Automatic merge from submit-queue (batch tested with PRs 63881, 64046, 63409, 63402, 63221). 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 responds to ConfigMap mutations for dynamic Kubelet config
This PR makes dynamic Kubelet config easier to reason about by leaving less room for silent skew scenarios. The new behavior is as follows:
- ConfigMap does not exist: Kubelet reports error status due to missing source
- ConfigMap is created: Kubelet starts using it
- ConfigMap is updated: Kubelet respects the update (but we discourage this pattern, in favor of incrementally migrating to a new ConfigMap)
- ConfigMap is deleted: Kubelet keeps using the config (non-disruptive), but reports error status due to missing source
- ConfigMap is recreated: Kubelet respects any updates (but, again, we discourage this pattern)
This PR also makes a small change to the config checkpoint file tree structure, because ResourceVersion is now taken into account when saving checkpoints. The new structure is as follows:
```
- dir named by --dynamic-config-dir (root for managing dynamic config)
| - meta
| - assigned (encoded kubeletconfig/v1beta1.SerializedNodeConfigSource object, indicating the assigned config)
| - last-known-good (encoded kubeletconfig/v1beta1.SerializedNodeConfigSource object, indicating the last-known-good config)
| - checkpoints
| - uid1 (dir for versions of object identified by uid1)
| - resourceVersion1 (dir for unpacked files from resourceVersion1)
| - ...
| - ...
```
fixes: #61643
```release-note
The dynamic Kubelet config feature will now update config in the event of a ConfigMap mutation, which reduces the chance for silent config skew. Only name, namespace, and kubeletConfigKey may now be set in Node.Spec.ConfigSource.ConfigMap. The least disruptive pattern for config management is still to create a new ConfigMap and incrementally roll out a new Node.Spec.ConfigSource.
```
Automatic merge from submit-queue (batch tested with PRs 63881, 64046, 63409, 63402, 63221). 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 config: Validate new config against future feature gates
This fixes an issue with KubeletConfiguration validation, where the
feature gates set by the new config were not taken into account.
Also fixes a validation issue with dynamic Kubelet config, where flag
precedence was not enforced prior to dynamic config validation in the
controller; this prevented rejection of dynamic configs that don't merge
well with values set via legacy flags.
Fixes#63305
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 63881, 64046, 63409, 63402, 63221). 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>.
Disable the public cadvisor port by default
**What this PR does / why we need it**:
Quoting @tallclair in #56523:
> We should add the deprecation warning in 1.10 along with a release note, but not change the default. The notes should warn that the default will change in 1.11. We disable the flag by default in 1.11, and remove it entirely in 1.12 or 1.13.
> If you currently depend on the UI or the API, speak up! Going forward, the recommended way of taking advantage of those features will be to run cAdvisor as a DaemonSet.
Disabling the publicly-available cAdvisor port is beneficial for security, as you might not want to expose the UI with lots of information about what your system is doing. We already did this for all kubeadm deployments in v1.7, and haven't recieved any issues for that. This should be okay to do at this stage, as this flag was deprecated in v1.10. Given we need to support this flag for one more release (v1.11), it makes perfect sense to instead switch it off in preparation for v1.12 when we can delete it (see the [deprecation policy](https://kubernetes.io/docs/reference/deprecation-policy/#deprecating-a-flag-or-cli))
**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 #56523
**Special notes for your reviewer**:
I removed the e2e test that expects cAdvisor to be running, as we don't expect it to be anymore.
**Release note**:
```release-note
[action required] The formerly publicly-available cAdvisor web UI that the kubelet ran on port 4194 by default is now turned off by default. The flag configuring what port to run this UI on `--cadvisor-port` was deprecated in v1.10. Now the default is `--cadvisor-port=0`, in other words, to not run the web server. The recommended way to run cAdvisor if you still need it, is via a DaemonSet. The `--cadvisor-port` will be removed in v1.12
```
cc @kubernetes/sig-cluster-lifecycle-pr-reviews @kubernetes/sig-auth-pr-reviews @kubernetes/sig-node-pr-reviews
This fixes an issue with KubeletConfiguration validation, where the
feature gates set by the new config were not taken into account.
Also fixes a validation issue with dynamic Kubelet config, where flag
precedence was not enforced prior to dynamic config validation in the
controller; this prevented rejection of dynamic configs that don't merge
well with values set via legacy flags.
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 to a structured status for dynamic kubelet config
This PR updates dynamic Kubelet config to use a structured status, rather than a node condition. This makes the status machine-readable, and thus more useful for config orchestration.
Fixes: #56896
```release-note
The status of dynamic Kubelet config is now reported via Node.Status.Config, rather than the KubeletConfigOk node condition.
```
Updates dynamic Kubelet config to use a structured status, rather than a
node condition. This makes the status machine-readable, and thus more
useful for config orchestration.
Fixes: #56896
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. 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 necessary explanation for container log rotation.
**What this PR does / why we need it**:
#59898
> Container log manager only starts when the container runtime is remote (not docker), because we can't implement ReopenContainerLog for docker.
