diff --git a/pkg/kubelet/cm/container_manager_linux.go b/pkg/kubelet/cm/container_manager_linux.go
index 78a496fd69..53136f7978 100644
--- a/pkg/kubelet/cm/container_manager_linux.go
+++ b/pkg/kubelet/cm/container_manager_linux.go
@@ -147,6 +147,8 @@ func setupKernelTunables(option KernelTunableBehavior) error {
 	desiredState := map[string]int{
 		utilsysctl.VmOvercommitMemory: utilsysctl.VmOvercommitMemoryAlways,
 		utilsysctl.VmPanicOnOOM:       utilsysctl.VmPanicOnOOMInvokeOOMKiller,
+		utilsysctl.KernelPanic:        utilsysctl.KernelPanicRebootTimeout,
+		utilsysctl.KernelPanicOnOops:  utilsysctl.KernelPanicOnOopsAlways,
 	}
 
 	errList := []error{}
diff --git a/pkg/kubelet/dockertools/fake_manager.go b/pkg/kubelet/dockertools/fake_manager.go
index 04e6a014ed..4f0a96cb58 100644
--- a/pkg/kubelet/dockertools/fake_manager.go
+++ b/pkg/kubelet/dockertools/fake_manager.go
@@ -44,7 +44,7 @@ func NewFakeDockerManager(
 	httpClient kubetypes.HttpGetter, imageBackOff *util.Backoff) *DockerManager {
 
 	fakeOOMAdjuster := oom.NewFakeOOMAdjuster()
-	fakeProcFs := procfs.NewFakeProcFs()
+	fakeProcFs := procfs.NewFakeProcFS()
 	dm := NewDockerManager(client, recorder, livenessManager, containerRefManager, machineInfo, podInfraContainerImage, qps,
 		burst, containerLogsDir, osInterface, networkPlugin, generator, httpClient, &NativeExecHandler{},
 		fakeOOMAdjuster, fakeProcFs, false, imageBackOff, true)
diff --git a/pkg/kubelet/dockertools/manager.go b/pkg/kubelet/dockertools/manager.go
index 3db3066fa7..ac1d8abda8 100644
--- a/pkg/kubelet/dockertools/manager.go
+++ b/pkg/kubelet/dockertools/manager.go
@@ -133,7 +133,7 @@ type DockerManager struct {
 	oomAdjuster *oom.OOMAdjuster
 
 	// Get information from /proc mount.
-	procFs procfs.ProcFsInterface
+	procFs procfs.ProcFSInterface
 
 	// If true, enforce container cpu limits with CFS quota support
 	cpuCFSQuota bool
@@ -158,7 +158,7 @@ func NewDockerManager(
 	httpClient kubetypes.HttpGetter,
 	execHandler ExecHandler,
 	oomAdjuster *oom.OOMAdjuster,
-	procFs procfs.ProcFsInterface,
+	procFs procfs.ProcFSInterface,
 	cpuCFSQuota bool,
 	imageBackOff *util.Backoff,
 	serializeImagePulls bool) *DockerManager {
diff --git a/pkg/kubelet/kubelet.go b/pkg/kubelet/kubelet.go
index 9c4f0807ac..74c59a2759 100644
--- a/pkg/kubelet/kubelet.go
+++ b/pkg/kubelet/kubelet.go
@@ -320,7 +320,7 @@ func NewMainKubelet(
 		return nil, err
 	}
 
-	procFs := procfs.NewProcFs()
+	procFs := procfs.NewProcFS()
 	imageBackOff := util.NewBackOff(resyncInterval, MaxContainerBackOff)
 
 	klet.livenessManager = proberesults.NewManager()
diff --git a/pkg/util/procfs/procfs.go b/pkg/util/procfs/procfs.go
index 22dd1470cc..c0a45725fb 100644
--- a/pkg/util/procfs/procfs.go
+++ b/pkg/util/procfs/procfs.go
@@ -24,10 +24,10 @@ import (
 	"strings"
 )
 
