Automatic merge from submit-queue (batch tested with PRs 66177, 66185, 67136, 67157, 65065). 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>.
kubelet: reduce logging for backoff situations
xref https://bugzilla.redhat.com/show_bug.cgi?id=1555057#c6
Pods that are in `ImagePullBackOff` or `CrashLoopBackOff` currently generate a lot of logging at the `glog.Info()` level. This PR moves some of that logging to `V(3)` and avoids logging in situations where the `SyncPod` only fails because pod are in a BackOff error condition.
@derekwaynecarr @liggitt
Automatic merge from submit-queue (batch tested with PRs 66491, 66587, 66856, 66657, 66923). 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>.
Revert #63905: Setup dns servers and search domains for Windows Pods
**What this PR does / why we need it**:
From https://github.com/kubernetes/kubernetes/pull/63905#issuecomment-396709775:
> I don't think this change does anything on Windows. On windows, the network endpoint configuration is taken care of completely by CNI. If you would like to pass on the custom dns polices from the pod spec, it should be dynamically going to the cni configuration that gets passed to CNI. From there, it would be passed down to platform and would be taken care of appropriately by HNS.
> etc\resolve.conf is very specific to linux and that should remain linux speicfic implementation. We should be trying to move away from platform specific code in Kubelet.
Docker is not managing the networking here for windows. So it doens't really care about any network settings. So passing it to docker shim's hostconfig also doens;t make sense here.
DNS for Windows containers will be set by CNI plugins. And this change also introduced two endpoints for sandbox container. So this PR reverts #63905 .
**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**:
The PR should also be cherry-picked to release-1.11.
Also, https://github.com/kubernetes/kubernetes/issues/66588 is opened to track the process of pushing this to CNI.
**Release note**:
```release-note
Revert #63905: Setup dns servers and search domains for Windows Pods. DNS for Windows containers will be set by CNI plugins.
```
/sig windows
/sig node
/kind bug
This allows kubelets to stop the necessary work when the context has
been canceled (e.g., connection closed), and not leaking a goroutine
and inotify watcher waiting indefinitely.
Automatic merge from submit-queue (batch tested with PRs 67177, 53042). 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>.
Adding unit tests to methods of pod's format
What this PR does / why we need it:
Add unit test cases, thank you!
Automatic merge from submit-queue (batch tested with PRs 66512, 66946, 66083). 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>.
kubelet/cm/cpumanager: Fix unused variable "skipIfPermissionsError"
The variable "skipIfPermissionsError" is not needed even when
permission error happened.
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>.
clean up useless variables in deviceplugin/types.go
**What this PR does / why we need it**:
some variables is useless for reasons, I think we need a clean up.
**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. 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>.
error out empty hostname
**What this PR does / why we need it**:
For linux, the hostname is read from file `/proc/sys/kernel/hostname` directly, which can be overwritten with whitespaces.
Should error out such invalid hostnames.
**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Fixeskubernetes/kubeadm#835
**Special notes for your reviewer**:
/cc luxas timothysc
**Release note**:
```release-note
nodes: improve handling of erroneous host names
```
Automatic merge from submit-queue (batch tested with PRs 62901, 66562, 66938, 66927, 66926). 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>.
kubelet: volumemanager: poll immediate when waiting for volume attachment
Currently, `WaitForAttachAndMount()` introduces a 300ms minimum delay by using `wait.Poll()` rather than `wait.PollImmediate()`. This wait constitutes >99% of the total processing time for `syncPod()`. Changing this reduced `syncPod()` processing time for a simple busybox pod with one emptyDir volume from 302ms to 2ms.
@derekwaynecarr @pmorie @smarterclayton @jsafrane
/sig node
/release-note-none
Automatic merge from submit-queue (batch tested with PRs 66190, 66871, 66617, 66293, 66891). 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 set cgroup parent when --cgroups-per-qos is disabled
When --cgroups-per-qos=false (default is true), kubelet sets pod
container management to podContainerManagerNoop implementation and
GetPodContainerName() returns '/' as cgroup parent (default cgroup root).
