From a8b8995ef241e93e9486d475126450f33f24ef4e Mon Sep 17 00:00:00 2001 From: Alexey Perevalov Date: Tue, 14 Jul 2020 18:03:56 +0300 Subject: [PATCH] Implement TopologyInfo and cpu_ids in podresources It covers deviceplugin & cpumanager. It has drawback, since cpuset and all other structs including cadvisor's keep cpu as int, but for protobuf based interface is better to have fixed int. This patch also introduces additional interface CPUsProvider, while DeviceProvider might have been extended too. Checkpoint not covered by unit test. Signed-off-by: Swati Sehgal Signed-off-by: Alexey Perevalov --- pkg/kubelet/apis/podresources/server_v1.go | 5 +- .../apis/podresources/server_v1_test.go | 12 ++++- .../apis/podresources/server_v1alpha1_test.go | 5 ++ pkg/kubelet/apis/podresources/types.go | 5 ++ pkg/kubelet/cm/container_manager.go | 3 ++ pkg/kubelet/cm/container_manager_linux.go | 4 ++ pkg/kubelet/cm/container_manager_stub.go | 4 ++ pkg/kubelet/cm/container_manager_windows.go | 4 ++ pkg/kubelet/cm/cpumanager/cpu_manager.go | 13 +++++ pkg/kubelet/cm/cpumanager/fake_cpu_manager.go | 5 ++ pkg/kubelet/cm/devicemanager/BUILD | 2 + pkg/kubelet/cm/devicemanager/checkpoint/BUILD | 1 + .../cm/devicemanager/checkpoint/checkpoint.go | 22 ++++++++- pkg/kubelet/cm/devicemanager/manager.go | 11 ++++- pkg/kubelet/cm/devicemanager/manager_test.go | 39 +++++++++------ pkg/kubelet/cm/devicemanager/pod_devices.go | 32 ++++++------- .../cm/devicemanager/pod_devices_test.go | 47 +++++++++++++++++++ .../cm/devicemanager/topology_hints_test.go | 2 +- pkg/kubelet/kubelet.go | 2 +- pkg/kubelet/server/server.go | 4 +- 20 files changed, 183 insertions(+), 39 deletions(-) create mode 100644 pkg/kubelet/cm/devicemanager/pod_devices_test.go diff --git a/pkg/kubelet/apis/podresources/server_v1.go b/pkg/kubelet/apis/podresources/server_v1.go index fe4666a969c..8482f51a19d 100644 --- a/pkg/kubelet/apis/podresources/server_v1.go +++ b/pkg/kubelet/apis/podresources/server_v1.go @@ -28,14 +28,16 @@ import ( type v1PodResourcesServer struct { podsProvider PodsProvider devicesProvider DevicesProvider + cpusProvider CPUsProvider } // NewV1PodResourcesServer returns a PodResourcesListerServer which lists pods provided by the PodsProvider // with device information provided by the DevicesProvider -func NewV1PodResourcesServer(podsProvider PodsProvider, devicesProvider DevicesProvider) v1.PodResourcesListerServer { +func NewV1PodResourcesServer(podsProvider PodsProvider, devicesProvider DevicesProvider, cpusProvider CPUsProvider) v1.PodResourcesListerServer { return &v1PodResourcesServer{ podsProvider: podsProvider, devicesProvider: devicesProvider, + cpusProvider: cpusProvider, } } @@ -58,6 +60,7 @@ func (p *v1PodResourcesServer) List(ctx context.Context, req *v1.ListPodResource pRes.Containers[j] = &v1.ContainerResources{ Name: container.Name, Devices: p.devicesProvider.GetDevices(string(pod.UID), container.Name), + CpuIds: p.cpusProvider.GetCPUs(string(pod.UID), container.Name), } } podResources[i] = &pRes diff --git a/pkg/kubelet/apis/podresources/server_v1_test.go b/pkg/kubelet/apis/podresources/server_v1_test.go index 40ace68f351..c25912ee3ba 100644 --- a/pkg/kubelet/apis/podresources/server_v1_test.go +++ b/pkg/kubelet/apis/podresources/server_v1_test.go @@ -31,24 +31,30 @@ func TestListPodResourcesV1(t *testing.T) { podNamespace := "pod-namespace" podUID := types.UID("pod-uid") containerName := "container-name" + numaID := int64(1) devs := []*podresourcesapi.ContainerDevices{ { ResourceName: "resource", DeviceIds: []string{"dev0", "dev1"}, + Topology: &podresourcesapi.TopologyInfo{Nodes: []*podresourcesapi.NUMANode{{ID: numaID}}}, }, } + cpus := []int64{12, 23, 30} + for _, tc := range []struct { desc string pods []*v1.Pod devices []*podresourcesapi.ContainerDevices + cpus []int64 expectedResponse *podresourcesapi.ListPodResourcesResponse }{ { desc: "no pods", pods: []*v1.Pod{}, devices: []*podresourcesapi.ContainerDevices{}, + cpus: []int64{}, expectedResponse: &podresourcesapi.ListPodResourcesResponse{}, }, { @@ -70,6 +76,7 @@ func TestListPodResourcesV1(t *testing.T) { }, }, devices: []*podresourcesapi.ContainerDevices{}, + cpus: []int64{}, expectedResponse: &podresourcesapi.ListPodResourcesResponse{ PodResources: []*podresourcesapi.PodResources{ { @@ -104,6 +111,7 @@ func TestListPodResourcesV1(t *testing.T) { }, }, devices: devs, + cpus: cpus, expectedResponse: &podresourcesapi.ListPodResourcesResponse{ PodResources: []*podresourcesapi.PodResources{ { @@ -113,6 +121,7 @@ func TestListPodResourcesV1(t *testing.T) { { Name: containerName, Devices: devs, + CpuIds: cpus, }, }, }, @@ -124,8 +133,9 @@ func TestListPodResourcesV1(t *testing.T) { m := new(mockProvider) m.On("GetPods").Return(tc.pods) m.On("GetDevices", string(podUID), containerName).Return(tc.devices) + m.On("GetCPUs", string(podUID), containerName).Return(tc.cpus) m.On("UpdateAllocatedDevices").Return() - server := NewV1PodResourcesServer(m, m) + server := NewV1PodResourcesServer(m, m, m) resp, err := server.List(context.TODO(), &podresourcesapi.ListPodResourcesRequest{}) if err != nil { t.Errorf("want err = %v, got %q", nil, err) diff --git a/pkg/kubelet/apis/podresources/server_v1alpha1_test.go b/pkg/kubelet/apis/podresources/server_v1alpha1_test.go index b87ac10fa8c..5fe6f966e9d 100644 --- a/pkg/kubelet/apis/podresources/server_v1alpha1_test.go +++ b/pkg/kubelet/apis/podresources/server_v1alpha1_test.go @@ -43,6 +43,11 @@ func (m *mockProvider) GetDevices(podUID, containerName string) []*podresourcesv return args.Get(0).([]*podresourcesv1.ContainerDevices) } +func (m *mockProvider) GetCPUs(podUID, containerName string) []int64 { + args := m.Called(podUID, containerName) + return args.Get(0).([]int64) +} + func (m *mockProvider) UpdateAllocatedDevices() { m.Called() } diff --git a/pkg/kubelet/apis/podresources/types.go b/pkg/kubelet/apis/podresources/types.go index 266c0c8af70..433d92c5996 100644 --- a/pkg/kubelet/apis/podresources/types.go +++ b/pkg/kubelet/apis/podresources/types.go @@ -31,3 +31,8 @@ type DevicesProvider interface { type PodsProvider interface { GetPods() []*v1.Pod } + +// CPUsProvider knows how to provide the cpus used by the given container +type CPUsProvider interface { + GetCPUs(podUID, containerName string) []int64 +} diff --git a/pkg/kubelet/cm/container_manager.go b/pkg/kubelet/cm/container_manager.go index 603a150a26a..bf8122610a2 100644 --- a/pkg/kubelet/cm/container_manager.go +++ b/pkg/kubelet/cm/container_manager.go @@ -106,6 +106,9 @@ type ContainerManager interface { // GetDevices returns information about the devices assigned to pods and containers GetDevices(podUID, containerName string) []*podresourcesapi.ContainerDevices + // GetCPUs returns information about the cpus assigned to pods and containers + GetCPUs(podUID, containerName string) []int64 + // ShouldResetExtendedResourceCapacity returns whether or not the extended resources should be zeroed, // due to node recreation. ShouldResetExtendedResourceCapacity() bool diff --git a/pkg/kubelet/cm/container_manager_linux.go b/pkg/kubelet/cm/container_manager_linux.go index f0c0e33cf1d..1c95b652918 100644 --- a/pkg/kubelet/cm/container_manager_linux.go +++ b/pkg/kubelet/cm/container_manager_linux.go @@ -1027,6 +1027,10 @@ func (cm *containerManagerImpl) GetDevices(podUID, containerName string) []*podr return cm.deviceManager.GetDevices(podUID, containerName) } +func (cm *containerManagerImpl) GetCPUs(podUID, containerName string) []int64 { + return cm.cpuManager.GetCPUs(podUID, containerName) +} + func (cm *containerManagerImpl) ShouldResetExtendedResourceCapacity() bool { return cm.deviceManager.ShouldResetExtendedResourceCapacity() } diff --git a/pkg/kubelet/cm/container_manager_stub.go b/pkg/kubelet/cm/container_manager_stub.go index 6d8fa08dcab..65eac7aadcf 100644 --- a/pkg/kubelet/cm/container_manager_stub.go +++ b/pkg/kubelet/cm/container_manager_stub.go @@ -125,6 +125,10 @@ func (cm *containerManagerStub) UpdateAllocatedDevices() { return } +func (cm *containerManagerStub) GetCPUs(_, _ string) []int64 { + return nil +} + func NewStubContainerManager() ContainerManager { return &containerManagerStub{shouldResetExtendedResourceCapacity: false} } diff --git a/pkg/kubelet/cm/container_manager_windows.go b/pkg/kubelet/cm/container_manager_windows.go index 3bc0f00e2a6..e6637b4d443 100644 --- a/pkg/kubelet/cm/container_manager_windows.go +++ b/pkg/kubelet/cm/container_manager_windows.go @@ -192,3 +192,7 @@ func (cm *containerManagerImpl) GetAllocateResourcesPodAdmitHandler() lifecycle. func (cm *containerManagerImpl) UpdateAllocatedDevices() { return } + +func (cm *containerManagerImpl) GetCPUs(_, _ string) []int64 { + return nil +} diff --git a/pkg/kubelet/cm/cpumanager/cpu_manager.go b/pkg/kubelet/cm/cpumanager/cpu_manager.go index 1e6cc936d82..5f68c5957cd 100644 --- a/pkg/kubelet/cm/cpumanager/cpu_manager.go +++ b/pkg/kubelet/cm/cpumanager/cpu_manager.go @@ -77,6 +77,10 @@ type Manager interface { // and is consulted to achieve NUMA aware resource alignment among this // and other resource controllers. GetTopologyHints(*v1.Pod, *v1.Container) map[string][]topologymanager.TopologyHint + + // GetCPUs implements the podresources.CPUsProvider interface to provide allocated + // cpus for the container + GetCPUs(podUID, containerName string) []int64 } type manager struct { @@ -461,3 +465,12 @@ func (m *manager) updateContainerCPUSet(containerID string, cpus cpuset.CPUSet) CpusetCpus: cpus.String(), }) } + +func (m *manager) GetCPUs(podUID, containerName string) []int64 { + cpus := m.state.GetCPUSetOrDefault(string(podUID), containerName) + result := []int64{} + for _, cpu := range cpus.ToSliceNoSort() { + result = append(result, int64(cpu)) + } + return result +} diff --git a/pkg/kubelet/cm/cpumanager/fake_cpu_manager.go b/pkg/kubelet/cm/cpumanager/fake_cpu_manager.go index d8d707173bd..f73d038298f 100644 --- a/pkg/kubelet/cm/cpumanager/fake_cpu_manager.go +++ b/pkg/kubelet/cm/cpumanager/fake_cpu_manager.go @@ -64,6 +64,11 @@ func (m *fakeManager) State() state.Reader { return m.state } +func (m *fakeManager) GetCPUs(podUID, containerName string) []int64 { + klog.Infof("[fake cpumanager] GetCPUs(podUID: %s, containerName: %s)", podUID, containerName) + return