internal

package
v1.5.0 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Jul 25, 2023 License: Apache-2.0 Imports: 49 Imported by: 0

Documentation

Overview

Package internal contains internal implementation details for the Kubeadm Control Plane.

Index

Constants

View Source 
const (
	// NodeBootstrapTokenAuthGroup specifies which group a Node Bootstrap Token should be authenticated in.
	NodeBootstrapTokenAuthGroup = "system:bootstrappers:kubeadm:default-node-token"

	// GetNodesClusterRoleName defines the name of the ClusterRole and ClusterRoleBinding to get nodes.
	GetNodesClusterRoleName = "kubeadm:get-nodes"

	// NodesGroup defines the well-known group for all nodes.
	NodesGroup = "system:nodes"

	// KubeletConfigMapRolePrefix defines base kubelet configuration ConfigMap role prefix.
	KubeletConfigMapRolePrefix = "kubeadm:"

	// KubeletConfigMapName defines base kubelet configuration ConfigMap name for kubeadm < 1.24.
	KubeletConfigMapName = "kubelet-config-%d.%d"

	// UnversionedKubeletConfigMapName defines base kubelet configuration ConfigMap for kubeadm >= 1.24.
	UnversionedKubeletConfigMapName = "kubelet-config"
)
View Source 
const (
	// KubeadmControlPlaneControllerName defines the controller used when creating clients.
	KubeadmControlPlaneControllerName = "kubeadm-controlplane-controller"
)

Variables

View Source 
var (

	// ErrControlPlaneMinNodes signals that a cluster doesn't meet the minimum required nodes
	// to remove an etcd member.
	ErrControlPlaneMinNodes = errors.New("cluster has fewer than 2 control plane nodes; removing an etcd member is not supported")
)

Functions

func ControlPlaneMachineLabelsForCluster added in v0.4.0 

func ControlPlaneMachineLabelsForCluster(kcp *controlplanev1.KubeadmControlPlane, clusterName string) map[string]string

ControlPlaneMachineLabelsForCluster returns a set of labels to add to a control plane machine for this specific cluster.

func ImageRepositoryFromClusterConfig added in v1.2.1 

func ImageRepositoryFromClusterConfig(clusterConfig *bootstrapv1.ClusterConfiguration, kubernetesVersion semver.Version) string

ImageRepositoryFromClusterConfig returns the image repository to use. It returns:

  • clusterConfig.ImageRepository if set.
  • else either k8s.gcr.io or registry.k8s.io depending on the default registry of the kubeadm binary of the given kubernetes version. This is only done for Kubernetes versions >= v1.22.0 and < v1.26.0 because in this version range the default registry was changed.

Note: Please see the following issue for more context: https://github.com/kubernetes-sigs/cluster-api/issues/7833 tl;dr is that the imageRepository must be in sync with the default registry of kubeadm. Otherwise kubeadm preflight checks will fail because kubeadm is trying to pull the CoreDNS image from the wrong repository (<registry>/coredns instead of <registry>/coredns/coredns).

func NeedsRollout added in v1.4.0 

func NeedsRollout(reconciliationTime, rolloutAfter *metav1.Time, rolloutBefore *controlplanev1.RolloutBefore, infraConfigs map[string]*unstructured.Unstructured, machineConfigs map[string]*bootstrapv1.KubeadmConfig, kcp *controlplanev1.KubeadmControlPlane, machine *clusterv1.Machine) (string, bool)

NeedsRollout checks if a Machine needs to be rolled out and returns the reason why.

Types

type ClusterStatus 

type ClusterStatus struct {
	// Nodes are a total count of nodes
	Nodes int32
	// ReadyNodes are the count of nodes that are reporting ready
	ReadyNodes int32
	// HasKubeadmConfig will be true if the kubeadm config map has been uploaded, false otherwise.
	HasKubeadmConfig bool
}

ClusterStatus holds stats information about the cluster.

type ControlPlane added in v0.3.2 

type ControlPlane struct {
	KCP      *controlplanev1.KubeadmControlPlane
	Cluster  *clusterv1.Cluster
	Machines collections.Machines

	// TODO: we should see if we can combine these with the Machine objects so we don't have all these separate lookups
	// See discussion on https://github.com/kubernetes-sigs/cluster-api/pull/3405
	KubeadmConfigs map[string]*bootstrapv1.KubeadmConfig
	InfraResources map[string]*unstructured.Unstructured
	// contains filtered or unexported fields
}

