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>.
apiextensions: normalize CR validation to return multiple errors
Deduplicated and use of the `allErrs` pattern we have elsewhere.
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>.
move cached_discovery to client-go/discovery
**Release note**:
```release-note
NONE
```
Moves the cmd/util CachedDiscoveryClient to client-go
cc @soltysh @deads2k
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>.
scheduler: remove nested retry loops and fix serving metrics under healthz server port
This PR fixes the issues of #63404:
- The Serve func for metrics and healthz does not block. Inside, there is a retry loop. This PR fixes this and gets rid of the error messages every 5 seconds.
- The separate metrics server is only to be started if it is configured on another port. Before this PR we were wrongly checking for the healthz service to be activated and then launched the metrics server instead. Because both server startups are in a go routine, they were racing against each other. If you were unlucky, the metrics endpoint was winning and it returned 404 on /healthz (while it should not have been active in the first place). The kubemark tests run the scheduler with a liveness probe which failed, restarting the scheduler after some minutes.
Automatic merge from submit-queue (batch tested with PRs 63792, 63495, 63742, 63332, 63779). 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>.
Track a metrics of number of registered watchers in apiserver
Automatic merge from submit-queue (batch tested with PRs 63792, 63495, 63742, 63332, 63779). 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 timeout for exec interface
This should get us away from situations like https://github.com/kubernetes/kubernetes/issues/63331.
A little bit more context, the `os/exec` package starts to accept `context.Context` in golang 1.7. We should leverage that so we can have a more predictable behavior, then.
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 63792, 63495, 63742, 63332, 63779). 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>.
Actually support service `publishNotReadyAddresses`
This was added and the annotation was deprecated, but it was never
implemented.
xref #63741
**Release note**:
```release-note
The annotation `service.alpha.kubernetes.io/tolerate-unready-endpoints` is deprecated. Users should use Service.spec.publishNotReadyAddresses instead.
```
Automatic merge from submit-queue (batch tested with PRs 63272, 63782, 63715, 63811, 63803). 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>.
switch to versioned objects only for set
Only used versioned types in the `kubectl set` commands. This has long been the goal and we finally make it so here.
@liggitt @soltysh as promised in earlier pulls when I asked you to go with it while I unwound the debt. This frees the entire package.
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 63272, 63782, 63715, 63811, 63803). 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>.
Refactor kubeadm config list-images
In preparation for creating a `kubeadm config images pull`
this commit refactors `kubeadm config list-images` into
`kubeadm config images list`.
Signed-off-by: Chuck Ha <ha.chuck@gmail.com>
**What this PR does / why we need it**:
Prepares the `kubeadm config images` subcommand so we can add more functionality to 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)*:
Related to kubernetes/kubeadm#812
**Special notes for your reviewer**:
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 63272, 63782, 63715, 63811, 63803). 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 support for GCP charm
**What this PR does / why we need it**: Continuing from #62354, add support for Juju charms to connect to GCP integration charm to automatically manage IAM labels and roles, and cloud-provider setting for GCE.
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 63272, 63782, 63715, 63811, 63803). 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>.
Stop installing kubeadm types in the generic, legacy scheme
**What this PR does / why we need it**:
Right now the kubeadm API types are (wrongly from a correctness perspective) installed in the "catch-all" and now legacy `pkg/api/legacyscheme`. Instead, we should use our own, kubeadm-specific `scheme` where our API types are registered. Doing this also cuts one dependency on the internal core API types.
**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Part of https://github.com/kubernetes/community/pull/2131
**Special notes for your reviewer**:
**Release note**:
```release-note
NONE
```
@kubernetes/sig-cluster-lifecycle-pr-reviews @sttts @liztio
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 mount unmount failure for a Windows pod
**What this PR does / why we need it**:
`IsLikelyNotMountPoint` func does not return correctly, for invalid symlink, it should return true(not a mount point), now it will always return false:
7711d88661/pkg/util/mount/mount_windows.go (L141-L148)7711d88661/pkg/volume/util/util.go (L147-L163)
**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#63249
**Special notes for your reviewer**:
**Release note**:
```
fix mount unmount failure for a Windows pod
```
/sig windows
/assign @msau42
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>.
avoid duplicate status in audit events
Fixes: https://github.com/kubernetes/kubernetes/issues/60108
**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 #
**Special notes for your reviewer**:
/assign @sttts @tallclair
**Release note**:
```
Action required: When Response is a metav1.Status, it is no longer copied into the audit.Event status. Only the "status", "reason" and "code" fields are set.
```
Automatic merge from submit-queue (batch tested with PRs 63603, 63557, 62015). 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>.
