package scheduler

import "k8s.io/kubernetes/test/integration/scheduler"

Index ¶

• func InitRegistryAndConfig(t *testing.T, factory func(plugin framework.Plugin) frameworkruntime.PluginFactory, plugins ...framework.Plugin) (frameworkruntime.Registry, schedulerconfig.KubeSchedulerProfile)
• func InitTestSchedulerForFrameworkTest(t *testing.T, testCtx *testutils.TestContext, nodeCount int, runScheduler bool, opts ...scheduler.Option) (*testutils.TestContext, testutils.ShutdownFunc)
• func NewPlugin(plugin framework.Plugin) frameworkruntime.PluginFactory

Functions ¶

func InitRegistryAndConfig ¶

func InitRegistryAndConfig(t *testing.T, factory func(plugin framework.Plugin) frameworkruntime.PluginFactory, plugins ...framework.Plugin) (frameworkruntime.Registry, schedulerconfig.KubeSchedulerProfile)

InitRegistryAndConfig returns registry and plugins config based on give plugins.

func InitTestSchedulerForFrameworkTest ¶

func InitTestSchedulerForFrameworkTest(t *testing.T, testCtx *testutils.TestContext, nodeCount int, runScheduler bool, opts ...scheduler.Option) (*testutils.TestContext, testutils.ShutdownFunc)

The returned shutdown func will delete created resources and scheduler, resources should be those that will affect the scheduling result, like nodes, pods, etc.. Namespaces should not be deleted here because it's created together with the apiserver, they should be deleted simultaneously or we'll have no namespace. This should only be called when you want to kill the scheduler alone, away from apiserver. For example, in scheduler integration tests, recreating apiserver is performance consuming, then shutdown the scheduler and recreate it between each test case is a better approach.

func NewPlugin ¶

func NewPlugin(plugin framework.Plugin) frameworkruntime.PluginFactory

NewPlugin returns a plugin factory with specified Plugin.

Source Files ¶

util.go

Directories ¶