nil +} + // NewFakeManager creates empty/fake cpu manager func NewFakeManager() Manager { return &fakeManager{ diff --git a/pkg/kubelet/cm/devicemanager/BUILD b/pkg/kubelet/cm/devicemanager/BUILD index a2a58b0b363..571f8b37389 100644 --- a/pkg/kubelet/cm/devicemanager/BUILD +++ b/pkg/kubelet/cm/devicemanager/BUILD @@ -48,11 +48,13 @@ go_test( srcs = [ "endpoint_test.go", "manager_test.go", + "pod_devices_test.go", "topology_hints_test.go", ], embed = [":go_default_library"], deps = [ "//pkg/kubelet/checkpointmanager:go_default_library", + "//pkg/kubelet/cm/devicemanager/checkpoint:go_default_library", "//pkg/kubelet/cm/topologymanager:go_default_library", "//pkg/kubelet/cm/topologymanager/bitmask:go_default_library", "//pkg/kubelet/config:go_default_library", diff --git a/pkg/kubelet/cm/devicemanager/checkpoint/BUILD b/pkg/kubelet/cm/devicemanager/checkpoint/BUILD index 91506a71f7c..afdace3db4f 100644 --- a/pkg/kubelet/cm/devicemanager/checkpoint/BUILD +++ b/pkg/kubelet/cm/devicemanager/checkpoint/BUILD @@ -8,6 +8,7 @@ go_library( deps = [ "//pkg/kubelet/checkpointmanager:go_default_library", "//pkg/kubelet/checkpointmanager/checksum:go_default_library", + "//staging/src/k8s.io/apimachinery/pkg/util/sets:go_default_library", ], ) diff --git a/pkg/kubelet/cm/devicemanager/checkpoint/checkpoint.go b/pkg/kubelet/cm/devicemanager/checkpoint/checkpoint.go index 441eebd6684..0430fcd1d2b 100644 --- a/pkg/kubelet/cm/devicemanager/checkpoint/checkpoint.go +++ b/pkg/kubelet/cm/devicemanager/checkpoint/checkpoint.go @@ -19,6 +19,7 @@ package checkpoint import ( "encoding/json" + "k8s.io/apimachinery/pkg/util/sets" "k8s.io/kubernetes/pkg/kubelet/checkpointmanager" "k8s.io/kubernetes/pkg/kubelet/checkpointmanager/checksum" ) @@ -29,12 +30,15 @@ type DeviceManagerCheckpoint interface { GetData() ([]PodDevicesEntry, map[string][]string) } +// DevicesPerNUMA represents device ids obtained from device plugin per NUMA node id +type DevicesPerNUMA map[int64][]string + // PodDevicesEntry connects pod information to devices type PodDevicesEntry struct { PodUID string ContainerName string ResourceName string - DeviceIDs []string + DeviceIDs DevicesPerNUMA AllocResp []byte } @@ -52,6 +56,22 @@ type Data struct { Checksum checksum.Checksum } +// NewDevicesPerNUMA is a function that creates DevicesPerNUMA map +func NewDevicesPerNUMA() DevicesPerNUMA { + return make(DevicesPerNUMA) +} + +// Devices is a function that returns all device ids for all NUMA nodes +// and represent it as sets.String +func (dev DevicesPerNUMA) Devices() sets.String { + result := sets.NewString() + + for _, devs := range dev { + result.Insert(devs...) + } + return result +} + // New returns an instance of Checkpoint func New(devEntries []PodDevicesEntry, devices map[string][]string) DeviceManagerCheckpoint { diff --git a/pkg/kubelet/cm/devicemanager/manager.go b/pkg/kubelet/cm/devicemanager/manager.go index c374e1a7c7a..626f25da9ef 100644 --- a/pkg/kubelet/cm/devicemanager/manager.go +++ b/pkg/kubelet/cm/devicemanager/manager.go @@ -900,8 +900,17 @@ func (m *ManagerImpl) allocateContainerResources(pod *v1.Pod, container *v1.Cont return fmt.Errorf("no containers return in allocation response %v", resp) } + allocDevicesWithNUMA := checkpoint.NewDevicesPerNUMA() // Update internal cached podDevices state. - m.podDevices.insert(podUID, contName, resource, allocDevices, resp.ContainerResponses[0]) + m.mutex.Lock() + for