Automatic merge from submit-queue
Validate if service has duplicate targetPort
**What this PR does / why we need it**:
Validate if a service has dup targetport
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes#47222
**Special notes for your reviewer**:
/cc @thockin
@kubernetes/sig-network-pr-reviews
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 47850, 47835, 46197, 47250, 48284)
Populate endpoints for headless service with no ports
**What this PR does / why we need it**:
- populate endpoints with headless service (thanks @fraenkel for the original PR!)
- allow ports with headless service
- nits
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes#32796https://github.com/kubernetes/kubernetes/issues/32796#issuecomment-270462724
**Special notes for your reviewer**:
/cc @thockin @fraenkel
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 47850, 47835, 46197, 47250, 48284)
Allocate clusterIP when change service type from ExternalName to ClusterIP
**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#35354#46190
**Special notes for your reviewer**:
/cc @smarterclayton @thockin
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 47850, 47835, 46197, 47250, 48284)
dockershim: checkpoint HostNetwork property
To ensure kubelet doesn't attempt network teardown on HostNetwork
containers that no longer exist but are still checkpointed, make
sure we preserve the HostNetwork property in checkpoints. If
the checkpoint indicates the container was a HostNetwork one,
don't tear down the network since that would fail anyway.
Related: https://github.com/kubernetes/kubernetes/issues/44307#issuecomment-299548609
@freehan @kubernetes/sig-network-misc
Automatic merge from submit-queue (batch tested with PRs 47619, 47951, 46260, 48277)
iSCSi plugin: Remove redundant nil check
**What this PR does / why we need it**:
This patch is for cleanup of redundant nil check in iSCSI plugin.
This was mentioned at the code review on origin github thread.
https://github.com/openshift/origin/pull/14565
**Which issue this PR fixes**
**Special notes for your reviewer**:
**Release note**:
```
NONE
```
Automatic merge from submit-queue
Add type conversion judgment
If do not type conversion judgment, there may be panic.
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 47286, 47729)
Set default User-Agent on http probe
**Set a default User-Agent on `httpGet` probes**:
Currently the default Go HTTP client sets a `User-Agent` specific to the language and version, but every Go client has the same one. In Kubernetes, users can override the User-Agent by setting a header in their probe definition, but its tedious to do this everywhere.
This PR changes the default User-Agent only if not set (or removed) in the probe definition.
**Which issue this PR fixes**
fixes#29025
**Special notes for your reviewer**:
**Release note**:
```release-note
Change default `httpGet` probe `User-Agent` to `kube-probe/<version major.minor>` if none specified, overriding the default Go `User-Agent`.
```
Automatic merge from submit-queue (batch tested with PRs 47286, 47729)
Add client certificate authentication to Azure cloud provider
This adds support for client cert authentication in Azure cloud provider. The certificate can be provided in PKCS #12 format with password protection. Not that this authentication will be active only when no client secret is configured.
cc @brendandburns @colemickens
Automatic merge from submit-queue
fixes to two pkg/util unit tests
Spun out from #46537. Trivial unit test fixes:
- `pkg/util/ebtables/ebtables_test.go`'s tests are incorrectly named and so weren't getting run by "go test". (And one of them had a typo in it so it failed after the name was fixed)
- In `pkg/util/iptables/iptables_test.go`, the deletion tests were apparently written by copying+pasting+editing the creation tests, but they forgot to change the names of the tests to match the behavior that was being tested in the new version
Automatic merge from submit-queue
Remove repeat type conversions
Here is the type of conversion for the variable is repeated.
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue
deletePod handler in the deployment controller shouldn't set owner refs
**What this PR does / why we need it**:
This PR stops the deletePod handler in the deployment controller from adopting replica sets when determining if a deployment needs to be requeued. It leaves this logic to the replication loop, removing the replica set adoption side effect.
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes#46933
**Special notes for your reviewer**:
@kargakis PR for delete pod handler setting owner refs issue
**Release note**:
```release-note
```
Automatic merge from submit-queue (batch tested with PRs 48183, 45611, 48065)
Fix Unstructured field accessor
Fix the Unstructured GetDeletionGracePeriodSeconds accessor which was
always returning nil regardless of the underlying stored value. The
field value always appearing nil prevents Custom Resource instances
from being deleted when garbage collection is enabled for CRs and
when DeletePropagationOrphan is used. More generally, this fix means that
delete-on-update now works for CR instances.
Add some test coverage for Unstructured metadata deserialization.
The Unstructured DeletionGracePeriodSeconds field marshals as a value
type from JSON and as a pointer type via SetDeletionGracePeriodSeconds.
The GetDeletionGracePeriodSeconds method now supports handling both
int64 and *int64 values so that either underlying value can be returned.
