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were relying on outdated state.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
It is possible to have multiple signals that point to the same type of
resource, e.g., both SignalNodeFsAvailable and
SignalAllocatableNodeFsAvailable refer to the same resource NodeFs.
Change the map from map[v1.ResourceName]evictionapi.Signal to
map[v1.ResourceName][]evictionapi.Signal
Automatic merge from submit-queue (batch tested with PRs 48976, 49474, 40050, 49426, 49430)
Remove duplicated import and wrong alias name of api package
**What this PR does / why we need it**:
**Which issue this PR fixes**: fixes#48975
**Special notes for your reviewer**:
/assign @caesarxuchao
**Release note**:
```release-note
NONE
```
This PR fixes the following issues:
1. Use ResourceStorageScratch instead of ResourceStorage API to represent
local storage capacity
2. In eviction manager, use container manager instead of node provider
(kubelet) to retrieve the node capacity and reserved resources. Node
provider (kubelet) has a feature gate so that storagescratch information
may not be exposed if feature gate is not set. On the other hand,
container manager has all the capacity and allocatable resource
information.
Automatic merge from submit-queue
Cleanup kubelet eviction manager tests
It cleans up kubelet eviction manager tests
Extracted parts of tests that were similar to each other to functions