ControlPlane holds business logic around control planes. It should never need to connect to a service, that responsibility lies outside of this struct. Going forward we should be trying to add more logic to here and reduce the amount of logic in the reconciler.

func NewControlPlane added in v0.3.2 

func NewControlPlane(ctx context.Context, managementCluster ManagementCluster, client client.Client, cluster *clusterv1.Cluster, kcp *controlplanev1.KubeadmControlPlane, ownedMachines collections.Machines) (*ControlPlane, error)

NewControlPlane returns an instantiated ControlPlane.

func (*ControlPlane) FailureDomainWithMostMachines added in v0.3.3 

func (c *ControlPlane) FailureDomainWithMostMachines(machines collections.Machines) *string

FailureDomainWithMostMachines returns a fd which exists both in machines and control-plane machines and has the most control-plane machines on it.

func (*ControlPlane) FailureDomains added in v0.3.3 

func (c *ControlPlane) FailureDomains() clusterv1.FailureDomains

FailureDomains returns a slice of failure domain objects synced from the infrastructure provider into Cluster.Status.

func (*ControlPlane) GetKubeadmConfig added in v1.3.0 

func (c *ControlPlane) GetKubeadmConfig(machineName string) (*bootstrapv1.KubeadmConfig, bool)

GetKubeadmConfig returns the KubeadmConfig of a given machine.

func (*ControlPlane) GetWorkloadCluster added in v1.5.0 

func (c *ControlPlane) GetWorkloadCluster(ctx context.Context) (WorkloadCluster, error)

GetWorkloadCluster builds a cluster object. The cluster comes with an etcd client generator to connect to any etcd pod living on a managed machine.

func (*ControlPlane) HasDeletingMachine added in v0.3.2 

func (c *ControlPlane) HasDeletingMachine() bool

HasDeletingMachine returns true if any machine in the control plane is in the process of being deleted.

func (*ControlPlane) HasUnhealthyMachine added in v0.3.11 

func (c *ControlPlane) HasUnhealthyMachine() bool

HasUnhealthyMachine returns true if any machine in the control plane is marked as unhealthy by MHC.

func (*ControlPlane) HealthyMachines added in v0.3.11 

func (c *ControlPlane) HealthyMachines() collections.Machines

HealthyMachines returns the list of control plane machines not marked as unhealthy by MHC.

func (*ControlPlane) InitialControlPlaneConfig added in v0.3.2 

func (c *ControlPlane) InitialControlPlaneConfig() *bootstrapv1.KubeadmConfigSpec

InitialControlPlaneConfig returns a new KubeadmConfigSpec that is to be used for an initializing control plane.

func (*ControlPlane) InjectTestManagementCluster added in v1.5.0 

func (c *ControlPlane) InjectTestManagementCluster(managementCluster ManagementCluster)

InjectTestManagementCluster allows to inject a test ManagementCluster during tests. NOTE: This approach allows to keep the managementCluster field private, which will prevent people from using managementCluster.GetWorkloadCluster because it creates a new instance of WorkloadCluster at every call. People instead should use ControlPlane.GetWorkloadCluster that creates only a single instance of WorkloadCluster for each reconcile.

func (*ControlPlane) IsEtcdManaged added in v0.3.9 

func (c *ControlPlane) IsEtcdManaged() bool

IsEtcdManaged returns true if the control plane relies on a managed etcd.

func (*ControlPlane) JoinControlPlaneConfig added in v0.3.2 

func (c *ControlPlane) JoinControlPlaneConfig() *bootstrapv1.KubeadmConfigSpec

JoinControlPlaneConfig returns a new KubeadmConfigSpec that is to be used for joining control planes.

func (*ControlPlane) MachineInFailureDomainWithMostMachines added in v0.3.4 

func (c *ControlPlane) MachineInFailureDomainWithMostMachines(machines collections.Machines) (*clusterv1.Machine, error)

MachineInFailureDomainWithMostMachines returns the first matching failure domain with machines that has the most control-plane machines on it.

func (*ControlPlane) MachineWithDeleteAnnotation added in v0.3.13 

func (c *ControlPlane) MachineWithDeleteAnnotation(machines collections.Machines) collections.Machines