-type ProcFs struct{}
+type ProcFS struct{}
 
-func NewProcFs() ProcFsInterface {
-	return &ProcFs{}
+func NewProcFS() ProcFSInterface {
+	return &ProcFS{}
 }
 
 func containerNameFromProcCgroup(content string) (string, error) {
@@ -44,7 +44,7 @@ func containerNameFromProcCgroup(content string) (string, error) {
 // getFullContainerName gets the container name given the root process id of the container.
 // Eg. If the devices cgroup for the container is stored in /sys/fs/cgroup/devices/docker/nginx,
 // return docker/nginx. Assumes that the process is part of exactly one cgroup hierarchy.
-func (pfs *ProcFs) GetFullContainerName(pid int) (string, error) {
+func (pfs *ProcFS) GetFullContainerName(pid int) (string, error) {
 	filePath := path.Join("/proc", strconv.Itoa(pid), "cgroup")
 	content, err := ioutil.ReadFile(filePath)
 	if err != nil {
diff --git a/pkg/util/procfs/procfs_fake.go b/pkg/util/procfs/procfs_fake.go
index b002a9d98c..8d16aa53cc 100644
--- a/pkg/util/procfs/procfs_fake.go
+++ b/pkg/util/procfs/procfs_fake.go
@@ -16,15 +16,15 @@ limitations under the License.
 
 package procfs
 
-type FakeProcFs struct{}
+type FakeProcFS struct{}
 
-func NewFakeProcFs() ProcFsInterface {
-	return &FakeProcFs{}
+func NewFakeProcFS() ProcFSInterface {
+	return &FakeProcFS{}
 }
 
-// getFullContainerName gets the container name given the root process id of the container.
+// GetFullContainerName gets the container name given the root process id of the container.
 // Eg. If the devices cgroup for the container is stored in /sys/fs/cgroup/devices/docker/nginx,
 // return docker/nginx. Assumes that the process is part of exactly one cgroup hierarchy.
-func (fakePfs *FakeProcFs) GetFullContainerName(pid int) (string, error) {
+func (fakePfs *FakeProcFS) GetFullContainerName(pid int) (string, error) {
 	return "", nil
 }
diff --git a/pkg/util/procfs/procfs_interface.go b/pkg/util/procfs/procfs_interface.go
index b776443dbf..d3bf14f0c7 100644
--- a/pkg/util/procfs/procfs_interface.go
+++ b/pkg/util/procfs/procfs_interface.go
@@ -16,7 +16,7 @@ limitations under the License.
 
 package procfs
 
-type ProcFsInterface interface {
-	// getFullContainerName gets the container name given the root process id of the container.
+type ProcFSInterface interface {
+	// GetFullContainerName gets the container name given the root process id of the container.
 	GetFullContainerName(pid int) (string, error)
 }
diff --git a/pkg/util/sysctl/sysctl.go b/pkg/util/sysctl/sysctl.go
index 784a20ff23..9398a7ac31 100644
--- a/pkg/util/sysctl/sysctl.go
+++ b/pkg/util/sysctl/sysctl.go
@@ -27,9 +27,14 @@ const (
 	sysctlBase         = "/proc/sys"
 	VmOvercommitMemory = "vm/overcommit_memory"
 	VmPanicOnOOM       = "vm/panic_on_oom"
+	KernelPanic        = "kernel/panic"
+	KernelPanicOnOops  = "kernel/panic_on_oops"
 
 	VmOvercommitMemoryAlways    = 1 // kernel performs no memory over-commit handling
 	VmPanicOnOOMInvokeOOMKiller = 0 // kernel calls the oom_killer function when OOM occurs
+
+	KernelPanicOnOopsAlways  = 1  // kernel panics on kernel oops
+	KernelPanicRebootTimeout = 10 // seconds after a panic for the kernel to reboot
 )
 
 // GetSysctl returns the value for the specified sysctl setting