Documentation
¶
Index ¶
- Constants
- func LocationPrioritizedComparison(currentLeaderRevision string, l *LeaderElection) bool
- type LeaderElection
- func NewLeaderElection(namespace, name, electionID, revision string, client kube.Client) *LeaderElection
- func NewLeaderElectionMulticluster(namespace, name, electionID, revision string, remote bool, client kube.Client) *LeaderElection
- func NewPerRevisionLeaderElection(namespace, name, electionID, revision string, client kube.Client) *LeaderElection
Constants ¶
View Source
const ( NamespaceController = "istio-namespace-controller-election" ServiceExportController = "istio-serviceexport-controller-election" // This holds the legacy name to not conflict with older control plane deployments which are just // doing the ingress syncing. IngressController = "istio-leader" // GatewayStatusController controls the status of gateway.networking.k8s.io objects. For the v1alpha1 // this was formally "istio-gateway-leader"; because they are a different API group we need a different // election to ensure we do not only handle one or the other. GatewayStatusController = "istio-gateway-status-leader" StatusController = "istio-status-leader" AnalyzeController = "istio-analyze-leader" // GatewayDeploymentController controls translating Kubernetes Gateway objects into various derived // resources (Service, Deployment, etc). // Unlike other types which use ConfigMaps, we use a Lease here. This is because: // * Others use configmap for backwards compatibility // * This type is per-revision, so it is higher cost. Leases are cheaper // * Other types use "prioritized leader election", which isn't implemented for Lease GatewayDeploymentController = "istio-gateway-deployment" NodeUntaintController = "istio-node-untaint" )
Various locks used throughout the code
Variables ¶
This section is empty.
Functions ¶
func LocationPrioritizedComparison ¶
func LocationPrioritizedComparison(currentLeaderRevision string, l *LeaderElection) bool
Types ¶
type LeaderElection ¶
type LeaderElection struct {
// contains filtered or unexported fields
}
func NewLeaderElection ¶
func NewLeaderElection(namespace, name, electionID, revision string, client kube.Client) *LeaderElection
NewLeaderElection creates a leader election instance with the provided ID. This follows standard Kubernetes elections, with one difference: the "default" revision will steal the lock from other revisions.
func NewLeaderElectionMulticluster ¶
func NewLeaderElectionMulticluster(namespace, name, electionID, revision string, remote bool, client kube.Client) *LeaderElection
func NewPerRevisionLeaderElection ¶
func NewPerRevisionLeaderElection(namespace, name, electionID, revision string, client kube.Client) *LeaderElection
NewPerRevisionLeaderElection creates a *per revision* leader election. This means there will be one leader for each revision.
func (*LeaderElection) AddRunFunction ¶
func (l *LeaderElection) AddRunFunction(f func(stop <-chan struct{})) *LeaderElection
AddRunFunction registers a function to run when we are the leader. These will be run asynchronously. To avoid running when not a leader, functions should respect the stop channel.
func (*LeaderElection) Run ¶
func (l *LeaderElection) Run(stop <-chan struct{})
Run will start leader election, calling all runFns when we become the leader. If leader election is disabled, it skips straight to the runFns.
Source Files
Directories
¶
| Path | Synopsis |
|---|---|
|
Package leaderelection implements leader election of a set of endpoints.
|
Package leaderelection implements leader election of a set of endpoints. |
Click to show internal directories.
Click to hide internal directories.