MachineWithDeleteAnnotation returns a machine that has been annotated with DeleteMachineAnnotation key.

func (*ControlPlane) MachinesNeedingRollout added in v0.3.8 

func (c *ControlPlane) MachinesNeedingRollout() (collections.Machines, map[string]string)

MachinesNeedingRollout return a list of machines that need to be rolled out.

func (*ControlPlane) NextFailureDomainForScaleUp added in v0.3.8 

func (c *ControlPlane) NextFailureDomainForScaleUp() *string

NextFailureDomainForScaleUp returns the failure domain with the fewest number of up-to-date machines.

func (*ControlPlane) PatchMachines added in v0.3.11 

func (c *ControlPlane) PatchMachines(ctx context.Context) error

PatchMachines patches all the machines conditions.

func (*ControlPlane) SetPatchHelpers added in v1.4.0 

func (c *ControlPlane) SetPatchHelpers(patchHelpers map[string]*patch.Helper)

SetPatchHelpers updates the patch helpers.

func (*ControlPlane) UnhealthyMachines added in v0.3.11 

func (c *ControlPlane) UnhealthyMachines() collections.Machines

UnhealthyMachines returns the list of control plane machines marked as unhealthy by MHC.

func (*ControlPlane) UpToDateMachines added in v0.3.8 

func (c *ControlPlane) UpToDateMachines() collections.Machines

UpToDateMachines returns the machines that are up to date with the control plane's configuration and therefore do not require rollout.

type CoreDNSMigrator 

type CoreDNSMigrator struct{}

CoreDNSMigrator is a shim that can be used to migrate CoreDNS files from one version to another.

func (*CoreDNSMigrator) Migrate 

func (c *CoreDNSMigrator) Migrate(fromCoreDNSVersion, toCoreDNSVersion, corefile string, deprecations bool) (string, error)

Migrate calls the CoreDNS migration library to migrate a corefile.

type EtcdClientGenerator added in v0.4.0 

type EtcdClientGenerator struct {
	// contains filtered or unexported fields
}

EtcdClientGenerator generates etcd clients that connect to specific etcd members on particular control plane nodes.

func NewEtcdClientGenerator added in v0.4.0 

func NewEtcdClientGenerator(restConfig *rest.Config, tlsConfig *tls.Config, etcdDialTimeout, etcdCallTimeout time.Duration) *EtcdClientGenerator

NewEtcdClientGenerator returns a new etcdClientGenerator instance.

type EtcdMemberStatus added in v0.3.11 

type EtcdMemberStatus struct {
	Name       string
	Responsive bool
}

EtcdMemberStatus contains status information for a single etcd member.

type Management 

type Management struct {
	Client              client.Reader
	SecretCachingClient client.Reader
	Tracker             *remote.ClusterCacheTracker
	EtcdDialTimeout     time.Duration
	EtcdCallTimeout     time.Duration
}

Management holds operations on the management cluster.

func (*Management) Get added in v0.3.7 

func (m *Management) Get(ctx context.Context, key client.ObjectKey, obj client.Object, opts ...client.GetOption) error

Get implements client.Reader.

func (*Management) GetMachinePoolsForCluster added in v0.4.0 

func (m *Management) GetMachinePoolsForCluster(ctx context.Context, cluster *clusterv1.Cluster) (*expv1.MachinePoolList, error)

GetMachinePoolsForCluster returns a list of machine pools owned by the cluster.

func (*Management) GetMachinesForCluster 

func (m *Management) GetMachinesForCluster(ctx context.Context, cluster *clusterv1.Cluster, filters ...collections.Func) (collections.Machines, error)

GetMachinesForCluster returns a list of machines that can be filtered or not. If no filter is supplied then all machines associated with the target cluster are returned.

func (*Management) GetWorkloadCluster 

func (m *Management) GetWorkloadCluster(ctx context.Context, clusterKey client.ObjectKey) (WorkloadCluster, error)

GetWorkloadCluster builds a cluster object. The cluster comes with an etcd client generator to connect to any etcd pod living on a managed machine.

func (*Management) List added in v0.3.7 

func (m *Management) List(ctx context.Context, list client.ObjectList, opts ...client.ListOption) error

