Merge pull request #27199 from derekwaynecarr/disk_eviction

Automatic merge from submit-queue

Initial support for pod eviction based on disk

This PR adds the following:

1. node reports disk pressure condition based on configured thresholds
1. scheduler does not place pods on nodes reporting disk pressure
1. kubelet will not admit any pod when it reports disk pressure
1. kubelet ranks pods for eviction when low on disk
1. kubelet evicts greediest pod

Follow-on PRs will need to handle:

1. integrate with new image gc PR (https://github.com/kubernetes/kubernetes/pull/27199)
1. container gc policy should always run (will not be launched from eviction, tbd who does that)
  1. this means kill pod is fine for all eviction code paths since container gc will remove dead container
1. min reclaim support will just poll summary provider (derek will do follow-on)
1. need to know if imagefs is same device as rootfs from summary (derek follow-on)

/cc @vishh @kubernetes/sig-node

plugin/pkg/scheduler/algorithm/predicates/error.go

@@ -39,6 +39,7 @@ var (
 	ErrServiceAffinityViolated   = newPredicateFailureError("CheckServiceAffinity")
 	ErrMaxVolumeCountExceeded    = newPredicateFailureError("MaxVolumeCount")
 	ErrNodeUnderMemoryPressure   = newPredicateFailureError("NodeUnderMemoryPressure")
+	ErrNodeUnderDiskPressure     = newPredicateFailureError("NodeUnderDiskPressure")
 	// ErrFakePredicate is used for test only. The fake predicates returning false also returns error
 	// as ErrFakePredicate.
 	ErrFakePredicate = newPredicateFailureError("FakePredicateError")

plugin/pkg/scheduler/algorithm/predicates/predicates.go

@@ -1049,3 +1049,21 @@ func CheckNodeMemoryPressurePredicate(pod *api.Pod, meta interface{}, nodeInfo *
 
 	return true, nil
 }
+
+// CheckNodeDiskPressurePredicate checks if a pod can be scheduled on a node
+// reporting disk pressure condition.
+func CheckNodeDiskPressurePredicate(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, error) {
+	node := nodeInfo.Node()
+	if node == nil {
+		return false, fmt.Errorf("node not found")
+	}
+
+	// is node under presure?
+	for _, cond := range node.Status.Conditions {
+		if cond.Type == api.NodeDiskPressure && cond.Status == api.ConditionTrue {
+			return false, ErrNodeUnderDiskPressure
+		}
+	}
+
+	return true, nil
+}

plugin/pkg/scheduler/algorithm/predicates/predicates_test.go

@@ -2906,3 +2906,72 @@ func TestPodSchedulesOnNodeWithMemoryPressureCondition(t *testing.T) {
 		}
 	}
 }
+
+func TestPodSchedulesOnNodeWithDiskPressureCondition(t *testing.T) {
+	pod := &api.Pod{
+		Spec: api.PodSpec{
+			Containers: []api.Container{
+				{
+					Name:            "container",
+					Image:           "image",
+					ImagePullPolicy: "Always",
+				},
+			},
+		},
+	}
+
+	// specify a node with no disk pressure condition on
+	noPressureNode := &api.Node{
+		Status: api.NodeStatus{
+			Conditions: []api.NodeCondition{
+				{
+					Type:   "Ready",
+					Status: "True",
+				},
+			},
+		},
+	}
+
+	// specify a node with pressure condition on
+	pressureNode := &api.Node{
+		Status: api.NodeStatus{
+			Conditions: []api.NodeCondition{
+				{
+					Type:   "DiskPressure",
+					Status: "True",
+				},
+			},
+		},
+	}
+
+	tests := []struct {
+		pod      *api.Pod
+		nodeInfo *schedulercache.NodeInfo
+		fits     bool
+		name     string
+	}{
+		{
+			pod:      pod,
+			nodeInfo: makeEmptyNodeInfo(noPressureNode),
+			fits:     true,
+			name:     "pod schedulable on node without pressure condition on",
+		},
+		{
+			pod:      pod,
+			nodeInfo: makeEmptyNodeInfo(pressureNode),
+			fits:     false,
+			name:     "pod not schedulable on node with pressure condition on",
+		},
+	}
+
+	for _, test := range tests {
+		fits, err := CheckNodeDiskPressurePredicate(test.pod, nil, test.nodeInfo)
+		if fits != test.fits {
+			t.Errorf("%s: expected %v got %v", test.name, test.fits, fits)
+		}
+
+		if err != nil && err != ErrNodeUnderDiskPressure {
+			t.Errorf("%s: unexpected error: %v", test.name, err)
+		}
+	}
+}