**Release note**:
```release-note
NONE
```
This enables a smooth transition to PSP. Today, users would have to
manually set --allow-privileged to true before transitioning to PSP,
which isn't a smooth deprecation path for the flag (we want people
to *stop* setting it). This PR makes the default behavior isomorphic
with what will happen after the flag is removed.
Defaulting --allow-privileged to true should be safe, because it simply
allows a superset of Pods to run (all workloads continue to work).
WRT https://github.com/kubernetes/kubernetes/issues/58010#issuecomment-383264473
the --allow-privileged flag is effectively useless for security, so this
shouldn't be a concern from that perspective.
I also bumped the deprecation timeline in the comment to 1.13.0, so that
we give people the full period of time to stop setting
--allow-privileged, now that the behavior makes it possible to do so.
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>.
check for commands in kubelet command line
**What this PR does / why we need it**:
Kubelet doesn't support subcommands and silently ignores them.
This causes a lot of confusion among users, for example:
Running 'kubelet logs' can produce this output:
```
I0206 16:20:40.010949 5712 feature_gate.go:220] feature gates: &{{} map[]}
I0206 16:20:40.011054 5712 controller.go:114] kubelet config controller: starting controller
...
error: failed to run Kubelet: failed to create kubelet:
misconfiguration: kubelet cgroup driver: "cgroupfs" is different from
docker cgroup driver: "systemd"
```
The output in this case looks like a valid log output. The error is not
caused by incorrectly configured cgroup driver though. It's caused by
the fact that kubelet ignores 'logs' subcommand, so 'kubelet logs' is
essentially the same as 'kubelet': it runs kubelet with the default set
of parameters, including default cgroup driver.
**The fix**:
Added check to prevent kubelet from running if user specifies
subcommands in the command line.
**Which issue(s) this PR fixes**:
Fixes: kubeadm issue 639
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 62982, 63075, 63067, 62877, 63141). 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 warnings on using pod-infra-container-image for remote container runtime
**What this PR does / why we need it**:
We should warn on using `--pod-infra-container-image` to avoid confusions, when users are using remote container runtime.
**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 #55676,#62388,#62732
**Special notes for your reviewer**:
/cc @kubernetes/sig-node-pr-reviews
**Release note**:
```release-note
add warnings on using pod-infra-container-image for remote container runtime
```
In our CI, we run kubekins image for most of the jobs. This is a
debian image with upstart and does not enable systemd. So we should:
* Bailout if any binary is missing other than systemd-run.
* SupportsSystemd should check the binary path to correctly
identify if the systemd-run is present or not
* Pass the errors back to the callers so kubelet is forced to
fail early when there is a problem. We currently assume
that all binaries are in the root directory by default which
is wrong.
Automatic merge from submit-queue (batch tested with PRs 61962, 58972, 62509, 62606). 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: move QOSReserved from experimental to alpha feature gate
Fixes https://github.com/kubernetes/kubernetes/issues/61665
**Release note**:
```release-note
The --experimental-qos-reserve kubelet flags is replaced by the alpha level --qos-reserved flag or QOSReserved field in the kubeletconfig and requires the QOSReserved feature gate to be enabled.
```
/sig node
/assign @derekwaynecarr
/cc @mtaufen
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>.
Show help for deprecated Kubelet flags
We recently deprecated a bunch of Kubelet flags, which caused them to disappear from `--help` output. This PR unhides these flags, so that the deprecation notice is clearly visible in `--help`.
Fixes: #62009
```release-note
NONE
```
/cc @eparis
This PR extends the client-side startup scripts to provision a Kubelet
config file instead of legacy flags. This PR also extends the
master/node init scripts to install this config file from the GCE
metadata server, and provide the --config argument to the Kubelet.
With CRI, kubelet no longer sets up networking for the pods. The
dockershim package is the rightful owner and the only user of the
newtork package. This change moves the package into dockershim to make
the distinction obvious, and untangles the codebase.
The`network/dns`is kept in the original package since it is only used by
kubelet.
The code was added to support rktnetes and non-CRI docker integrations.
These legacy integrations have already been removed from the codebase.
This change removes the compatibility code existing soley for the
legacy integrations.
Kubelet doesn't support subcommands and silently ignores them.
This causes a lot of confusion among users, for example:
Running 'kubelet logs' can produce this output:
I0206 16:20:40.010949 5712 feature_gate.go:220] feature gates: &{{} map[]}
I0206 16:20:40.011054 5712 controller.go:114] kubelet config controller: starting controller
...
error: failed to run Kubelet: failed to create kubelet:
misconfiguration: kubelet cgroup driver: "cgroupfs" is different from
docker cgroup driver: "systemd"
The output in this case looks like a valid log output. The error is not
caused by incorrectly configured cgroup driver though. It's caused by
the fact that kubelet ignores 'logs' subcommand, so 'kubelet logs' is
essentially the same as 'kubelet': it runs kubelet with the default set
of parameters, including default cgroup driver.
Added check to prevent kubelet from running if user specifies
subcommands in the command line.
Fixes: #639
rktnetes is scheduled to be deprecated in 1.10 (#53601). According to
the deprecation policy for beta CLI and flags, we can remove the feature
in 1.11.