List implements client.Reader.

type ManagementCluster 

type ManagementCluster interface {
	client.Reader

	GetMachinesForCluster(ctx context.Context, cluster *clusterv1.Cluster, filters ...collections.Func) (collections.Machines, error)
	GetMachinePoolsForCluster(ctx context.Context, cluster *clusterv1.Cluster) (*expv1.MachinePoolList, error)
	GetWorkloadCluster(ctx context.Context, clusterKey client.ObjectKey) (WorkloadCluster, error)
}

ManagementCluster defines all behaviors necessary for something to function as a management cluster.

type RemoteClusterConnectionError added in v0.3.7 

type RemoteClusterConnectionError struct {
	Name string
	Err  error
}

RemoteClusterConnectionError represents a failure to connect to a remote cluster.

func (*RemoteClusterConnectionError) Error added in v0.3.7 

func (e *RemoteClusterConnectionError) Error() string

Error satisfies the error interface.

func (*RemoteClusterConnectionError) Unwrap added in v0.3.7 

func (e *RemoteClusterConnectionError) Unwrap() error

Unwrap satisfies the unwrap error inteface.

type Workload 

type Workload struct {
	Client          ctrlclient.Client
	CoreDNSMigrator coreDNSMigrator
	// contains filtered or unexported fields
}

Workload defines operations on workload clusters.

func (*Workload) AllowBootstrapTokensToGetNodes added in v0.3.5 

func (w *Workload) AllowBootstrapTokensToGetNodes(ctx context.Context) error

AllowBootstrapTokensToGetNodes creates RBAC rules to allow Node Bootstrap Tokens to list nodes.

func (*Workload) ClusterStatus 

func (w *Workload) ClusterStatus(ctx context.Context) (ClusterStatus, error)

ClusterStatus returns the status of the cluster.

func (*Workload) EnsureResource added in v0.3.5 

func (w *Workload) EnsureResource(ctx context.Context, obj client.Object) error

EnsureResource creates a resoutce if the target resource doesn't exist. If the resource exists already, this function will ignore the resource instead.

func (*Workload) EtcdMembers added in v0.3.11 

func (w *Workload) EtcdMembers(ctx context.Context) ([]string, error)

EtcdMembers returns the current set of members in an etcd cluster.

NOTE: This methods uses control plane machines/nodes only to get in contact with etcd, but then it relies on etcd as ultimate source of truth for the list of members. This is intended to allow informed decisions on actions impacting etcd quorum.

func (*Workload) ForwardEtcdLeadership added in v0.3.2 

func (w *Workload) ForwardEtcdLeadership(ctx context.Context, machine *clusterv1.Machine, leaderCandidate *clusterv1.Machine) error

ForwardEtcdLeadership forwards etcd leadership to the first follower.

func (*Workload) GetAPIServerCertificateExpiry added in v1.3.0 

func (w *Workload) GetAPIServerCertificateExpiry(ctx context.Context, kubeadmConfig *bootstrapv1.KubeadmConfig, nodeName string) (*time.Time, error)

GetAPIServerCertificateExpiry returns the certificate expiry of the apiserver on the given node.

func (*Workload) ReconcileEtcdMembers added in v0.3.4 

func (w *Workload) ReconcileEtcdMembers(ctx context.Context, nodeNames []string, version semver.Version) ([]string, error)

ReconcileEtcdMembers iterates over all etcd members and finds members that do not have corresponding nodes. If there are any such members, it deletes them from etcd and removes their nodes from the kubeadm configmap so that kubeadm does not run etcd health checks on them.

func (*Workload) ReconcileKubeletRBACBinding 

func (w *Workload) ReconcileKubeletRBACBinding(ctx context.Context, version semver.Version) error

ReconcileKubeletRBACBinding will create a RoleBinding for the new kubelet version during upgrades. If the role binding already exists this function is a no-op.

func (*Workload) ReconcileKubeletRBACRole 

func (w *Workload) ReconcileKubeletRBACRole(ctx context.Context, version semver.Version) error

ReconcileKubeletRBACRole will create a Role for the new kubelet version during upgrades. If the role already exists this function is a no-op.

func (*Workload) RemoveEtcdMemberForMachine 

func (w *Workload) RemoveEtcdMemberForMachine(ctx context.Context, machine *clusterv1.Machine) error

