ func (kl *Kubelet) HandlePodRemoves(pods []*v1.Pod) {
if err := kl.deletePod(pod); err != nil {
klog.V(2).InfoS("Failed to delete pod", "pod", klog.KObj(pod), "err", err)
}
// TODO: move inside syncTerminatingPod|syncTerminatedPod (we should stop probing
// once the pod kill is acknowledged and during eviction)
// https://github.com/kubernetes/kubernetes/issues/105014
kl.probeManager.RemovePod(pod)
}
}