dev := range allocDevices { + for idx := range m.allDevices[resource][dev].Topology.Nodes { + node := m.allDevices[resource][dev].Topology.Nodes[idx] + allocDevicesWithNUMA[node.ID] = append(allocDevicesWithNUMA[node.ID], dev) + } + } + m.mutex.Unlock() + m.podDevices.insert(podUID, contName, resource, allocDevicesWithNUMA, resp.ContainerResponses[0]) } if needsUpdateCheckpoint { diff --git a/pkg/kubelet/cm/devicemanager/manager_test.go b/pkg/kubelet/cm/devicemanager/manager_test.go index ca45ebc87e8..8f1ec0bd362 100644 --- a/pkg/kubelet/cm/devicemanager/manager_test.go +++ b/pkg/kubelet/cm/devicemanager/manager_test.go @@ -38,6 +38,7 @@ import ( pluginapi "k8s.io/kubelet/pkg/apis/deviceplugin/v1beta1" watcherapi "k8s.io/kubelet/pkg/apis/pluginregistration/v1" "k8s.io/kubernetes/pkg/kubelet/checkpointmanager" + "k8s.io/kubernetes/pkg/kubelet/cm/devicemanager/checkpoint" "k8s.io/kubernetes/pkg/kubelet/cm/topologymanager" "k8s.io/kubernetes/pkg/kubelet/config" "k8s.io/kubernetes/pkg/kubelet/lifecycle" @@ -432,10 +433,10 @@ func TestUpdateCapacityAllocatable(t *testing.T) { as.True(testManager.isDevicePluginResource(resourceName2)) } -func constructDevices(devices []string) sets.String { - ret := sets.NewString() +func constructDevices(devices []string) checkpoint.DevicesPerNUMA { + ret := checkpoint.DevicesPerNUMA{} for _, dev := range devices { - ret.Insert(dev) + ret[0] = append(ret[0], dev) } return ret } @@ -621,13 +622,11 @@ func getTestManager(tmpDir string, activePods ActivePodsFunc, testRes []TestReso activePods: activePods, sourcesReady: &sourcesReadyStub{}, checkpointManager: ckm, + allDevices: make(map[string]map[string]pluginapi.Device), } for _, res := range testRes { - testManager.healthyDevices[res.resourceName] = sets.NewString() - for _, dev := range res.devs { - testManager.healthyDevices[res.resourceName].Insert(dev) - } + testManager.healthyDevices[res.resourceName] = sets.NewString(res.devs.Devices().UnsortedList()...) if res.resourceName == "domain1.com/resource1" { testManager.endpoints[res.resourceName] = endpointInfo{ e: &MockEndpoint{allocateFunc: allocateStubFunc()}, @@ -657,6 +656,8 @@ func getTestManager(tmpDir string, activePods ActivePodsFunc, testRes []TestReso opts: nil, } } + testManager.allDevices[res.resourceName] = makeDevice(res.devs) + } return testManager, nil } @@ -664,19 +665,19 @@ func getTestManager(tmpDir string, activePods ActivePodsFunc, testRes []TestReso type TestResource struct { resourceName string resourceQuantity resource.Quantity - devs []string + devs checkpoint.DevicesPerNUMA } func TestPodContainerDeviceAllocation(t *testing.T) { res1 := TestResource{ resourceName: "domain1.com/resource1", resourceQuantity: *resource.NewQuantity(int64(2), resource.DecimalSI), - devs: []string{"dev1", "dev2"}, + devs: checkpoint.DevicesPerNUMA{0: []string{"dev1", "dev2"}}, } res2 := TestResource{ resourceName: "domain2.com/resource2", resourceQuantity: *resource.NewQuantity(int64(1), resource.DecimalSI), - devs: []string{"dev3", "dev4"}, + devs: checkpoint.DevicesPerNUMA{0: []string{"dev3", "dev4"}}, } testResources := make([]TestResource, 2) testResources = append(testResources, res1) @@ -767,12 +768,12 @@ func TestInitContainerDeviceAllocation(t *testing.T) { res1 := TestResource{ resourceName: "domain1.com/resource1", resourceQuantity: *resource.NewQuantity(int64(2), resource.DecimalSI), - devs: []string{"dev1", "dev2"}, + devs: checkpoint.DevicesPerNUMA{0: []string{"dev1", "dev2"}}, } res2 := TestResource{ resourceName: "domain2.com/resource2", resourceQuantity: *resource.NewQuantity(int64(1), resource.DecimalSI), - devs: []string{"dev3", "dev4"}, + devs: checkpoint.DevicesPerNUMA{0: []string{"dev3", "dev4"}}, } testResources := make([]TestResource, 2) testResources = append(testResources, res1) @@ -920,7 +921,7 @@ func TestDevicePreStartContainer(t *testing.T) { res1 := TestResource{ resourceName: "domain1.com/resource1", resourceQuantity: *resource.NewQuantity(int64(2), resource.DecimalSI), - devs: []string{"dev1", "dev2"}, + devs: checkpoint.DevicesPerNUMA{0: []string{"dev1", "dev2"}}, } as := require.New(t) podsStub := activePodsStub{ @@ -960,7 +961,7 @@ func TestDevicePreStartContainer(t *testing.T) { as.Contains(initializedDevs, "dev1") as.Contains(initializedDevs, "dev2") - as.Equal(len(initializedDevs), len(res1.devs)) + as.Equal(len(initializedDevs), res1.devs.Devices().Len()) expectedResps, err := allocateStubFunc()([]string{"dev1", "dev2"}) as.Nil(err) @@ -1057,3 +1058,13 @@ func allocateStubFunc() func(devs []string) (*pluginapi.AllocateResponse, error) return resps, nil } } + +func makeDevice(devOnNUMA checkpoint.DevicesPerNUMA) map[string]pluginapi.Device { + res := make(map[string]pluginapi.Device) + for node, devs := range devOnNUMA { + for idx := range devs { + res[devs[idx]] = pluginapi.Device{ID: devs[idx], Topology: &pluginapi.TopologyInfo{Nodes: []*pluginapi.NUMANode{{ID: node}}}} + } + } + return res +} diff --git a/pkg/kubelet/cm/devicemanager/pod_devices.go b/pkg/kubelet/cm/devicemanager/pod_devices.go index d3ffab80da4..6f640666211 100644 --- a/pkg/kubelet/cm/devicemanager/pod_devices.go +++ b/pkg/kubelet/cm/devicemanager/pod_devices.go @@ -30,7 +30,7 @@ import ( type deviceAllocateInfo struct { // deviceIds contains device Ids allocated to this container for the given resourceName. - deviceIds sets.String + deviceIds checkpoint.DevicesPerNUMA // allocResp contains cached rpc AllocateResponse. allocResp *pluginapi.ContainerAllocateResponse } @@ -70,7 +70,7 @@ func (pdev *podDevices) hasPod(podUID string) bool { return podExists } -func (pdev *podDevices) insert(podUID, contName, resource string, devices sets.String, resp *pluginapi.ContainerAllocateResponse) { +func (pdev *podDevices) insert(podUID, contName, resource string, devices checkpoint.DevicesPerNUMA, resp *pluginapi.ContainerAllocateResponse) { pdev.Lock() defer pdev.Unlock() if _, podExists := pdev.devs[podUID]; !podExists { @@ -108,7 +108,7 @@ func (pdev *podDevices) containerDevices(podUID, contName, resource string) sets if !resourceExists { return nil } - return devs.deviceIds + return devs.deviceIds.Devices() } // Populates allocatedResources with the device resources allocated to the specified . @@ -124,7 +124,7 @@ func (pdev *podDevices) addContainerAllocatedResources(podUID, contName string, return } for resource, devices := range resources { - allocatedResources[resource] = allocatedResources[resource].Union(devices.deviceIds) + allocatedResources[resource] = allocatedResources[resource].Union(devices.deviceIds.Devices()) } } @@ -141,7 +141,7 @@ func (pdev *podDevices) removeContainerAllocatedResources(podUID, contName strin return } for resource, devices := range resources { - allocatedResources[resource] = allocatedResources[resource].Difference(devices.deviceIds) + allocatedResources[resource] = allocatedResources[resource].Difference(devices.deviceIds.Devices()) } } @@ -157,7 +157,7 @@ func (pdev *podDevices) devices() map[string]sets.String { ret[resource] = sets.NewString() } if devices.allocResp != nil { - ret[resource] = ret[resource].Union(devices.deviceIds) + ret[resource] = ret[resource].Union(devices.deviceIds.Devices()) } } } @@ -173,7 +173,6 @@ func (pdev *podDevices) toCheckpointData() []checkpoint.PodDevicesEntry { for podUID, containerDevices := range pdev.devs { for conName, resources := range containerDevices { for resource, devices := range resources { - devIds := devices.deviceIds.UnsortedList() if devices.allocResp == nil { klog.Errorf("Can't marshal allocResp for %v %v %v: allocation response is missing", podUID, conName, resource) continue @@ -188,7 +187,7 @@ func (pdev *podDevices) toCheckpointData() []checkpoint.PodDevicesEntry { PodUID: podUID, ContainerName: conName, ResourceName: resource, - DeviceIDs: devIds, + DeviceIDs: devices.deviceIds, AllocResp: allocResp}) } } @@ -201,17 +200,13 @@ func (pdev *podDevices) fromCheckpointData(data []checkpoint.PodDevicesEntry) { for _, entry := range data { klog.V(2).Infof("Get checkpoint entry: %v %v %v %v %v\n", entry.PodUID, entry.ContainerName, entry.ResourceName, entry.DeviceIDs, entry.AllocResp) - devIDs := sets.NewString() - for _, devID := range entry.DeviceIDs { - devIDs.Insert(devID) - } allocResp := &pluginapi.ContainerAllocateResponse{} err := allocResp.Unmarshal(entry.AllocResp) if err != nil { klog.Errorf("Can't unmarshal allocResp for %v %v %v: %v", entry.PodUID, entry.ContainerName, entry.ResourceName, err) continue } - pdev.insert(entry.PodUID, entry.ContainerName, entry.ResourceName, devIDs, allocResp) + pdev.insert(entry.PodUID, entry.ContainerName, entry.ResourceName, entry.DeviceIDs, allocResp) } } @@ -328,10 +323,13 @@ func (pdev *podDevices) getContainerDevices(podUID, contName string) []*podresou } cDev := []*podresourcesapi.ContainerDevices{} for resource, allocateInfo := range pdev.devs[podUID][contName] { - cDev = append(cDev, &podresourcesapi.ContainerDevices{ - ResourceName: resource, - DeviceIds: allocateInfo.deviceIds.UnsortedList(), - }) + for numaid, devlist := range allocateInfo.deviceIds { + cDev = append(cDev, &podresourcesapi.ContainerDevices{ + ResourceName: resource, + DeviceIds: devlist, + Topology: &podresourcesapi.TopologyInfo{Nodes: []*podresourcesapi.NUMANode{{ID: numaid}}}, + }) + } } return cDev } diff --git a/pkg/kubelet/cm/devicemanager/pod_devices_test.go b/pkg/kubelet/cm/devicemanager/pod_devices_test.go new file mode 100644 index 00000000000..b2ff8376a73 --- /dev/null +++ b/pkg/kubelet/cm/devicemanager/pod_devices_test.go @@ -0,0 +1,47 @@ +/* +Copyright 2020 The Kubernetes Authors. + +Licensed under the Apache License, Version 2.0 (the "License"); +you may not use this file except in compliance with the License. +You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, software +distributed under the License is distributed on an "AS IS" BASIS, +WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +See the License for the specific language governing permissions and +limitations under the License. +*/ + +package devicemanager + +import ( + "testing" + + "github.com/stretchr/testify/require" + + "k8s.io/kubernetes/pkg/kubelet/cm/devicemanager/checkpoint" +) + +func TestGetContainerDevices(t *testing.T) { + podDevices := newPodDevices() + resourceName1 := "domain1.com/resource1" + podID := "pod1" + contID := "con1" + devices := checkpoint.DevicesPerNUMA{0: []string{"dev1"}, 1: []string{"dev1"}} + + podDevices.insert(podID, contID, resourceName1, + devices, + constructAllocResp(map[string]string{"/dev/r1dev1": "/dev/r1dev1", "/dev/r1dev2": "/dev/r1dev2"}, map[string]string{"/home/r1lib1": "/usr/r1lib1"}, map[string]string{})) + + contDevices := podDevices.getContainerDevices(podID, contID) + require.Equal(t, len(devices), len(contDevices), "Incorrect container devices") + for _, contDev := range contDevices { + for _, node := range contDev.Topology.Nodes { + dev, ok := devices[node.ID] + require.True(t, ok, "NUMA id %v doesn't exist in result", node.ID) + require.Equal(t, contDev.DeviceIds[0], dev[0], "Can't find device %s in result", dev[0]) + } + } +} diff --git a/pkg/kubelet/cm/devicemanager/topology_hints_test.go b/pkg/kubelet/cm/devicemanager/topology_hints_test.go index 5f2db322953..84f000a8cc6 100644 --- a/pkg/kubelet/cm/devicemanager/topology_hints_test.go +++ b/pkg/kubelet/cm/devicemanager/topology_hints_test.go @@ -404,7 +404,7 @@ func TestGetTopologyHints(t *testing.T) { for p := range tc.allocatedDevices { for c := range tc.allocatedDevices[p] { for r, devices := range tc.allocatedDevices[p][c] { - m.podDevices.insert(p, c, r, sets.NewString(devices...), nil) + m.podDevices.insert(p, c, r, constructDevices(devices), nil) m.allocatedDevices[r] = sets.NewString() for _, d := range devices { diff --git a/pkg/kubelet/kubelet.go b/pkg/kubelet/kubelet.go index d6d620790c0..f8f0fc0a0ef 100644 --- a/pkg/kubelet/kubelet.go +++ b/pkg/kubelet/kubelet.go @@ -2193,7 +2193,7 @@ func (kl *Kubelet) ListenAndServePodResources() { klog.V(2).Infof("Failed to get local endpoint for PodResources endpoint: %v", err) return } - server.ListenAndServePodResources(socket, kl.podManager, kl.containerManager) + server.ListenAndServePodResources(socket, kl.podManager, kl.containerManager, kl.containerManager) } // Delete the eligible dead container instances in a pod. Depending on the configuration, the latest dead containers may be kept around. diff --git a/pkg/kubelet/server/server.go b/pkg/kubelet/server/server.go index 23e7c362afc..fb8c7e12e7c 100644 --- a/pkg/kubelet/server/server.go +++ b/pkg/kubelet/server/server.go @@ -179,10 +179,10 @@ func ListenAndServeKubeletReadOnlyServer(host HostInterface, resourceAnalyzer st } // ListenAndServePodResources initializes a gRPC server to serve the PodResources service -func ListenAndServePodResources(socket string, podsProvider podresources.PodsProvider, devicesProvider podresources.DevicesProvider) { +func ListenAndServePodResources(socket string, podsProvider podresources.PodsProvider, devicesProvider podresources.DevicesProvider, cpusProvider podresources.CPUsProvider) { server := grpc.NewServer() podresourcesapiv1alpha1.RegisterPodResourcesListerServer(server, podresources.NewV1alpha1PodResourcesServer(podsProvider, devicesProvider)) - podresourcesapi.RegisterPodResourcesListerServer(server, podresources.NewV1PodResourcesServer(podsProvider, devicesProvider)) + podresourcesapi.RegisterPodResourcesListerServer(server, podresources.NewV1PodResourcesServer(podsProvider, devicesProvider, cpusProvider)) l, err := util.CreateListener(socket) if err != nil { klog.Fatalf("Failed to create listener for podResources endpoint: %v", err)