assign -998 as the oom_score_adj for critical pods.

pkg/kubelet/qos/policy.go

@@ -16,14 +16,20 @@ limitations under the License.
 
 package qos
 
-import "k8s.io/kubernetes/pkg/api/v1"
+import (
+	"k8s.io/kubernetes/pkg/api/v1"
+	kubepod "k8s.io/kubernetes/pkg/kubelet/pod"
+)
 
 const (
 	// PodInfraOOMAdj is very docker specific. For arbitrary runtime, it may not make
 	// sense to set sandbox level oom score, e.g. a sandbox could only be a namespace
 	// without a process.
 	// TODO: Handle infra container oom score adj in a runtime agnostic way.
+	// TODO: Should handle critical pod oom score adj with a proper preemption priority.
+	// This is the workaround for https://github.com/kubernetes/kubernetes/issues/38322.
 	PodInfraOOMAdj        int = -998
+	CriticalPodOOMAdj     int = -998
 	KubeletOOMScoreAdj    int = -999
 	DockerOOMScoreAdj     int = -999
 	KubeProxyOOMScoreAdj  int = -999
@@ -38,6 +44,10 @@ const (
 // and 1000. Containers with higher OOM scores are killed if the system runs out of memory.
 // See https://lwn.net/Articles/391222/ for more information.
 func GetContainerOOMScoreAdjust(pod *v1.Pod, container *v1.Container, memoryCapacity int64) int {
+	if kubepod.IsCriticalPod(pod) {
+		return CriticalPodOOMAdj
+	}
+
 	switch GetPodQOS(pod) {
 	case Guaranteed:
 		// Guaranteed containers should be the last to get killed.

pkg/kubelet/qos/policy_test.go

@@ -22,6 +22,7 @@ import (
 
 	"k8s.io/kubernetes/pkg/api/resource"
 	"k8s.io/kubernetes/pkg/api/v1"
+	kubetypes "k8s.io/kubernetes/pkg/kubelet/types"
 )
 
 const (
@@ -135,6 +136,25 @@ var (
 			},
 		},
 	}
+	criticalPodWithNoLimit = v1.Pod{
+		ObjectMeta: v1.ObjectMeta{
+			Annotations: map[string]string{
+				kubetypes.CriticalPodAnnotationKey: "",
+			},
+		},
+		Spec: v1.PodSpec{
+			Containers: []v1.Container{
+				{
+					Resources: v1.ResourceRequirements{
+						Requests: v1.ResourceList{
+							v1.ResourceName(v1.ResourceMemory): resource.MustParse(strconv.Itoa(standardMemoryAmount - 1)),
+							v1.ResourceName(v1.ResourceCPU):    resource.MustParse("5m"),
+						},
+					},
+				},
+			},
+		},
+	}
 )
 
 type oomTest struct {
@@ -188,6 +208,12 @@ func TestGetContainerOOMScoreAdjust(t *testing.T) {
 			lowOOMScoreAdj:  2,
 			highOOMScoreAdj: 2,
 		},
+		{
+			pod:             &criticalPodWithNoLimit,
+			memoryCapacity:  standardMemoryAmount,
+			lowOOMScoreAdj:  -998,
+			highOOMScoreAdj: -998,
+		},
 	}
 	for _, test := range oomTests {
 		oomScoreAdj := GetContainerOOMScoreAdjust(test.pod, &test.pod.Spec.Containers[0], test.memoryCapacity)