CRI: update documents for container logpath
**What this PR does / why we need it**:
The container log path has been changed from `containername_attempt#.log` to `containername/attempt#.log` in #59906. This PR updates CRI documents for 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 #
**Special notes for your reviewer**:
**Release note**:
```release-note
CRI: update documents for container logpath. The container log path has been changed from containername_attempt#.log to containername/attempt#.log
```
Automatic merge from submit-queue (batch tested with PRs 63603, 63557, 62015). 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>.
apiserver: Fail if dry-run query param is specified
Adds a dry-run query parameter now that does nothing but reject the request. The sooner we have this check in master, the safer it will be for clients to send dry-run requests that are not going to be applied nonetheless.
```release-note
Create a new `dryRun` query parameter for mutating endpoints. If the parameter is set, then the query will be rejected, as the feature is not implemented yet. This will allow forward compatibility with future clients; otherwise, future clients talking with older apiservers might end up modifying a resource even if they include the `dryRun` query parameter.
```
Automatic merge from submit-queue (batch tested with PRs 63603, 63557, 62015). 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>.
Clean up equiv cache with a simple implementation instead of LRU
**What this PR does / why we need it**:
The original version of equiv cache use pod hash as cache key, also, the predicate order is not fixed. So I used a LRU cache to improve hit rate.
While now we've already refactored it to use predicates as keys, and its order was also fixed in scheduler, we can use a simplest cache instead now.
**Special notes for your reviewer**:
The question is brought up by @misterikkit
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 57536, 63812). 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: Contact the kubelet on its healthz port instead of its readonly port
**What this PR does / why we need it**:
In order for us to disable the kubelet's readonly port in v1.11 (https://github.com/kubernetes/kubeadm/issues/732), we need to cut the dependency on that port being open. Instead, we can use the dedicated healthz port (using the defaults `--healthz-bind-address=127.0.0.1` and `--healthz-port=10248`, xref: https://kubernetes.io/docs/reference/command-line-tools-reference/kubelet/)
**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Part of https://github.com/kubernetes/kubeadm/issues/732
**Special notes for your reviewer**:
**Release note**:
```release-note
NONE
```
@kubernetes/sig-cluster-lifecycle-pr-reviews @liztio
Currently kubeadm only performs an upgrade if the etcd cluster is
colocated with the control plane node. As this is only one possible
configuration, kubeadm should support upgrades with etcd clusters
that are not local to the node.
Signed-off-by: Craig Tracey <craigtracey@gmail.com>
Automatic merge from submit-queue (batch tested with PRs 63492, 62379, 61984, 63805, 63807). 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>.
Sort arguments before joining them, for reproducible return string
**What this PR does / why we need it**:
This PR makes kubeadm static pod manifest generation consistent. Right now when `kubeadm init` is called repeatedly, the generated pod manifest files under /etc/kubernetes/manifest/ are changing. Its really hard to test how a configuration change effects the manifest files.
The current implementation is ranging over a map[string]string which will be happening in a random order, generating different pod manifests even without changing any configuration.
The suggested solution makes pom manifest generation idempotent. It opens up integration test possibilities, like testing whole yaml result of `kubeadm alpha phase controlplane`.
**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 #
**Special notes for your reviewer**:
**Release note**:
```release-note
NONE
```
/sig cluster-lifecycle
/assign @luxas
Automatic merge from submit-queue (batch tested with PRs 63492, 62379, 61984, 63805, 63807). 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 e2e test to verify that GPU pool is not scaled up if GPUs are not requested
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 63492, 62379, 61984, 63805, 63807). 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>.
validation: improve ProjectedVolume validation errors
* only report "may not specify more than 1 volume type" once
* fix incorrectly reported field paths
* continue to traverse into projections to report further errors.
@kubernetes/sig-storage-pr-reviews
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 63492, 62379, 61984, 63805, 63807). 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>.
remove PodPreset and enable resources for Priority admission plugins in e2e-gce
**What this PR does / why we need it**:
e2e-gce start kube-apiserver without admission PodPreset and enable resources for Priority
**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#62377
**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>.
track/close kubelet->API connections on heartbeat failure
xref #48638
xref https://github.com/kubernetes-incubator/kube-aws/issues/598
we're already typically tracking kubelet -> API connections and have the ability to force close them as part of client cert rotation. if we do that tracking unconditionally, we gain the ability to also force close connections on heartbeat failure as well. it's a big hammer (means reestablishing pod watches, etc), but so is having all your pods evicted because you didn't heartbeat.
this intentionally does minimal refactoring/extraction of the cert connection tracking transport in case we want to backport this
* first commit unconditionally sets up the connection-tracking dialer, and moves all the cert management logic inside an if-block that gets skipped if no certificate manager is provided (view with whitespace ignored to see what actually changed)
* second commit plumbs the connection-closing function to the heartbeat loop and calls it on repeated failures
follow-ups:
* consider backporting this to 1.10, 1.9, 1.8
* refactor the connection managing dialer to not be so tightly bound to the client certificate management
/sig node
/sig api-machinery
```release-note
kubelet: fix hangs in updating Node status after network interruptions/changes between the kubelet and API server
```
Automatic merge from submit-queue (batch tested with PRs 55511, 63372, 63400, 63100, 63769). 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>.
print result object from kubectl taint correctly
fixes#63767
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 55511, 63372, 63400, 63100, 63769). 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>.