RemoveEtcdMemberForMachine removes the etcd member from the target cluster's etcd cluster. Removing the last remaining member of the cluster is not supported.

func (*Workload) RemoveMachineFromKubeadmConfigMap 

func (w *Workload) RemoveMachineFromKubeadmConfigMap(ctx context.Context, machine *clusterv1.Machine, version semver.Version) error

RemoveMachineFromKubeadmConfigMap removes the entry for the machine from the kubeadm configmap.

func (*Workload) RemoveNodeFromKubeadmConfigMap added in v0.3.4 

func (w *Workload) RemoveNodeFromKubeadmConfigMap(ctx context.Context, name string, v semver.Version) error

RemoveNodeFromKubeadmConfigMap removes the entry for the node from the kubeadm configmap.

func (*Workload) UpdateAPIServerInKubeadmConfigMap added in v0.3.13 

func (w *Workload) UpdateAPIServerInKubeadmConfigMap(ctx context.Context, apiServer bootstrapv1.APIServer, version semver.Version) error

UpdateAPIServerInKubeadmConfigMap updates api server configuration in kubeadm config map.

func (*Workload) UpdateControllerManagerInKubeadmConfigMap added in v0.3.13 

func (w *Workload) UpdateControllerManagerInKubeadmConfigMap(ctx context.Context, controllerManager bootstrapv1.ControlPlaneComponent, version semver.Version) error

UpdateControllerManagerInKubeadmConfigMap updates controller manager configuration in kubeadm config map.

func (*Workload) UpdateCoreDNS 

func (w *Workload) UpdateCoreDNS(ctx context.Context, kcp *controlplanev1.KubeadmControlPlane, version semver.Version) error

UpdateCoreDNS updates the kubeadm configmap, coredns corefile and coredns deployment.

func (*Workload) UpdateEtcdConditions added in v0.3.11 

func (w *Workload) UpdateEtcdConditions(ctx context.Context, controlPlane *ControlPlane)

UpdateEtcdConditions is responsible for updating machine conditions reflecting the status of all the etcd members. This operation is best effort, in the sense that in case of problems in retrieving member status, it sets the condition to Unknown state without returning any error.

func (*Workload) UpdateEtcdExtraArgsInKubeadmConfigMap added in v0.4.1 

func (w *Workload) UpdateEtcdExtraArgsInKubeadmConfigMap(ctx context.Context, extraArgs map[string]string, version semver.Version) error

UpdateEtcdExtraArgsInKubeadmConfigMap sets extraArgs in the kubeadm config map.

func (*Workload) UpdateEtcdVersionInKubeadmConfigMap 

func (w *Workload) UpdateEtcdVersionInKubeadmConfigMap(ctx context.Context, imageRepository, imageTag string, version semver.Version) error

UpdateEtcdVersionInKubeadmConfigMap sets the imageRepository or the imageTag or both in the kubeadm config map.

func (*Workload) UpdateImageRepositoryInKubeadmConfigMap added in v0.3.3 

func (w *Workload) UpdateImageRepositoryInKubeadmConfigMap(ctx context.Context, imageRepository string, version semver.Version) error

UpdateImageRepositoryInKubeadmConfigMap updates the image repository in the kubeadm config map.

func (*Workload) UpdateKubeProxyImageInfo 

func (w *Workload) UpdateKubeProxyImageInfo(ctx context.Context, kcp *controlplanev1.KubeadmControlPlane, version semver.Version) error

UpdateKubeProxyImageInfo updates kube-proxy image in the kube-proxy DaemonSet.

func (*Workload) UpdateKubeletConfigMap 

func (w *Workload) UpdateKubeletConfigMap(ctx context.Context, version semver.Version) error

UpdateKubeletConfigMap will create a new kubelet-config-1.x config map for a new version of the kubelet. This is a necessary process for upgrades.

func (*Workload) UpdateKubernetesVersionInKubeadmConfigMap 

func (w *Workload) UpdateKubernetesVersionInKubeadmConfigMap(ctx context.Context, version semver.Version) error

UpdateKubernetesVersionInKubeadmConfigMap updates the kubernetes version in the kubeadm config map.

