github/kubernetes
1
0
Fork 0
mirror of https://github.com/k3s-io/kubernetes.git synced 2025-07-19 09:52:49 +00:00
Code Issues Projects Releases Wiki Activity

node: topologymgr: add metrics about admission requests and errors

Browse Source 
Signed-off-by: Swati Sehgal <swsehgal@redhat.com>
This commit is contained in:
Swati Sehgal 2023-01-17 17:39:21 +00:00
parent 5550064bc2
commit 172c55d310
4 changed files with 35 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
pkg/kubelet/cm/topologymanager/scope_container.go
View File

@ -22,6 +22,7 @@ import (
"k8s.io/kubernetes/pkg/kubelet/cm/admission"
"k8s.io/kubernetes/pkg/kubelet/cm/containermap"
"k8s.io/kubernetes/pkg/kubelet/lifecycle"
"k8s.io/kubernetes/pkg/kubelet/metrics"
)
type containerScope struct {
@ -54,6 +55,7 @@ func (s *containerScope) Admit(pod *v1.Pod) lifecycle.PodAdmitResult {
klog.InfoS("Best TopologyHint", "bestHint", bestHint, "pod", klog.KObj(pod), "containerName", container.Name)
if !admit {
metrics.TopologyManagerAdmissionErrorsTotal.Inc()
return admission.GetPodAdmitResult(&TopologyAffinityError{})
}
klog.InfoS("Topology Affinity", "bestHint", bestHint, "pod", klog.KObj(pod), "containerName", container.Name)
@ -61,6 +63,7 @@ func (s *containerScope) Admit(pod *v1.Pod) lifecycle.PodAdmitResult {
err := s.allocateAlignedResources(pod, &container)
if err != nil {
metrics.TopologyManagerAdmissionErrorsTotal.Inc()
return admission.GetPodAdmitResult(err)
}
}

3
pkg/kubelet/cm/topologymanager/scope_pod.go
View File

@ -22,6 +22,7 @@ import (
"k8s.io/kubernetes/pkg/kubelet/cm/admission"
"k8s.io/kubernetes/pkg/kubelet/cm/containermap"
"k8s.io/kubernetes/pkg/kubelet/lifecycle"
"k8s.io/kubernetes/pkg/kubelet/metrics"
)
type podScope struct {
@ -52,6 +53,7 @@ func (s *podScope) Admit(pod *v1.Pod) lifecycle.PodAdmitResult {
bestHint, admit := s.calculateAffinity(pod)
klog.InfoS("Best TopologyHint", "bestHint", bestHint, "pod", klog.KObj(pod))
if !admit {
metrics.TopologyManagerAdmissionErrorsTotal.Inc()
return admission.GetPodAdmitResult(&TopologyAffinityError{})
}
@ -61,6 +63,7 @@ func (s *podScope) Admit(pod *v1.Pod) lifecycle.PodAdmitResult {
err := s.allocateAlignedResources(pod, &container)
if err != nil {
metrics.TopologyManagerAdmissionErrorsTotal.Inc()
return admission.GetPodAdmitResult(err)
}
}

3
pkg/kubelet/cm/topologymanager/topology_manager.go
View File

@ -24,6 +24,7 @@ import (
"k8s.io/klog/v2"
"k8s.io/kubernetes/pkg/kubelet/cm/topologymanager/bitmask"
"k8s.io/kubernetes/pkg/kubelet/lifecycle"
"k8s.io/kubernetes/pkg/kubelet/metrics"
)
const (
@ -208,6 +209,8 @@ func (m *manager) RemoveContainer(containerID string) error {
func (m *manager) Admit(attrs *lifecycle.PodAdmitAttributes) lifecycle.PodAdmitResult {
klog.InfoS("Topology Admit Handler")
metrics.TopologyManagerAdmissionRequestsTotal.Inc()
pod := attrs.Pod
return m.scope.Admit(pod)

26
pkg/kubelet/metrics/metrics.go
View File

@ -91,6 +91,10 @@ const (
CPUManagerPinningRequestsTotalKey = "cpu_manager_pinning_requests_total"
CPUManagerPinningErrorsTotalKey = "cpu_manager_pinning_errors_total"
// Metrics to track the Topology manager behavior
TopologyManagerAdmissionRequestsTotalKey = "topology_manager_admission_requests_total"
TopologyManagerAdmissionErrorsTotalKey = "topology_manager_admission_errors_total"
// Values used in metric labels
Container = "container"
InitContainer = "init_container"
@ -549,6 +553,26 @@ var (
StabilityLevel: metrics.ALPHA,
},
)
// TopologyManagerAdmissionRequestsTotal tracks the number of times the pod spec will cause the topology manager to admit a pod
TopologyManagerAdmissionRequestsTotal = metrics.NewCounter(
&metrics.CounterOpts{
Subsystem: KubeletSubsystem,
Name: TopologyManagerAdmissionRequestsTotalKey,
Help: "The number of admission requests where resources have to be aligned.",
StabilityLevel: metrics.ALPHA,
},
)
// TopologyManagerAdmissionErrorsTotal tracks the number of times the pod spec required the topology manager to admit a pod, but the admission failed
TopologyManagerAdmissionErrorsTotal = metrics.NewCounter(
&metrics.CounterOpts{
Subsystem: KubeletSubsystem,
Name: TopologyManagerAdmissionErrorsTotalKey,
Help: "The number of admission request failures where resources could not be aligned.",
StabilityLevel: metrics.ALPHA,
},
)
)
var registerMetrics sync.Once
@ -600,6 +624,8 @@ func Register(collectors ...metrics.StableCollector) {
legacyregistry.MustRegister(RunPodSandboxErrors)
legacyregistry.MustRegister(CPUManagerPinningRequestsTotal)
legacyregistry.MustRegister(CPUManagerPinningErrorsTotal)
legacyregistry.MustRegister(TopologyManagerAdmissionRequestsTotal)
legacyregistry.MustRegister(TopologyManagerAdmissionErrorsTotal)
for _, collector := range collectors {
legacyregistry.CustomMustRegister(collector)