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>.
Ring buffer for shared informer notifications
**What this PR does / why we need it**:
Improves memory allocation for shared informer listeners. Instead of always appending to the slice use as a ring buffer, avoiding reslice operations as long as there is room in the slice. See https://github.com/kubernetes/kubernetes/pull/47045#issuecomment-317621259 for details. This is a follow up PR for #47045.
Results from BenchmarkListener:
```
Current code (from the #47045):
1000000 1540 ns/op 109 B/op 1 allocs/op
```
```
New code:
1000000 1162 ns/op 16 B/op 1 allocs/op
```
**Special notes for your reviewer**:
Only review the last commit, this branch is based on #47045 PR. I'll rebase onto master once it is merged.
**Release note**:
```release-note
NONE
```
/kind enhancement
/sig api-machinery
/cc @deads2k @ncdc
Automatic merge from submit-queue (batch tested with PRs 53234, 53252, 53267, 53276, 53107). 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 get alpha backend service into cloud provider
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 53234, 53252, 53267, 53276, 53107). 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>.
Prepull images after disk eviction tests
Example failure: https://k8s-gubernator.appspot.com/build/kubernetes-jenkins/logs/ci-kubernetes-node-kubelet-flaky/2855
Disk eviction tests trigger image garbage collection. It can remove images required for subsequent tests.
This results in the error during pod creation:
`timed out waiting for the condition`
You can see in the events after the test:
`I0929 15:47:05.884] I0929 15:17:09.376591 2309 util.go:4734] Event(v1.ObjectReference{Kind:"Pod", Namespace:"e2e-tests-localstorage-eviction-test-mn5v4", Name:"container-disk-hog-pod", UID:"8dba851c-a528-11e7-a9a6-42010a800fd7", APIVersion:"v1", ResourceVersion:"116", FieldPath:"spec.containers{container-disk-hog-container}"}): type: 'Warning' reason: 'ErrImageNeverPull' Container image "busybox" is not present with pull policy of Never`
/assign @Random-Liu
Automatic merge from submit-queue (batch tested with PRs 53234, 53252, 53267, 53276, 53107). 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>.
kubeadm: Let kubelets in the cluster rotate their certificates
**What this PR does / why we need it**:
As of 1.8.0 kubelet in kubeadm configuration ships with enabled feature
of node certificate autorotation, it makes sense to enable automatic
certificate rotation csr signing. It will help to avoid issues like
described in #53231 and #53237.
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes#53237.
**Special notes for your reviewer**:
We need that patch to be applied both to master branch and cherry-pick into 1.8, so kubeadm in 1.8.1 release will create correct role bindings on upgrades between 1.7 -> 1.8 and initialization of new clusters with 1.8.x
cc @luxas @liggitt @pipejakob
**Release note**:
```release-note
kubeadm allows the kubelets in the cluster to automatically renew their client certificates
```
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>.
Fixes a flakiness in GPUDevicePlugin e2e test.
Waits till nvidia gpu disappears from all nodes after deleting the
device plug DaemonSet to make sure its pods are deleted from all nodes.
**What this PR does / why we need it**:
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes #
https://github.com/kubernetes/kubernetes/issues/53281
**Special notes for your reviewer**:
**Release note**:
```release-note
```
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>.
Fix broken links in kubelet after moving proposals to subdirs
**What this PR does / why we need it**:
fix incorrect links in kubelet after kubernetes/community#1010
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes kubernetes/community#918
**Special notes for your reviewer**:
CC @bgrant0607
**Release note**:
```
NONE
```
Automatic merge from submit-queue (batch tested with PRs 53101, 53158, 52165). 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>.
More descriptive error message for `make test`
**Which issue this PR fixes**
Fix#52025
**What this PR does / why we need it**:
The `[packages]` argument to `go test` can be confusing. If the
`package` does not begin with `./`, `go test` considers it relative to
`$GOPATH/src`. If it does begin with `./`, `go test` considers it
relative to `pwd`. `hack/make-rules/test.sh`, and thus `make test`, use
`go test`.
