Expose accelerator metrics in the summary API.

pkg/kubelet/apis/stats/v1alpha1/types.go

@@ -109,6 +109,8 @@ type ContainerStats struct {
 	// Stats pertaining to memory (RAM) resources.
 	// +optional
 	Memory *MemoryStats `json:"memory,omitempty"`
+	// Metrics for Accelerators. Each Accelerator corresponds to one element in the array.
+	Accelerators []AcceleratorStats `json:"accelerators,omitempty"`
 	// Stats pertaining to container rootfs usage of filesystem resources.
 	// Rootfs.UsedBytes is the number of bytes used for the container write layer.
 	// +optional
@@ -188,6 +190,30 @@ type MemoryStats struct {
 	MajorPageFaults *uint64 `json:"majorPageFaults,omitempty"`
 }
 
+// AcceleratorStats contains stats for accelerators attached to the container.
+type AcceleratorStats struct {
+	// Make of the accelerator (nvidia, amd, google etc.)
+	Make string `json:"make"`
+
+	// Model of the accelerator (tesla-p100, tesla-k80 etc.)
+	Model string `json:"model"`
+
+	// ID of the accelerator.
+	ID string `json:"id"`
+
+	// Total accelerator memory.
+	// unit: bytes
+	MemoryTotal uint64 `json:"memory_total"`
+
+	// Total accelerator memory allocated.
+	// unit: bytes
+	MemoryUsed uint64 `json:"memory_used"`
+
+	// Percent of time over the past sample period (10s) during which
+	// the accelerator was actively processing.
+	DutyCycle uint64 `json:"duty_cycle"`
+}
+
 // VolumeStats contains data about Volume filesystem usage.
 type VolumeStats struct {
 	// Embedded FsStats

pkg/kubelet/server/stats/summary_test.go

@@ -92,6 +92,7 @@ func TestSummaryProvider(t *testing.T) {
 		StartTime:          cgroupStatsMap["/kubelet"].cs.StartTime,
 		CPU:                cgroupStatsMap["/kubelet"].cs.CPU,
 		Memory:             cgroupStatsMap["/kubelet"].cs.Memory,
+		Accelerators:       cgroupStatsMap["/kubelet"].cs.Accelerators,
 		UserDefinedMetrics: cgroupStatsMap["/kubelet"].cs.UserDefinedMetrics,
 	})
 	assert.Contains(summary.Node.SystemContainers, statsapi.ContainerStats{
@@ -99,6 +100,7 @@ func TestSummaryProvider(t *testing.T) {
 		StartTime:          cgroupStatsMap["/misc"].cs.StartTime,
 		CPU:                cgroupStatsMap["/misc"].cs.CPU,
 		Memory:             cgroupStatsMap["/misc"].cs.Memory,
+		Accelerators:       cgroupStatsMap["/misc"].cs.Accelerators,
 		UserDefinedMetrics: cgroupStatsMap["/misc"].cs.UserDefinedMetrics,
 	})
 	assert.Contains(summary.Node.SystemContainers, statsapi.ContainerStats{
@@ -106,6 +108,7 @@ func TestSummaryProvider(t *testing.T) {
 		StartTime:          cgroupStatsMap["/runtime"].cs.StartTime,
 		CPU:                cgroupStatsMap["/runtime"].cs.CPU,
 		Memory:             cgroupStatsMap["/runtime"].cs.Memory,
+		Accelerators:       cgroupStatsMap["/runtime"].cs.Accelerators,
 		UserDefinedMetrics: cgroupStatsMap["/runtime"].cs.UserDefinedMetrics,
 	})
 	assert.Equal(summary.Pods, podStats)

pkg/kubelet/stats/helper.go

@@ -119,6 +119,17 @@ func cadvisorInfoToContainerStats(name string, info *cadvisorapiv2.ContainerInfo
 		}
 	}
 
+	for _, acc := range cstat.Accelerators {
+		result.Accelerators = append(result.Accelerators, statsapi.AcceleratorStats{
+			Make:        acc.Make,
+			Model:       acc.Model,
+			ID:          acc.ID,
+			MemoryTotal: acc.MemoryTotal,
+			MemoryUsed:  acc.MemoryUsed,
+			DutyCycle:   acc.DutyCycle,
+		})
+	}
+
 	result.UserDefinedMetrics = cadvisorInfoToUserDefinedMetrics(info)
 
 	return result

test/e2e_node/summary_test.go

@@ -97,6 +97,7 @@ var _ = framework.KubeDescribe("Summary API", func() {
 						"PageFaults":      bounded(1000, 1E9),
 						"MajorPageFaults": bounded(0, 100000),
 					}),
+					"Accelerators":       BeEmpty(),
 					"Rootfs":             BeNil(),
 					"Logs":               BeNil(),
 					"UserDefinedMetrics": BeEmpty(),
@@ -145,6 +146,7 @@ var _ = framework.KubeDescribe("Summary API", func() {
 							"PageFaults":      bounded(100, 1000000),
 							"MajorPageFaults": bounded(0, 10),
 						}),
+						"Accelerators": BeEmpty(),
 						"Rootfs": ptrMatchAllFields(gstruct.Fields{
 							"Time":           recent(maxStatsAge),
 							"AvailableBytes": fsCapacityBounds,