Add a reflection-based unit test which attempts to exercise all the
Object Get/Set methods for nil handling.
```release-note
Registries backed by the generic Store's `Update` implementation support delete-on-update, which allows resources to be automatically deleted during an update provided:
* Garbage collection is enabled for the Store
* The resource being updated has no finalizers
* The resource being updated has a non-nil DeletionGracePeriodSeconds equal to 0
With this fix, Custom Resource instances now also support delete-on-update behavior under the same circumstances.
```
Automatic merge from submit-queue (batch tested with PRs 48183, 45611, 48065)
kubectl: 'apply view-last-applied' must not use printf() semantics
**What this PR does / why we need it**:
This fixes `kubectl apply view-last-applied` to not use `fmt.Fprintf()`, as this will cause format codes in the YAML/JSON to be interpreted. For example, if a resource manifest contains `%r`, this would cause `view-last-applied` so print `%!r(MISSING)`.
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes#45608.
**Special notes for your reviewer**:
**Release note**:
```release-note
Fixes an edge case where "kubectl apply view-last-applied" would emit garbage if the data contained Go format codes.
```
Fix the Unstructured GetDeletionGracePeriodSeconds accessor which was
always returning nil regardless of the underlying stored value. The
field value always appearing nil prevents Custom Resource instances
from being deleted when garbage collection is enabled for CRs and
when DeletePropagationOrphan is used. More generally, this fix means that
delete-on-update now works for CR instances.
Add some test coverage for Unstructured metadata deserialization.
The Unstructured DeletionGracePeriodSeconds field marshals as a value
type from JSON and as a pointer type via SetDeletionGracePeriodSeconds.
The GetDeletionGracePeriodSeconds method now supports handling both
int64 and *int64 values so that either underlying value can be returned.
Add a reflection-based unit test which attempts to exercise all the
Object Get/Set methods for nil handling.
Automatic merge from submit-queue (batch tested with PRs 48123, 48079)
kubectl/cmd: many small refactors
#### Improve Code Quality in `pkg/kubectl/cmd`
___
* [x] Rename variables and functions to match Go convention. For example, UsageError --> UsageErrorf.
* [x] Remove redundant or unreachable code.
* [x] Simplify some utility functions (no functionality changes).
* [x] Fix hanging 'if { return } else { return }' constructs.
* [x] Fix several incorrect printf verbs.
These changes were extracted from the refactoring and flag parity work I'm doing in kubectl.
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 48123, 48079)
[Kubelet] Fix race condition in container manager
**What this PR does / why we need it**:
This fixes a race condition where the container manager capacity map was being updated without synchronization. It moves the storage capacity detection to kubelet initialization, which happens serially in one thread.
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes#48045
**Release note**:
```release-note
Fixes kubelet race condition in container manager.
```
Centralize Capacity discovery of standard resources in Container manager.
Have storage derive node capacity from container manager.
Move certain cAdvisor interfaces to the cAdvisor package in the process.
This patch fixes a bug in container manager where it was writing to a map without synchronization.
Signed-off-by: Vishnu kannan <vishnuk@google.com>
Automatic merge from submit-queue (batch tested with PRs 45610, 47628)
Add Priority to Kubernetes API
**What this PR does / why we need it**: This is the first in a series of PRs to add priority to Kubernetes API. Subsequent PRs will add priority name resolution to admission controller.
**Release note**:
```release-note
Add PriorityClassName and Priority fields to PodSpec.
```
* Rename variables and functions to match Go convention.
For example, UsageError --> UsageErrorf.
* Remove redundant or unreachable code.
* Simplify some utility functions (no functionality changes).
* Fix hanging 'if { return } else { return }' constructs.
* Fix several incorrect printf verbs.
Automatic merge from submit-queue (batch tested with PRs 47675, 48001)
cmd/create_deployment: refactor long function
Refactor the `createDeployment` function under `pkg/kubectl/cmd`.
- [x] Behavior has been extracted to two helper functions.
- [x] Behavior remains identical.
- [x] Logic has been made explicit through function naming and comments.
This is essentially the pattern I've been following in my larger branches (the ones that are pending the merge of #46468). Want to get some design feedback before I get too far away from `master`.
