Align metrics-server metrics-resolution with the upstream manifests so
that scalability tests are running a similar configuration of
metrics-server as the one we are running in the e2e tests.
Signed-off-by: Damien Grisonnet <dgrisonn@redhat.com>
grpclog.SetLoggerV is not thread-safe and may only be called before code starts
using GRPC. Calling RunCustomEtcd multiple times, for example in
k8s.io/kubernetes/test/integration/apiserver.TestWatchCacheUpdatedByEtcd,
causes a data race:
WARNING: DATA RACE
Read at 0x00000c8e8d20 by goroutine 135612:
k8s.io/kubernetes/vendor/google.golang.org/grpc/grpclog.V()
/home/prow/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/vendor/google.golang.org/grpc/grpclog/grpclog.go:41 +0x30
k8s.io/kubernetes/vendor/google.golang.org/grpc/grpclog.(*componentData).V()
/home/prow/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/vendor/google.golang.org/grpc/grpclog/component.go:103 +0x4e
k8s.io/kubernetes/vendor/google.golang.org/grpc/internal/transport.(*loopyWriter).run.func1()
/home/prow/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/vendor/google.golang.org/grpc/internal/transport/controlbuf.go:528 +0xf1
runtime.deferreturn()
/home/prow/go/src/k8s.io/kubernetes/_output/local/.gimme/versions/go1.20.2.linux.amd64/src/runtime/panic.go:476 +0x32
k8s.io/kubernetes/vendor/google.golang.org/grpc/internal/transport.newHTTP2Client.func6()
/home/prow/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/vendor/google.golang.org/grpc/internal/transport/http2_client.go:442 +0x112
Previous write at 0x00000c8e8d20 by goroutine 140228:
k8s.io/kubernetes/vendor/google.golang.org/grpc/grpclog.SetLoggerV2()
/home/prow/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/vendor/google.golang.org/grpc/grpclog/loggerv2.go:76 +0xc6a
k8s.io/kubernetes/test/integration/framework.RunCustomEtcd()
/home/prow/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/test/integration/framework/etcd.go:153 +0xb89
k8s.io/kubernetes/test/integration/apiserver.multiEtcdSetup()
/home/prow/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/test/integration/apiserver/watchcache_test.go:40 +0xac
k8s.io/kubernetes/test/integration/apiserver.TestWatchCacheUpdatedByEtcd()
/home/prow/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/test/integration/apiserver/watchcache_test.go:88 +0x4a
testing.tRunner()
/home/prow/go/src/k8s.io/kubernetes/_output/local/.gimme/versions/go1.20.2.linux.amd64/src/testing/testing.go:1576 +0x216
testing.(*T).Run.func1()
/home/prow/go/src/k8s.io/kubernetes/_output/local/.gimme/versions/go1.20.2.linux.amd64/src/testing/testing.go:1629 +0x47
Since the codebase already migrates to genericiooptions.IOStreams,
external tools will automatically start consuming the new one.
This PR deletes deprecated struct because codebase does not rely on it
already. We keep `NewTestIOStreams` and `NewTestIOStreamsDiscard` functions
to provide users a smooth migration.
Currently, `genericclioptions` package imports `resource` package in cli-runtime
(i.e. builder_flags uses builder object in resource). Therefore, it is not allowed
that `resource` can import any package in `genericclioptions`(due to disallowed import cycles).
It is already reasonable burden except `genericclioptions.IOStreams`.
There are some cases we want to raise a warning to user in builder but
it can not be achieved due to resource package can not depend on
IOStreams. Since IOStreams solely contains go primitives, this PR
deprecates `genericclioptions.IOStreams` and adds `genericiooptions.IOStreams`.
Thanks to that, that will add capability of using IOStreams also in
builders, etc.
In most cases `dir` arg of `SetUpAt()` method of `volume.Mounter` interface is the same as `mounter.GetPath()` because we usually call `SetUpAt()` from `SetUp()` like this:"
```
func (ed *emptyDir) SetUp(mounterArgs volume.MounterArgs) error {
return ed.SetUpAt(ed.GetPath(), mounterArgs)
}
```
(this example is from `volume/emptydir/empty_dir.go`, but there are plenty other examples like that in `volume/*`)
However, there is currently one exception. This is from `volume/projected/projected.go`:
```
if err := wrapped.SetUpAt(dir, mounterArgs); err != nil {
return err
}
```
(see 96306f144a/pkg/volume/projected/projected.go (L203))
In this case `dir` is not equal to `wrapped.GetPath()` and `volume.SetVolumeOwnership()` fails when called from `SetUpAt()` of wrapped volume:
```
lstat /var/lib/kubelet/pods/a2f6e58f-7edf-4c48-a97c-ef1b8fd3caf6/volumes/kubernetes.io~empty-dir/wrapped_kube-api-access-knvkv: no such file or directory
```
To fix the issue let's pass `dir` arg to `volume.SetVolumeOwnership()` explicitly, and use it instead of `mounter.GetPath()`.
Users can pass resources into `kubectl events` command via `--for` flag,
if they have desire to only get events for the resource they specify.
However, current `kubectl events` does not support passing fully qualified
names(e.g. `replicasets.apps`, `cronjobs.v1.batch`, etc.). This PR adds support
for this.
The newly added `MirrorPodWithGracePeriod when create a mirror pod and
the container runtime is temporarily down during pod termination` test
is currently flaking because in some cases when it is run there are
other pods from other tests that are still in progress of being
terminated. This results in the test failing because it asserts metrics
that assume that there is only one pod running on the node.
To fix the flake, prior to starting the test, verify that no pods exist
in the api server other then the newly created mirror pod.
Signed-off-by: David Porter <david@porter.me>
A pod that cannot be started yet (due to static pod fullname
exclusion when UIDs are reused) must be accounted for in the
pod worker since it is considered to have been admitted and will
eventually start.
Due to a bug we accidentally cleared pendingUpdate for pods that
cannot start yet which means we can't report the right metric to
users in kubelet_working_pods and in theory we might fail to start
the pod in the future (although we currently have not observed
that in tests that should catch such an error). Describe, implement,
and test the invariant that when startPodSync returns in every path
that either activeUpdate OR pendingUpdate is set on the status, but
never both, and is only nil when the pod can never start.
This bug was detected by a "programmer error" assertion we added
on metrics that were not being reported, suggesting that we should
be more aggressive on using log assertions and automating detection
in tests.