This commit adds a verify step to `hack/make-rules/test.sh`. Before we
run `go test`, we check the packages under test exist. If the user
specified the package path in the incorrect format - for example they
didn't prepend with `./` when they should have, the check logs a
suggestion of the alternative path they should use.
Running the test suites for a package is an activity many all first time
contributors will take. Hopefully including a more descriptive error
message, with a suggestion for a fix if applicable, will make this an
easier experience.
Signed-off-by: mattjmcnaughton <mattjmcnaughton@gmail.com>
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 53101, 53158, 52165). 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>.
Calculate patches for commands using input version
Fixes#53040
the encoder used for encoding these objects while calculating patches does not have sufficient information to select a correct version when the object does not exist in all versions of a target group (like replicasets not existing in apps/v1beta1)
this PR wraps the encoder to first convert to the same version used to read the object (based on the mapping's GroupVersion)
long-term, we should switch UpdatePodSpecForObject to work on versioned objects and v1.PodSpec and avoid conversion altogether
```release-note
Fixes an issue with `kubectl set` commands encountering conversion errors for ReplicaSet and DaemonSet objects
```
Automatic merge from submit-queue (batch tested with PRs 53101, 53158, 52165). 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>.
[OpenStack] Service LoadBalancer defaults to external
**What this PR does / why we need it**:
Let "service.beta.kubernetes.io/openstack-internal-load-balancer" default to false.
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*:
fixes#53078
**Special notes for your reviewer**:
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 53263, 52967, 53262, 52654, 53187). 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 sample-apiserver artifacts
The example yaml file for sample-apiserver has some errors. Fix them.
**Release note**:
```
NONE
```
Automatic merge from submit-queue (batch tested with PRs 53263, 52967, 53262, 52654, 53187). 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 pod disruption budgets to admin/edit/view roles
Closes#50767
cc @kubernetes/sig-auth-pr-reviews @kubernetes/sig-node-pr-reviews
```release-note
RBAC: The default `admin` and `edit` roles now include read/write permissions and the `view` role includes read permissions on `poddisruptionbudget.policy` resources.
```
Automatic merge from submit-queue (batch tested with PRs 53263, 52967, 53262, 52654, 53187). 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 sed command to not try shell redirection
RunCmd uses Go's os/exec library to run commands directly. Since these
are not run through a shell, we can't use shell syntax for piping or
file redirection. The proper way to do that is to create a Command
object and set the Std{in,out,err} pipes appropriately. Luckily sed
can handle the behavior we need without having to manually set this up.
fixes https://github.com/kubernetes/kubeadm/issues/472
Automatic merge from submit-queue (batch tested with PRs 51021, 53225, 53094, 53219). 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>.
Change ImageGCManage to consume ImageFS stats from StatsProvider
Fixes#53083.
**Release note**:
```
Change ImageGCManage to consume ImageFS stats from StatsProvider
```
/assign @Random-Liu
Automatic merge from submit-queue (batch tested with PRs 51021, 53225, 53094, 53219). 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 issue(#49965)kubectl scale also says that it can work based on a label selector or all"
**What this PR does / why we need it**:
Fixes#49965#44800
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes #
**Special notes for your reviewer**:
**Release note**:
```release-note
NONE
```
As of 1.8.0 kubelet in kubeadm configuration ships with enabled feature
of node certificate autorotation, it makes sense to enable automatic
certificate rotation csr signing. It will help to avoid issues like
described in #53231 and #53237.
RunCmd uses Go's os/exec library to run commands directly. Since these
are not run through a shell, we can't use shell syntax for piping for
file redirection. The proper way to do that is to create a Command
object and set the Std{in,out,err} pipes appropriately. Luckily sed
can handle the behavior we need without having to manually set this up.
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>.
In cluster autoscaling tests, improve error logging on enable autoscaling failure
This adds logging command and request output in addition to error.
