Automatic merge from submit-queue (batch tested with PRs 47883, 47179, 46966, 47982, 47945)
add level for print flags
Signed-off-by: sakeven <jc5930@sina.cn>
**What this PR does / why we need it**:
It's ugly to print all flags whenever.
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*:
fixes#47932
**Special notes for your reviewer**:
**Release note**:
```
NONE
```
Automatic merge from submit-queue (batch tested with PRs 47883, 47179, 46966, 47982, 47945)
Add feature gating to REST Compression
**What this PR does / why we need it**: Adds feature gating to opt out of REST API compression
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes#46963
**Special notes for your reviewer**: This PR is a fix / addendum to #45666
**Release note**:
```release-note
```
Automatic merge from submit-queue
Lower etcd compacted loglevel
Fixes#47941
When we run apiserver, it will consistently produce logs like:
> I0620 15:27:44.627978 2765 compact.go:159] etcd: compacted rev (7825), endpoints ([http://127.0.0.1:2379])
I0620 15:32:44.632311 2765 compact.go:159] etcd: compacted rev (8193), endpoints ([http://127.0.0.1:2379])
I0620 15:37:44.643658 2765 compact.go:159] etcd: compacted rev (8552), endpoints ([http://127.0.0.1:2379])
...
which does not mean much overall, we may change the loglevel and lower the frequency.
**Release note**:
-->
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 46151, 47602, 47507, 46203, 47471)
rebase gophercloud to support HTTP status 300 in pagination, so listingng Cinder v1/v2 API versions won't break
**What this PR does / why we need it**:
Since Cinder block v1/v2 support lands in 1.7, we have to ensure Cinder block api list won't fail due to https://github.com/gophercloud/gophercloud/pull/384.
**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes#47651
**Special notes for your reviewer**:
@anguslees @NickrenREN
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue
Fix link to apiextensions client-go example
Fixes link to client-go example for apiextensions-apiserver.
Fixing #47211.
**Special notes for your reviewer**:
When I run `hack/update-staging-client-go.sh`, I get this error:
```
!!! Error in staging/copy.sh:132
Error in staging/copy.sh:132. 'git commit -q -m "Snapshot" > /dev/null' exited with status 1
Call stack:
1: staging/copy.sh:132 main(...)
Exiting with status 1
```
What am I missing here?
/cc @caesarxuchao @sttts
**Release note**:
```
NONE
```
Automatic merge from submit-queue (batch tested with PRs 42252, 42251, 42249, 47512, 47887)
fix HostAliases' json keys to be hostAlias instead of hostMapping to reflect actual feature name
**What this PR does / why we need it**: a rename was introduce during the middle of #44641 to change from `hostMappings` to `hostAliases`. the Go structs were updated, but I neglected to update the json keys. They should be in sync.
**Special notes for your reviewer**: I messed up. This is an API change. I hope this is still ok to be in the 1.7 release.
**Release note**:
```release-note
HostAliases is now parsed with `hostAliases` json keys to be in line with the feature's name.
```
Automatic merge from submit-queue (batch tested with PRs 47878, 47503, 47857)
restore working aggregator and avoid duplicate informers
Fixes https://github.com/kubernetes/kubernetes/issues/47866
This runs the informer all the way through and makes sure its started.
@lavalamp ptal
@kubernetes/sig-api-machinery-bugs
Automatic merge from submit-queue (batch tested with PRs 47851, 47824, 47858, 46099)
Revert 44714 manually
#44714 broke backward compatibility for old swagger spec that kubectl still uses. The decision on #47448 was to revert this change but the change was not automatically revertible. Here I semi-manually remove all references to UnixUserID and UnixGroupID and updated generated files accordingly.
Please wait for tests to pass then review that as there may still be tests that are failing.
Fixes#47448
Adding release note just because the original PR has a release note. If possible, we should remove both release notes as they cancel each other.
**Release note**: (removed by caesarxuchao)
UnixUserID and UnixGroupID is reverted back as int64 to keep backward compatibility.
Automatic merge from submit-queue (batch tested with PRs 47851, 47824, 47858, 46099)
Revert "Merge pull request #43946 from jhorwit2/jah/host-path-psp"
fixes#47863
This reverts commit b5eadb5d6b, reversing
changes made to 1889d654f5.
**What this PR does / why we need it**:
Revert whitelist host paths in psp due to API concerns. Please refer to https://github.com/kubernetes/kubernetes/pull/47811 for the concerns.
**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**:
cc @liggitt @ericchiang @smarterclayton
**Release note**:
```release-note
```
Automatic merge from submit-queue (batch tested with PRs 34515, 47236, 46694, 47819, 47792)
Fix rawextension decoding in update
Fixes https://github.com/kubernetes/kubernetes/issues/47842
The `Create` handler was correctly decoding into the internal version, but the `Update` handler was not.
Top-level objects were not affected, because the type of the `New()` object returned by the rest handler governs the destination group/version/kind.
If a field within the object is of type `RawExtension`, and converts to a `runtime.Object` field in the internal object, the `runtime.Object` field provides no information about the desired group/version/kind, so the decoder's groupversioner governs.
This would manifest as the resthandler's Update function being given an internal top-level object with it's runtime.Object field containing an external object.
Automatic merge from submit-queue (batch tested with PRs 34515, 47236, 46694, 47819, 47792)
Clear auth config when gcp app default credentials fail
**What this PR does / why we need it**:
Specific use case is when utilizing multiple gcp accounts, the user may provide credentials for the wrong account.
This change ensures the incorrect credentials are not cached in auth config, and logs an appropriate message.
**Which issue this PR fixes** : fixes#38075
**Special notes for your reviewer**:
**Release note**:
```release-note
Tokens retrieved from Google Cloud with application default credentials will not be cached if the client fails authorization
```
Automatic merge from submit-queue (batch tested with PRs 46604, 47634)
Added example for status errors in go client
This PR adds status error handling examples to the go client examples, for both in-cluster and out-of-cluster usage. Fixes https://github.com/kubernetes/client-go/issues/163
Automatic merge from submit-queue (batch tested with PRs 47726, 47693, 46909, 46812)
Plumb service resolver into webhook AC
This is the last piece of plumbing needed for https://github.com/kubernetes/features/issues/209
Automatic merge from submit-queue (batch tested with PRs 47669, 40284, 47356, 47458, 47701)
Serialized protobuf should have stable map order
Leverage `stable_marshaller_all` tag in gogo-protobuf to ensure all of our maps have stable ordering when serialized to protobuf. This preserves the behavior from JSON that we rely on to avoid writing to etcd except when the content has changed.
Will slightly increase allocations (1 slice per non-empty map in an object) during Encode, which has a minor impact on CPU. However, avoiding writes when a client issues a blind put results in significantly less CPU across the whole cluster (avoiding a new etcd version for an Endpoints object might save ~1 core/sec on large clusters).
Fixes#47678
```release-note
The protobuf serialization of API objects has been updated to store maps in a predictable order to ensure that the representation of that object does not change when saved into etcd. This prevents the same object from being seen as being modified, even when no values have changed.
```
Specific use case is when utilizing multiple
gcp accounts, the user may provide credentials
for the wrong account.
This change ensures the incorrect credentials
are not cached in auth config, and logs an
appropriate message.