Create pkg/scheduling/apis/v1beta1 and move priorityClass to beta
**What this PR does / why we need it**:
This is for creating pkg/apis/scheduling/v1beta1 so that priorityClasses could be moved to beta.
**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Part of #57471
**Special notes for your reviewer**:
/cc @bsalamat @aveshagarwal
**Release note**:
```release-note
The `PriorityClass` API is promoted to `scheduling.k8s.io/v1beta1`
```
Automatic merge from submit-queue (batch tested with PRs 55511, 63372, 63400, 63100, 63769). 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 kubeadm upgrade path for CoreDNS
**What this PR does / why we need it**:
This PR achieves the following:
1. Retains the CoreDNS ConfigMap defined by the user during `kubeadm upgrade`.
2. Removes the CoreDNS deployment when user decides to install kube-dns during `kubeadm upgrade`.
**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 #
**Special notes for your reviewer**:
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 55511, 63372, 63400, 63100, 63769). 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>.
[GCE] check for new backend naming scheme
**What this PR does / why we need it**:
Checks for both the old Backend naming scheme (Nodeport in the name) and the new naming scheme (same scheme used to name NEGs). We need to check for both, in order for both the tests against Ingress head (once https://github.com/kubernetes/ingress-gce/pull/239 gets merged) and tests against prior Ingress versions to pass.
See https://github.com/kubernetes/ingress-gce/pull/239 .
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 55511, 63372, 63400, 63100, 63769). 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>.
k8s.io/client-go/{kubernetes,rest}: remove dependency on pflag
This change removes an unused pflag utility which cuts a client-go transitive dependency on that library.
Verified using `go list`
```
go list -f '{{join .Deps "\n"}}' k8s.io/kubernetes/vendor/k8s.io/client-go/kubernetes \
| xargs go list -f '{{if not .Standard}}{{.ImportPath}}{{end}}' \
| grep pflag
```
/cc @sttts @kubernetes/sig-api-machinery-pr-reviews
Edit: it looks like k8s.io/client-go/tools/clientcmd still imports this. So practically, if you're loading from a kubeconfig you're going to get the import. Might still be a good cleanup to make though.
```release-note
NONE
```
In preparation for creating a `kubeadm config images pull`
this commit refactors `kubeadm config list-images` into
`kubeadm config images list`.
Signed-off-by: Chuck Ha <ha.chuck@gmail.com>
Automatic merge from submit-queue (batch tested with PRs 63787, 62003). 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 localport open with --nodeport-addresses specified
**What this PR does / why we need it**:
Fix localport open with --nodeport-addresses specified.
**Which issue(s) this PR fixes**:
Fixes#61953
**Special notes for your reviewer**:
@ephur
**Release note**:
```release-note
Services can listen on same host ports on different interfaces with --nodeport-addresses specified
```
Automatic merge from submit-queue (batch tested with PRs 63787, 62003). 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: Use the API machinery for marshalling
**What this PR does / why we need it**:
Similar to https://github.com/kubernetes/kubernetes/pull/63723, we should not use a generic `yaml`
package for marshalling, but use the API machinery we have already in place. This now correctly includes `apiVersion` and `kind` when printing the configuration in the upgrade procedure.
**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Part of https://github.com/kubernetes/community/pull/2131
**Special notes for your reviewer**:
**Release note**:
```release-note
NONE
```
@kubernetes/sig-cluster-lifecycle-pr-reviews @liztio
Automatic merge from submit-queue (batch tested with PRs 63588, 63806). 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 - do not generate etcd ca/certs for external etcd
**What this PR does / why we need it**:
Currently we generate an etcd CA and certificates even if we are specifying an external etcd cluster when running `kubeadm init`, this PR changes this behavior to skip generating the etcd CA and certificates if configured for an external etcd cluster.
**Which issue(s) this PR fixes**
Fixes https://github.com/kubernetes/kubeadm/issues/807
**Release note**:
```release-note
kubeadm will no longer generate an unused etcd CA and certificates when configured to use an external etcd cluster.
```
Automatic merge from submit-queue (batch tested with PRs 63588, 63806). 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>.
Run FSGroup tests by default.
There is no special feature flag for FSGroup and the tests can run in all test suites. They're reasonably fast too.
**Release note**:
```release-note
NONE
```
cc: @jeffvance