Automatic merge from submit-queue (batch tested with PRs 51311, 52575, 53169). 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 a scheduler flaky e2e test
**What this PR does / why we need it**:
Makes a scheduler e2e test that verifies the resource limit predicate more robust.
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes#53066
**Release note**:
```release-note
NONE
```
@kubernetes/sig-scheduling-pr-reviews
Automatic merge from submit-queue (batch tested with PRs 51311, 52575, 53169). 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>.
Unable to detach the vSphere volume from Powered off node
With the existing implementation when a vSphere node is powered off, the node is not deleted by the node controller and is in "NotReady" state. Following the approach similar to GCE as mentioned here - https://github.com/kubernetes/kubernetes/issues/46442.
I observe the following issues:
- The pods on the powered off node are not **instantaneously** created on the other available node. Only after 5 minutes timeout, the pods will be created on other available nodes with the volume attached to it. This means an application downtime of around 5 minutes which is not good at all.
- The volume on the powered off node are not detached at all when the pod with the volume is already moved to other available node. Hence any attempt to restart the powered off node will fail as the same volume is attached to other node which is present on this powered off node. (Please note that the volumes are not automatically detached from powered off in vSphere as opposed to GCE, AWS where volume is automatically detached from when node is powered off).
So inorder to resolve this problem, we have decided to back with the approach where the powered off node will be removed by the Node controller. So the above 2 problems will be resolved as follows:
- Since the node is deleted, the pod on the powered off node becomes instantaneously available on other available nodes with the volume attached to the new nodes. Hence there is no application downtime at all.
- After a period of 6 minutes (timeout period), the volumes are automatically detached from the powered off node. Hence any restarts after 6 minutes on the powered off node would work and not cause any problems as volumes are already detached.
For now, we would want to go ahead with deleting the node from node controller when a node is powered off in vCenter until we have a better approach. I think the best possible solution would be to introduce power handler in volume controller to see if the node is powered off before we can take any appropriate for attach/detach operations.
```release-note
None
```
@jingxu97 @saad-ali @divyenpatel @luomiao @rohitjogvmw
Automatic merge from submit-queue (batch tested with PRs 44596, 52708, 53163, 53167, 52692). 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 the bad code comment and make the format unify
**What this PR does / why we need it**:
Fix the bad code comment and make the format unify
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes #
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 44596, 52708, 53163, 53167, 52692). 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>.
Do not GC exited containers in running pods
This fixes a regression introduced by #45896, and was identified by #52462.
This bug causes the kubelet to garbage collect exited containers in a running pod.
This manifests in strange and confusing state when viewing the cluster. For example, it can show running pods as having no init container (see #52462), if that container has exited and been removed.
This PR solves this problem by only removing containers and sandboxes from terminated pods.
The important line change is:
` if cgc.podDeletionProvider.IsPodDeleted(podUID) || evictNonDeletedPods {` --->
`if cgc.podStateProvider.IsPodDeleted(podUID) || (cgc.podStateProvider.IsPodTerminated(podUID) && evictTerminatedPods) {`
cc @MrHohn @yujuhong @kubernetes/sig-node-bugs
```release-note
BugFix: Exited containers are not Garbage Collected by the kubelet while the pod is running
```
Automatic merge from submit-queue (batch tested with PRs 44596, 52708, 53163, 53167, 52692). 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>.
Improve deb and rpm packaging in bazel build
**What this PR does / why we need it**:
- kubernetes-cni package now has proper version (0.5.1)
- Synchronize post-1.8 version of 10-kubeadm.conf file from release
repository.
- Fix dependencies
- Improve descriptions in produced packages
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes #
**Special notes for your reviewer**:
This PR will be safe to cherry-pick to 1.8 branch. After that we will be able to generate from bazel automatically usable packages for both 1.8 and master branch out of bazel builds.
cc @ixdy @mikedanese @luxas
**Release note**:
```release-note
- Improved generation of deb and rpm packages in bazel build
```
