Merge pull request #77598 from danielqsj/unreserve

Add Un-reserve extension point for the scheduling framework

pkg/scheduler/framework/v1alpha1/framework.go

@@ -33,6 +33,7 @@ type framework struct {
 	plugins          map[string]Plugin // a map of initialized plugins. Plugin name:plugin instance.
 	reservePlugins   []ReservePlugin
 	prebindPlugins   []PrebindPlugin
+	unreservePlugins []UnreservePlugin
 }
 
 var _ = Framework(&framework{})
@@ -64,6 +65,9 @@ func NewFramework(r Registry, _ *runtime.Unknown) (Framework, error) {
 		if pp, ok := p.(PrebindPlugin); ok {
 			f.prebindPlugins = append(f.prebindPlugins, pp)
 		}
+		if up, ok := p.(UnreservePlugin); ok {
+			f.unreservePlugins = append(f.unreservePlugins, up)
+		}
 	}
 	return f, nil
 }
@@ -105,6 +109,14 @@ func (f *framework) RunReservePlugins(
 	return nil
 }
 
+// RunUnreservePlugins runs the set of configured unreserve plugins.
+func (f *framework) RunUnreservePlugins(
+	pc *PluginContext, pod *v1.Pod, nodeName string) {
+	for _, pl := range f.unreservePlugins {
+		pl.Unreserve(pc, pod, nodeName)
+	}
+}
+
 // NodeInfoSnapshot returns the latest NodeInfo snapshot. The snapshot
 // is taken at the beginning of a scheduling cycle and remains unchanged until a
 // pod finishes "Reserve". There is no guarantee that the information remains

pkg/scheduler/framework/v1alpha1/interface.go

@@ -113,6 +113,17 @@ type PrebindPlugin interface {
 	Prebind(pc *PluginContext, p *v1.Pod, nodeName string) *Status
 }
 
+// UnreservePlugin is an interface for Unreserve plugins. This is an informational
+// extension point. If a pod was reserved and then rejected in a later phase, then
+// un-reserve plugins will be notified. Un-reserve plugins should clean up state
+// associated with the reserved Pod.
+type UnreservePlugin interface {
+	Plugin
+	// Unreserve is called by the scheduling framework when a reserved pod was
+	// rejected in a later phase.
+	Unreserve(pc *PluginContext, p *v1.Pod, nodeName string)
+}
+
 // Framework manages the set of plugins in use by the scheduling framework.
 // Configured plugins are called at specified points in a scheduling context.
 type Framework interface {
@@ -128,6 +139,9 @@ type Framework interface {
 	// plugins returns an error, it does not continue running the remaining ones and
 	// returns the error. In such case, pod will not be scheduled.
 	RunReservePlugins(pc *PluginContext, pod *v1.Pod, nodeName string) *Status
+
+	// RunUnreservePlugins runs the set of configured unreserve plugins.
+	RunUnreservePlugins(pc *PluginContext, pod *v1.Pod, nodeName string)
 }
 
 // FrameworkHandle provides data and some tools that plugins can use. It is

pkg/scheduler/scheduler.go

@@ -515,6 +515,8 @@ func (sched *Scheduler) scheduleOne() {
 	if err != nil {
 		klog.Errorf("error assuming pod: %v", err)
 		metrics.PodScheduleErrors.Inc()
+		// trigger un-reserve plugins to clean up state associated with the reserved Pod
+		fwk.RunUnreservePlugins(pluginContext, assumedPod, scheduleResult.SuggestedHost)
 		return
 	}
 	// bind the pod to its host asynchronously (we can do this b/c of the assumption step above).
@@ -525,6 +527,8 @@ func (sched *Scheduler) scheduleOne() {
 			if err != nil {
 				klog.Errorf("error binding volumes: %v", err)
 				metrics.PodScheduleErrors.Inc()
+				// trigger un-reserve plugins to clean up state associated with the reserved Pod
+				fwk.RunUnreservePlugins(pluginContext, assumedPod, scheduleResult.SuggestedHost)
 				return
 			}
 		}
@@ -543,6 +547,8 @@ func (sched *Scheduler) scheduleOne() {
 				klog.Errorf("scheduler cache ForgetPod failed: %v", forgetErr)
 			}
 			sched.recordSchedulingFailure(assumedPod, prebindStatus.AsError(), reason, prebindStatus.Message())
+			// trigger un-reserve plugins to clean up state associated with the reserved Pod
+			fwk.RunUnreservePlugins(pluginContext, assumedPod, scheduleResult.SuggestedHost)
 			return
 		}
 
@@ -558,6 +564,8 @@ func (sched *Scheduler) scheduleOne() {
 		if err != nil {
 			klog.Errorf("error binding pod: %v", err)
 			metrics.PodScheduleErrors.Inc()
+			// trigger un-reserve plugins to clean up state associated with the reserved Pod
+			fwk.RunUnreservePlugins(pluginContext, assumedPod, scheduleResult.SuggestedHost)
 		} else {
 			klog.V(2).Infof("pod %v/%v is bound successfully on node %v, %d nodes evaluated, %d nodes were found feasible", assumedPod.Namespace, assumedPod.Name, scheduleResult.SuggestedHost, scheduleResult.EvaluatedNodes, scheduleResult.FeasibleNodes)
 			metrics.PodScheduleSuccesses.Inc()