Thanks!
cc @apelisse @mengqiy @droot
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 48106, 46761)
removing unnecessary shallow copy on SyncService
**What this PR does / why we need it**: remove shallow copy + avoid using same index name as the nested loop
**Which issue this PR fixes**: fixes#46703
**Special notes for your reviewer**:
**Release note**:
```release-note
```
Automatic merge from submit-queue (batch tested with PRs 48139, 48042, 47645, 48054, 48003)
Pipe clusterID into gce_loadbalancer_external.go
**What this PR does / why we need it**: Small cleanup for GCE ELB codes.
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes#48002
**Special notes for your reviewer**:
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 47038, 47105)
extending DefaultExternalHost for any registred cloud provider
**What this PR does / why we need it**: this PR enables DefaultExternalHost to work with any registered cloud provider.
**Which issue this PR fixes** : fixes#46567
**Special notes for your reviewer**:
**Release note**:
```release-note
When determining the default external host of the kube apiserver, any configured cloud provider is now consulted
```
Automatic merge from submit-queue
Use endpoints informer for the endpoint controller
This substantially reduces the number of API calls made by the endpoint
controller. Currently the controller makes an API call per endpoint for
each service that is synced. When the 30s resync is triggered, this
results in an API call for every single endpoint in the cluster. This
quickly exceeds the default qps/burst limit of 20/30 even in small
clusters, leading to delays in endpoint updates.
This change modifies the controller to use the endpoint informer cache
for all endpoint GETs. This means we only make API calls for changes in
endpoints. As a result, qps only depends on the pod activity in the
cluster, rather than the number of services.
**What this PR does / why we need it**:
Address endpoint update delays as described in https://github.com/kubernetes/kubernetes/issues/47597.
**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/47597
**Special notes for your reviewer**:
**Release note**:
```release-note
```
This substantially reduces the number of API calls made by the endpoint
controller. Currently the controller makes an API call per endpoint for
each service that is synced. When the 30s resync is triggered, this
results in an API call for every single endpoint in the cluster. This
quickly exceeds the default qps/burst limit of 20/30 even in small
clusters, leading to delays in endpoint updates.
This change modifies the controller to use the endpoint informer cache
for all endpoint GETs. This means we only make API calls for changes in
endpoints. As a result, qps only depends on the pod activity in the
cluster, rather than the number of services.
Automatic merge from submit-queue (batch tested with PRs 47484, 47904, 48034)
fix nits in kubelet server
Signed-off-by: allencloud <allen.sun@daocloud.io>
**What this PR does / why we need it**:
fix nits in kubelet server
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes #
NONE
**Special notes for your reviewer**:
NONE
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 44058, 48085, 48077, 48076, 47823)
don't pass CRI error through to waiting state reason
Raw gRPC errors are getting into the `Reason` field of the container status `State`, causing it to be output inline on a `kubectl get pod`
xref https://bugzilla.redhat.com/show_bug.cgi?id=1449820
Basically the issue is that the err and msg are reversed in `startContainer()`. The msg is short and the err is long. It should be the other way around.
This PR changes `startContainer()` to return a short error that becomes the Reason and the extracted gPRC error description that becomes the Message.
@derekwaynecarr @smarterclayton @eparis
Automatic merge from submit-queue (batch tested with PRs 44058, 48085, 48077, 48076, 47823)
Move iptables logging in kubeproxy from Errorf to V(2).Infof
Fixes https://github.com/kubernetes/kubernetes/issues/48052
This will stop fluentd from OOM'ing in reasonably large clusters with services due to kube-proxy. You'll still get iptables printed on setups which run at >= v2, but we can at least optout.
@bowei Does this look reasonable?
cc @kubernetes/sig-network-misc
Automatic merge from submit-queue (batch tested with PRs 44058, 48085, 48077, 48076, 47823)
Make background garbage collection cascading
Fix#44046, fix#47843 where user reported that the garbage collector didn't delete pods when a deployment was deleted with PropagationPolicy=Background.
The cause is that when propagating background garbage collection request, the garbage collector deletes dependents with DeleteOptions.PropagationPolicy=nil, which means the default GC policy of a resource (defined by its REST strategy) and the existing GC-related finalizers will decide how the delete request is propagated further. Unfortunately, the default GC policy for RS is orphaning, so the pods are behind when a deployment is deleted.
This PR changes the garbage collector to delete dependents with DeleteOptions.PropagationPolicy=Background when the owner is deleted in background. This means the dependent's existing GC finalizers will be overridden, making orphaning less flexible (see this made-up [case](https://github.com/kubernetes/kubeadm/issues/149#issuecomment-278942012)). I think sacrificing the flexibility of orphaning is worthwhile, because making the behavior of background garbage collection matching users' expectation is more important.
cc @lavalamp @kargakis @krmayankk @enisoc
```release-note
The garbage collector now cascades deletion properly when deleting an object with propagationPolicy="background". This resolves issue [#44046](https://github.com/kubernetes/kubernetes/issues/44046), so that when a deployment is deleted with propagationPolicy="background", the garbage collector ensures dependent pods are deleted as well.
```
Kubernetes Submit Queue