(1) In case of 'systemd' cgroup driver, '/' is invalid parent as
docker daemon expects '.slice' suffix and throws this error:
'cgroup-parent for systemd cgroup should be a valid slice named as \"xxx.slice\"'
(5fc12449d8/daemon/daemon_unix.go (L618))
'/' corresponds to '-.slice' (root slice) in systemd but I don't think
we want to assign root slice instead of runtime specific default value.
In case of docker runtime, this will be 'system.slice'
(e2593239d9/daemon/oci_linux.go (L698))
(2) In case of 'cgroupfs' cgroup driver, '/' is valid parent but I don't
think we want to assign root instead of runtime specific default value.
In case of docker runtime, this will be '/docker'
(e2593239d9/daemon/oci_linux.go (L695))
Current fix will not set the cgroup parent when --cgroups-per-qos is disabled.
```release-note
Fix pod launch by kubelet when --cgroups-per-qos=false and --cgroup-driver="systemd"
```
Automatic merge from submit-queue (batch tested with PRs 66190, 66871, 66617, 66293, 66891). 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 nil pointer dereference in node_container_manager#enforceExisting
**What this PR does / why we need it**:
fix nil pointer dereference in node_container_manager#enforceExisting
**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#66189
**Special notes for your reviewer**:
NONE
**Release note**:
```release-note
kubelet: fix nil pointer dereference while enforce-node-allocatable flag is not config properly
```
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>.
Skip checking when failSwapOn=false
**What this PR does / why we need it**:
Skip checking when failSwapOn=false
**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**:
NONE
**Release note**:
```
NONE
```
Automatic merge from submit-queue (batch tested with PRs 66850, 66902, 66779, 66864, 66912). 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 methods to apimachinery to easy unit testing
When unit testing, you often want a selective scheme and codec factory. Rather than writing the vars and the init function and the error handling, you can simply do
`scheme, codecs := testing.SchemeForInstallOrDie(install.Install)`
@kubernetes/sig-api-machinery-misc
@sttts
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 65730, 66615, 66684, 66519, 66510). 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>.
kubelet: add image-gc low/high validation check
Currently, there is no protection against setting the high watermark <= the low watermark for image GC
This PR adds a validation rule for that.
@smarterclayton
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 feature gate checks inside IsCriticalPod
Currently `IsCriticalPod()` calls throughout the code are protected by `utilfeature.DefaultFeatureGate.Enabled(features.ExperimentalCriticalPodAnnotation)`.
However, with Pod Priority, this gate could be disabled which skips the priority check inside IsCriticalPod().
This PR moves the feature gate checking inside `IsCriticalPod()` and handles both situations properly.
@aveshagarwal @ravisantoshgudimetla @derekwaynecarr
/sig node
/sig scheduling
/king bug
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>.
delete unused events
**What this PR does / why we need it**:
events (HostNetworkNotSupported, UndefinedShaper) is unused since #47058
**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 66623, 66718). 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>.
cpumanager: validate topology in static policy
**What this PR does / why we need it**:
This patch adds a check for the static policy state validation. The check fails if the CPU topology obtained from cadvisor doesn't match with the current topology in the state file.
If the CPU topology has changed in a node, cpumanager static policy might try to assign non-present cores to containers.
For example in my test case, static policy had the default CPU set of `0-1,4-7`. Then kubelet was shut down and CPU 7 was offlined. After restarting the kubelet, CPU manager tries to assign the non-existent CPU 7 to containers which don't have exclusive allocations assigned to them:
Error response from daemon: Requested CPUs are not available - requested 0-1,4-7, available: 0-6)
This breaks the exclusivity, since the CPUs from the shared pool don't get assigned to non-exclusive containers, meaning that they can execute on the exclusive CPUs.
**Release note**:
```release-note
Added CPU Manager state validation in case of changed CPU topology.