func (*Workload) UpdateSchedulerInKubeadmConfigMap added in v0.3.13 

func (w *Workload) UpdateSchedulerInKubeadmConfigMap(ctx context.Context, scheduler bootstrapv1.ControlPlaneComponent, version semver.Version) error

UpdateSchedulerInKubeadmConfigMap updates scheduler configuration in kubeadm config map.

func (*Workload) UpdateStaticPodConditions added in v0.3.11 

func (w *Workload) UpdateStaticPodConditions(ctx context.Context, controlPlane *ControlPlane)

UpdateStaticPodConditions is responsible for updating machine conditions reflecting the status of all the control plane components running in a static pod generated by kubeadm. This operation is best effort, in the sense that in case of problems in retrieving the pod status, it sets the condition to Unknown state without returning any error.

type WorkloadCluster 

type WorkloadCluster interface {
	// Basic health and status checks.
	ClusterStatus(ctx context.Context) (ClusterStatus, error)
	UpdateStaticPodConditions(ctx context.Context, controlPlane *ControlPlane)
	UpdateEtcdConditions(ctx context.Context, controlPlane *ControlPlane)
	EtcdMembers(ctx context.Context) ([]string, error)
	GetAPIServerCertificateExpiry(ctx context.Context, kubeadmConfig *bootstrapv1.KubeadmConfig, nodeName string) (*time.Time, error)

	// Upgrade related tasks.
	ReconcileKubeletRBACBinding(ctx context.Context, version semver.Version) error
	ReconcileKubeletRBACRole(ctx context.Context, version semver.Version) error
	UpdateKubernetesVersionInKubeadmConfigMap(ctx context.Context, version semver.Version) error
	UpdateImageRepositoryInKubeadmConfigMap(ctx context.Context, imageRepository string, version semver.Version) error
	UpdateEtcdVersionInKubeadmConfigMap(ctx context.Context, imageRepository, imageTag string, version semver.Version) error
	UpdateEtcdExtraArgsInKubeadmConfigMap(ctx context.Context, extraArgs map[string]string, version semver.Version) error
	UpdateAPIServerInKubeadmConfigMap(ctx context.Context, apiServer bootstrapv1.APIServer, version semver.Version) error
	UpdateControllerManagerInKubeadmConfigMap(ctx context.Context, controllerManager bootstrapv1.ControlPlaneComponent, version semver.Version) error
	UpdateSchedulerInKubeadmConfigMap(ctx context.Context, scheduler bootstrapv1.ControlPlaneComponent, version semver.Version) error
	UpdateKubeletConfigMap(ctx context.Context, version semver.Version) error
	UpdateKubeProxyImageInfo(ctx context.Context, kcp *controlplanev1.KubeadmControlPlane, version semver.Version) error
	UpdateCoreDNS(ctx context.Context, kcp *controlplanev1.KubeadmControlPlane, version semver.Version) error
	RemoveEtcdMemberForMachine(ctx context.Context, machine *clusterv1.Machine) error
	RemoveMachineFromKubeadmConfigMap(ctx context.Context, machine *clusterv1.Machine, version semver.Version) error
	RemoveNodeFromKubeadmConfigMap(ctx context.Context, nodeName string, version semver.Version) error
	ForwardEtcdLeadership(ctx context.Context, machine *clusterv1.Machine, leaderCandidate *clusterv1.Machine) error
	AllowBootstrapTokensToGetNodes(ctx context.Context) error

	// State recovery tasks.
	ReconcileEtcdMembers(ctx context.Context, nodeNames []string, version semver.Version) ([]string, error)
}

WorkloadCluster defines all behaviors necessary to upgrade kubernetes on a workload cluster

TODO: Add a detailed description to each of these method definitions.

Source Files

View all Source files

Directories

Path Synopsis
Package controllers implements the Kubeadm controllers.
 Package controllers implements the Kubeadm controllers.
Package etcd provides a connection to an etcd member.
 Package etcd provides a connection to an etcd member.
fake
Package fake implements testing fakes.
 Package fake implements testing fakes.
util
Package util implements etcd utility functions.
 Package util implements etcd utility functions.
Package proxy implements kubeadm proxy functionality.
 Package proxy implements kubeadm proxy functionality.
Package webhooks provides the validating webhook for KubeadmControlPlane scale subresource.
 Package webhooks provides the validating webhook for KubeadmControlPlane scale subresource.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.