feature:add preemptionpolicy in preempt and reclaim

Signed-off-by: jessestutler <jesseincomparable@hotmail.com>

pkg/controllers/job/job_controller_util.go

@@ -62,6 +62,11 @@ func createJobPod(job *batch.Job, template *v1.PodTemplateSpec, topologyPolicy b
 		pod.Spec.SchedulerName = job.Spec.SchedulerName
 	}
 
+	// If no priority class specified in pod template, use priority class specified in job
+	if len(pod.Spec.PriorityClassName) == 0 && len(job.Spec.PriorityClassName) != 0 {
+		pod.Spec.PriorityClassName = job.Spec.PriorityClassName
+	}
+
 	volumeMap := make(map[string]string)
 	for _, volume := range job.Spec.Volumes {
 		vcName := volume.VolumeClaimName

pkg/scheduler/actions/preempt/preempt.go

@@ -19,6 +19,7 @@ package preempt
 import (
 	"fmt"
 
+	v1 "k8s.io/api/core/v1"
 	"k8s.io/klog/v2"
 
 	"volcano.sh/volcano/pkg/scheduler/api"
@@ -231,6 +232,10 @@ func (pmpt *Action) preempt(
 	filter func(*api.TaskInfo) bool,
 	predicateHelper util.PredicateHelper,
 ) (bool, error) {
+	if preemptor.Pod.Spec.PreemptionPolicy != nil && *preemptor.Pod.Spec.PreemptionPolicy == v1.PreemptNever {
+		return false, fmt.Errorf("not eligible to preempt other tasks due to preemptionPolicy is Never")
+	}
+
 	assigned := false
 
 	if err := ssn.PrePredicateFn(preemptor); err != nil {

pkg/scheduler/actions/preempt/preempt_test.go

@@ -233,6 +233,26 @@ func TestPreempt(t *testing.T) {
 			ExpectEvicted:  []string{"c1/preemptee1"},
 			ExpectEvictNum: 1,
 		},
+		{
+			// case about #3642
+			Name: "can not preempt resources when task preemption policy is never",
+			PodGroups: []*schedulingv1beta1.PodGroup{
+				util.BuildPodGroupWithPrio("pg1", "c1", "q1", 0, nil, schedulingv1beta1.PodGroupInqueue, "low-priority"),
+				util.BuildPodGroupWithPrio("pg2", "c1", "q1", 0, nil, schedulingv1beta1.PodGroupInqueue, "high-priority"),
+			},
+			Pods: []*v1.Pod{
+				util.BuildPod("c1", "preemptee1", "n1", v1.PodRunning, api.BuildResourceList("1", "1G"), "pg1", map[string]string{schedulingv1beta1.PodPreemptable: "true"}, make(map[string]string)),
+				util.BuildPodWithPreeemptionPolicy("c1", "preemptor1", "", v1.PodPending, api.BuildResourceList("1", "1G"), "pg2", make(map[string]string), make(map[string]string), v1.PreemptNever),
+			},
+			Nodes: []*v1.Node{
+				util.BuildNode("n1", api.BuildResourceList("1", "1Gi", []api.ScalarResource{{Name: "pods", Value: "1"}}...), make(map[string]string)),
+			},
+			Queues: []*schedulingv1beta1.Queue{
+				util.BuildQueue("q1", 1, nil),
+			},
+			ExpectEvictNum: 0,
+			ExpectEvicted:  []string{}, // no victims should be reclaimed
+		},
 	}
 
 	trueValue := true

pkg/scheduler/actions/reclaim/reclaim.go

@@ -17,6 +17,7 @@ limitations under the License.
 package reclaim
 
 import (
+	v1 "k8s.io/api/core/v1"
 	"k8s.io/klog/v2"
 
 	"volcano.sh/volcano/pkg/scheduler/api"
@@ -113,6 +114,15 @@ func (ra *Action) Execute(ssn *framework.Session) {
 			task = tasks.Pop().(*api.TaskInfo)
 		}
 