```
Test the cases where the number of CPUs available in the system is
smaller or larger than the number of CPUs known in the state, which
should lead to a panic. This covers both CPU onlining and offlining. The
case where the number of CPUs matches is already covered by the
"non-corrupted state" test.
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 the` k8s.io/kubernetes/pkg/util/pointer` package to` k8s.io/utils/pointer`
**What this PR does / why we need it**:
Move `k8s.io/kubernetes/pkg/util/pointer` to `shared utils` directory, so that we can use it easily.
Close#66010 accidentally, and can't reopen it, so the same as #66010
**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
```
This patch adds a check for the static policy state validation. The
check fails if the CPU topology obtained from cadvisor doesn't match
with the current topology in the state file.
If the CPU topology has changed in a node, cpu manager static policy
might try to assign non-present cores to containers.
For example in my test case, static policy had the default CPU set of
0-1,4-7. Then kubelet was shut down and CPU 7 was offlined. After
restarting the kubelet, CPU manager tries to assign the non-existent CPU
7 to containers which don't have exclusive allocations assigned to them:
Error response from daemon: Requested CPUs are not available - requested 0-1,4-7, available: 0-6)
This breaks the exclusivity, since the CPUs from the shared pool don't
get assigned to non-exclusive containers, meaning that they can execute
on the exclusive CPUs.
the caching layer on endpoint is redundant.
Here are the 3 related objects in picture:
devicemanager <-> endpoint <-> plugin
Plugin is the source of truth for devices
and device health status.
devicemanager maintain healthyDevices,
unhealthyDevices, allocatedDevices based on updates
from plugin.
So there is no point for endpoint caching devices,
this patch is removing this caching layer on endpoint,
Also removing the Manager.Devices() since i didn't
find any caller of this other than test, i am adding a
notification channel to facilitate testing,
If we need to get all devices from manager in future,
it just need to return healthyDevices + unhealthyDevices,
we don't have to call endpoint after all.
This patch makes code more readable, data model been simplified.
Automatic merge from submit-queue (batch tested with PRs 66593, 66727, 66558). 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 the outdate comments in tryRegisterWithAPIServer
**What this PR does / why we need it**:
some judgement about ExternalID removed in #61877, so remove the outdate comments in tryRegisterWithAPIServer
**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 58755, 66414). 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>.
Use probe based plugin watcher mechanism in Device Manager
**What this PR does / why we need it**:
Uses this probe based utility in the device plugin manager.
**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#56944
**Notes For Reviewers**:
Changes are backward compatible and existing device plugins will continue to work. At the same time, any new plugins that has required support for probing model (Identity service implementation), will also work.
**Release note**
```release-note
Add support kubelet plugin watcher in device manager.
```
/sig node
/area hw-accelerators
/cc /cc @jiayingz @RenaudWasTaken @vishh @ScorpioCPH @sjenning @derekwaynecarr @jeremyeder @lichuqiang @tengqm @saad-ali @chakri-nelluri @ConnorDoyle
Automatic merge from submit-queue (batch tested with PRs 66665, 66707, 66596). 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 kubelet npe panic on device plugin return zero container
Signed-off-by: bingshen.wbs <bingshen.wbs@alibaba-inc.com>
**What this PR does / why we need it**:
Fix kubelet panic when device plugin return zero containers. Panic logs like follows:
```
Jul 17 12:50:24 iZwz9bqgzuo4i8qu435zk8Z kubelet[25815]: /workspace/anago-v1.10.4-beta.0.68+5ca598b4ba5abb/src/k8s.io/kubernetes/_output/dockerized/go/src/
k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/runtime/runtime.go:51
Jul 17 12:50:24 iZwz9bqgzuo4i8qu435zk8Z kubelet[25815]: /workspace/anago-v1.10.4-beta.0.68+5ca598b4ba5abb/src/k8s.io/kubernetes/_output/dockerized/go/src/
k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/runtime/runtime.go:65
Jul 17 12:50:24 iZwz9bqgzuo4i8qu435zk8Z kubelet[25815]: /workspace/anago-v1.10.4-beta.0.68+5ca598b4ba5abb/src/k8s.io/kubernetes/_output/dockerized/go/src/
k8s.io/kubernetes/vendor/k8s.io/apimachinery/pkg/util/runtime/runtime.go:72
Jul 17 12:50:24 iZwz9bqgzuo4i8qu435zk8Z kubelet[25815]: E0717 12:50:24.726856 25815 runtime.go:66] Observed a panic: "index out of range" (runtime error
: index out of range)
```
**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>.
Taint node when initializing node.
