Automatic merge from submit-queue (batch tested with PRs 56650, 55813, 56911, 56921, 56871). If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.
fix deviceplugin test file create leak file problem
When execute make test, this test file will create a file named "kubelet_internal_checkpoint" in k8s directory and not delete it.
This patch fix this error
**What this PR does / why we need it**:
**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Fixes#56365
**Special notes for your reviewer**:
**Release note**:
```release-note
```
Automatic merge from submit-queue (batch tested with PRs 57172, 55382, 56147, 56146, 56158). If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.
Use file store utility for device plugin checkpointing
Partially address issue #54088
cc @sjenning @jeremyeder @jiayingz @vishh
/sig node
- Instead of using cm.capacity field to communicate device plugin resource
capacity, this PR changes to use an explicit cm.GetDevicePluginResourceCapacity()
function that returns device plugin resource capacity as well as any inactive
device plugin resource. Kubelet syncNodeStatus call this function during its
periodic run to update node status capacity and allocatable. After this call,
device plugin can remove the inactive device plugin resource from its allDevices
field as the update is already pushed to API server.
- Extends device plugin checkpoint data to record registered resources
so that we can finish resource removing even upon kubelet restarts.
- Passes sourcesReady from kubelet to device plugin to avoid removing
inactive pods during grace period of kubelet restart.
Automatic merge from submit-queue (batch tested with PRs 55839, 54495, 55884, 55983, 56069). If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.
deviceplugin: fix race when multiple plugins are registered
**What this PR does / why we need it**:
When registering multiple device plugins to Kubelet concurrently, there exists a race that crashes the Kubelet.
Consider two plugins: D1 and D2. The call order method is roughly
D1 -> manager.go:register -> endpoint.go:listAndWatch -> device_plugin_handler.go:(*D1).callback
D2 -> manager.go:register -> endpoint.go:listAndWatch -> device_plugin_handler.go:(*D2).callback
The callback function accesses HandlerImpl's allDevices map that maps (resourceName -> DeviceID). If both plugins reach these accesses at the same time, Kubelet crashes with "fatal error: concurrent map read and map write".
This can be solved by making sure handler is locked when allDevices are being updated. The functionality is needed to avoid Kubelet crashes when multiple device plugins are trying to register with Kubelet at the same moment. Occurs frequently when single binary tries to register itself as multiple plugins.
**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
**Special notes for your reviewer**:
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.
Deviceplugin refactoring: merge func list and listwatch in endpoint into one
**What this PR does / why we need it**:
merge func list and listwatch in endpoint into one, since we won't call list func individually
**Which issue this PR fixes**
fixes#51993
Part2
**Special notes for your reviewer**:
/cc @jiayingz @RenaudWasTaken @vishh
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 55283, 55461, 55288, 53970, 55487). If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.
Add more comments for DevicePluginHandlerImpl struct
**What this PR does / why we need it**:
Add more comments
**Special notes for your reviewer**:
@jiayingz PTAL.
**Release note**:
```
NONE
```
Automatic merge from submit-queue. If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>.
Add admission handler for device resources allocation
**What this PR does / why we need it**:
Add admission handler for device resources allocation to fail fast during pod creation
**Which issue this PR fixes**
fixes#51592
**Special notes for your reviewer**:
@jiayingz Sorry, there is something wrong with my branch in #51895. And I think the existing comments in the PR might be too long for others to view. So I closed it and opened the new one, as we have basically reach an agreement on the implement :)
I have covered the functionality and unit test part here, and would set about the e2e part ASAP
/cc @jiayingz @vishh @RenaudWasTaken
**Release note**:
```release-note
NONE
```
While moving device_plugin_handler_test.go from pkg/kubelet/cm/ to
pkg/kubelet/cm/deviceplugin/, we can no longer uses cm in its tests
because that would cause a cycle dependency. To solve this problem,
I moved the main cm GetResources functionality as well as part of the
current device plugin handler Allocate functionality into a new device
plugin handler function, GetDeviceRunContainerOptions(). This
refactoring is also needed by another PR 51895 that moves device
allocation into admission phase. Now device plugin handler Allocate()
first checks whether there is cached device runtime state and only
issues Allocate grpc call if there is no cached state available.
The new GetDeviceRunContainerOptions() function simply returns device
runtime config from the cached state. To support this change, extended the
podDevices struct and checkpoint data structure with device runtime state.