+		if task.Pod.Spec.PreemptionPolicy != nil && *task.Pod.Spec.PreemptionPolicy == v1.PreemptNever {
+			klog.V(3).Infof("Task %s/%s is not eligible to preempt other tasks due to preemptionPolicy is Never", task.Namespace, task.Name)
+			// TODO: In order to avoid blocking other tasks in the job or other jobs in the queue to reclaim resources, the job and queue need
+			// to be pushed back to the priority queue. Need to refactor the framework of reclaim action, see issue: https://github.com/volcano-sh/volcano/issues/3738
+			jobs.Push(job)
+			queues.Push(queue)
+			continue
+		}
+
 		if !ssn.Allocatable(queue, task) {
 			klog.V(3).Infof("Queue <%s> is overused when considering task <%s>, ignore it.", queue.Name, task.Name)
 			continue

pkg/scheduler/actions/reclaim/reclaim_test.go

@@ -129,6 +129,36 @@ func TestReclaim(t *testing.T) {
 			ExpectEvictNum: 1,
 			ExpectEvicted:  []string{"c1/preemptee1-1"}, // low queue priority job's preemptable pod is evicted
 		},
+		{
+			// case about #3642
+			Name: "can not reclaim resources when task preemption policy is never",
+			Plugins: map[string]framework.PluginBuilder{
+				conformance.PluginName: conformance.New,
+				gang.PluginName:        gang.New,
+				proportion.PluginName:  proportion.New,
+			},
+			PriClass: []*schedulingv1.PriorityClass{
+				util.BuildPriorityClass("low-priority", 100),
+				util.BuildPriorityClassWithPreemptionPolicy("high-priority", 1000, v1.PreemptNever),
+			},
+			PodGroups: []*schedulingv1beta1.PodGroup{
+				util.BuildPodGroupWithPrio("pg1", "c1", "q1", 0, nil, schedulingv1beta1.PodGroupInqueue, "low-priority"),
+				util.BuildPodGroupWithPrio("pg2", "c1", "q2", 0, nil, schedulingv1beta1.PodGroupInqueue, "high-priority"),
+			},
+			Pods: []*v1.Pod{
+				util.BuildPod("c1", "preemptee1", "n1", v1.PodRunning, api.BuildResourceList("1", "1G"), "pg1", map[string]string{schedulingv1beta1.PodPreemptable: "true"}, make(map[string]string)),
+				util.BuildPodWithPreeemptionPolicy("c1", "preemptor1", "", v1.PodPending, api.BuildResourceList("1", "1G"), "pg2", make(map[string]string), make(map[string]string), v1.PreemptNever),
+			},
+			Nodes: []*v1.Node{
+				util.BuildNode("n1", api.BuildResourceList("1", "1Gi", []api.ScalarResource{{Name: "pods", Value: "1"}}...), make(map[string]string)),
+			},
+			Queues: []*schedulingv1beta1.Queue{
+				util.BuildQueue("q1", 5, nil),
+				util.BuildQueue("q2", 10, nil),
+			},
+			ExpectEvictNum: 0,
+			ExpectEvicted:  []string{}, // no victims should be reclaimed
+		},
 	}
 
 	reclaim := New()

pkg/scheduler/util/test_utils.go

@@ -190,6 +190,13 @@ func BuildPodWithPriority(namespace, name, nodeName string, p v1.PodPhase, req v
 	return pod
 }
 
+// BuildPodWithPreemptionPolicy builds a pod with preemptionPolicy
+func BuildPodWithPreeemptionPolicy(namespace, name, nodeName string, p v1.PodPhase, req v1.ResourceList, groupName string, labels map[string]string, selector map[string]string, preemptionPolicy v1.PreemptionPolicy) *v1.Pod {
+	pod := BuildPod(namespace, name, nodeName, p, req, groupName, labels, selector)
+	pod.Spec.PreemptionPolicy = &preemptionPolicy
+	return pod
+}
+
 // BuildPodGroup return podgroup with base spec and phase status
 func BuildPodGroup(name, ns, queue string, minMember int32, taskMinMember map[string]int32, status schedulingv1beta1.PodGroupPhase) *schedulingv1beta1.PodGroup {
 	return &schedulingv1beta1.PodGroup{
@@ -304,6 +311,13 @@ func BuildPriorityClass(name string, value int32) *schedulingv1.PriorityClass {
 	}
 }
 
+// BuildPriorityClassWithPreemptionPolicy return a priorityClass with value and preemptionPolicy
+func BuildPriorityClassWithPreemptionPolicy(name string, value int32, preemptionPolicy v1.PreemptionPolicy) *schedulingv1.PriorityClass {
+	pc := BuildPriorityClass(name, value)
+	pc.PreemptionPolicy = &preemptionPolicy
+	return pc
+}
+
 // FakeBinder is used as fake binder
 type FakeBinder struct {
 	sync.RWMutex