Signed-off-by: Da K. Ma <klaus1982.cn@gmail.com>
**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#63897
**Release note**:
```release-note
If `TaintNodesByCondition` enabled, taint node with `TaintNodeUnschedulable` when
initializing node to avoid race condition.
```
The --docker-disable-shared-pid flag has been deprecated since 1.10 and
has been superceded by ShareProcessNamespace in the pod API, which is
scheduled for beta in 1.12.
Automatic merge from submit-queue (batch tested with PRs 66152, 66406, 66218, 66278, 65660). 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 kubelet node status setters, add test coverage
This internal refactor moves the node status setters to a new package, explicitly injects dependencies to facilitate unit testing, and adds individual unit tests for the setters.
I gave each setter a distinct commit to facilitate review.
Non-goals:
- I intentionally excluded the class of setters that return a "modified" boolean, as I want to think more carefully about how to cleanly handle the behavior, and this PR is already rather large.
- I would like to clean up the status update control loops as well, but that belongs in a separate PR.
```release-note
NONE
```
When --cgroups-per-qos=false (default is true), kubelet sets pod
container management to podContainerManagerNoop implementation and
GetPodContainerName() returns '/' as cgroup parent (default cgroup root).
(1) In case of 'systemd' cgroup driver, '/' is invalid parent as
docker daemon expects '.slice' suffix and throws this error:
'cgroup-parent for systemd cgroup should be a valid slice named as \"xxx.slice\"'
(5fc12449d8/daemon/daemon_unix.go (L618))
'/' corresponds to '-.slice' (root slice) in systemd but I don't think
we want to assign root slice instead of runtime specific default value.
In case of docker runtime, this will be 'system.slice'
(e2593239d9/daemon/oci_linux.go (L698))
(2) In case of 'cgroupfs' cgroup driver, '/' is valid parent but I don't
think we want to assign root instead of runtime specific default value.
In case of docker runtime, this will be '/docker'
(e2593239d9/daemon/oci_linux.go (L695))
Current fix will not set the cgroup parent when --cgroups-per-qos is disabled.
Automatic merge from submit-queue (batch tested with PRs 65771, 65849). 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 a new conversion path to replace GenericConversionFunc
reflect.Call is very expensive. We currently use a switch block as part of AddGenericConversionFunc to avoid the bulk of top level a->b conversion for our primary types which is hand-written. Instead of having these be handwritten, we should generate them.
The pattern for generating them looks like:
```
scheme.AddConversionFunc(&v1.Type{}, &internal.Type{}, func(a, b interface{}, scope conversion.Scope) error {
return Convert_v1_Type_to_internal_Type(a.(*v1.Type), b.(*internal.Type), scope)
})
```
which matches AddDefaultObjectFunc (which proved out the approach last year). The
conversion machinery should then do a simple map lookup based on the incoming types and invoke the function. Like defaulting, it's up to the caller to match the types to arguments, which we do by generating this code. This bypasses reflect.Call and in the future allows Golang mid-stack inlining to optimize this code.
As part of this change I strengthened registration of custom functions to be generated instead of hand registered, and also strengthened error checking of the generator when it sees a manual conversion to error out. Since custom functions are automatically used by the generator, we don't really have a case for not registering the functions.
Once this is fully tested out, we can remove the reflection based path and the old registration methods, and all conversion will work from point to point methods (whether generated or custom).
Much of the need for the reflection path has been removed by changes to generation (to omit fields) and changes to Go (to make assigning equivalent structs easy).
```release-note
NONE
```