Fixes#58721
Automatic merge from submit-queue (batch tested with PRs 59740, 59728, 60080, 60086, 58714). 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: make --cni-bin-dir accept a comma-separated list of CNI plugin directories
Allow CNI-related network plugin drivers (kubenet, cni) to search a list of
directories for plugin binaries instead of just one. This allows using an
administrator-provided path and fallbacks to others (like the previous default
of /opt/cni/bin) for backwards compatibility.
```release-note
kubelet's --cni-bin-dir option now accepts multiple comma-separated CNI binary directory paths, which are search for CNI plugins in the given order.
```
@kubernetes/rh-networking @kubernetes/sig-network-misc @freehan @pecameron @rajatchopra
Automatic merge from submit-queue (batch tested with PRs 51423, 53880). 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>.
Disable ImageGC when high threshold is set to 100
**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#51268
**Special notes for your reviewer**:
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 60737, 60739, 61080, 60968, 60951). 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>.
Update Kubelet command option description for IPv6
**What this PR does / why we need it**:
The restriction for a /66 cidr was removed in PR #60089.
Removing this reference from the command options description.
**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#60734
**Special notes for your reviewer**:
**Release note**:
```release-note
NONE
```
This patch adds support for kubernetes to integrate
with Windows SCM.
As a first step both `kubelet` and `kube-proxy` can be registered as a service.
To create the service:
PS > sc.exe create <component_name> binPath= "<path_to_binary> --service <other_args>"
CMD > sc create <component_name> binPath= "<path_to_binary> --service <other_args>"
Please note that if the arguments contain spaces, it must be escaped.
Example:
PS > sc.exe create kubelet binPath= "C:\kubelet.exe --service --hostname-override 'minion' <other_args>"
CMD > sc create kubelet binPath= "C:\kubelet.exe --service --hostname-override 'minion' <other_args>"
Example to start the service:
PS > Start-Service kubelet; Start-Service kube-proxy
CMD > net start kubelet && net start kube-proxy
Example to stop the service:
PS > Stop-Service kubelet (-Force); Stop-Service kube-proxy (-Force)
CMD > net stop kubelet && net stop kube-proxy
Example to query the service:
PS > Get-Service kubelet; Get-Service kube-proxy;
CMD > sc.exe queryex kubelet && sc qc kubelet && sc.exe queryex kube-proxy && sc.exe qc kube-proxy
Signed-off-by: Alin Balutoiu <abalutoiu@cloudbasesolutions.com>
Signed-off-by: Alin Gabriel Serdean <aserdean@ovn.org>
Co-authored-by: Alin Gabriel Serdean <aserdean@ovn.org>
Allow CNI-related network plugin drivers (kubenet, cni) to search a list of
directories for plugin binaries instead of just one. This allows using an
administrator-provided path and fallbacks to others (like the previous default
of /opt/cni/bin) for backwards compatibility.
It's only used for the test code and after talking with Rajat, the
vendor stuff was never really used anyway. So convert the vendor
code into a plain array of plugin binary search paths, which is all
the vendor code was doing anyway.
Automatic merge from submit-queue (batch tested with PRs 60324, 60269, 59771, 60314, 59941). 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>.
expunge the word 'manifest' from Kubelet's config API
The word 'manifest' technically refers to a container-group specification
that predated the Pod abstraction. We should avoid using this legacy
terminology where possible. Fortunately, the Kubelet's config API will
be beta in 1.10 for the first time, so we still had the chance to make
this change.
I left the flags alone, since they're deprecated anyway.
I changed a few var names in files I touched too, but this PR is the
just the first shot, not the whole campaign
(`git grep -i manifest | wc -l -> 1248`).
```release-note
Some field names in the Kubelet's now v1beta1 config API differ from the v1alpha1 API: PodManifestPath is renamed to PodPath, ManifestURL is renamed to PodURL, ManifestURLHeader is renamed to PodURLHeader.
```
Automatic merge from submit-queue (batch tested with PRs 60302, 57921, 59042, 60126, 59561). 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>.
Removing Flexvolume feature tag in e2e tests because Flexvolume is now GA
**What this PR does / why we need it**: Flexvolume e2e tests should be run in serial test suite after the feature was marked GA.
**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#57466
/release-note-none
/sig storage
/cc @wongma7
The word 'manifest' technically refers to a container-group specification
that predated the Pod abstraction. We should avoid using this legacy
terminology where possible. Fortunately, the Kubelet's config API will
be beta in 1.10 for the first time, so we still had the chance to make
this change.
I left the flags alone, since they're deprecated anyway.
I changed a few var names in files I touched too, but this PR is the
just the first shot, not the whole campaign
(`git grep -i manifest | wc -l -> 1248`).
Automatic merge from submit-queue (batch tested with PRs 60148, 60022, 59125, 60068, 60154). 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>.
Deprecate KubeletConfiguration flags
This is a follow-up to #53833 to deprecate the old flags in favor of the
config file.
```release-note
Flags that can be set via the Kubelet's --config file are now deprecated in favor of the file.
```
Lucas Käldström