github/kubernetes
1
0
Fork 0
mirror of https://github.com/k3s-io/kubernetes.git synced 2025-08-25 11:31:44 +00:00
Code Issues Projects Releases Wiki Activity

Merge pull request #129685 from swatisehgal/cpu-mgr-logs-improvements

Browse Source 
CPU Manager logging improvements
This commit is contained in:
Kubernetes Prow Robot 2025-02-07 03:50:02 -08:00 committed by GitHub
commit 20b12ad5c3
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 15 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
pkg/kubelet/cm/cpumanager/cpu_manager.go
View File

@ -239,6 +239,8 @@ func (m *manager) Start(activePods ActivePodsFunc, sourcesReady config.SourcesRe
return err
}
klog.V(4).InfoS("CPU manager started", "policy", m.policy.Name())
m.allocatableCPUs = m.policy.GetAllocatableCPUs(m.state)
if m.policy.Name() == string(PolicyNone) {
@ -465,7 +467,7 @@ func (m *manager) reconcileState() (success []reconciledContainer, failure []rec
cset := m.state.GetCPUSetOrDefault(string(pod.UID), container.Name)
if cset.IsEmpty() {
// NOTE: This should not happen outside of tests.
klog.V(2).InfoS("ReconcileState: skipping container; assigned cpuset is empty", "pod", klog.KObj(pod), "containerName", container.Name)
klog.V(2).InfoS("ReconcileState: skipping container; empty cpuset assigned", "pod", klog.KObj(pod), "containerName", container.Name)
failure = append(failure, reconciledContainer{pod.Name, container.Name, containerID})
continue
}

9
pkg/kubelet/cm/cpumanager/policy_static.go
View File

@ -388,6 +388,7 @@ func (p *staticPolicy) Allocate(s state.State, pod *v1.Pod, container *v1.Contai
p.updateCPUsToReuse(pod, container, cpuset)
p.updateMetricsOnAllocate(cpuset)
klog.V(4).InfoS("Allocated exclusive CPUs", "pod", klog.KObj(pod), "containerName", container.Name, "cpuset", cpuset)
return nil
}
@ -455,7 +456,9 @@ func (p *staticPolicy) allocateCPUs(s state.State, numCPUs int, numaAffinity bit
}
func (p *staticPolicy) guaranteedCPUs(pod *v1.Pod, container *v1.Container) int {
if v1qos.GetPodQOS(pod) != v1.PodQOSGuaranteed {
qos := v1qos.GetPodQOS(pod)
if qos != v1.PodQOSGuaranteed {
klog.V(5).InfoS("Exclusive CPU allocation skipped, pod QoS is not guaranteed", "pod", klog.KObj(pod), "containerName", container.Name, "qos", qos)
return 0
}
cpuQuantity := container.Resources.Requests[v1.ResourceCPU]
@ -474,7 +477,9 @@ func (p *staticPolicy) guaranteedCPUs(pod *v1.Pod, container *v1.Container) int
cpuQuantity = cs.AllocatedResources[v1.ResourceCPU]
}
}
if cpuQuantity.Value()*1000 != cpuQuantity.MilliValue() {
cpuValue := cpuQuantity.Value()
if cpuValue*1000 != cpuQuantity.MilliValue() {
klog.V(5).InfoS("Exclusive CPU allocation skipped, pod requested non-integral CPUs", "pod", klog.KObj(pod), "containerName", container.Name, "cpu", cpuValue)
return 0
}
// Safe downcast to do for all systems with < 2.1 billion CPUs.

10
pkg/kubelet/cm/cpumanager/state/state_checkpoint.go
View File

@ -201,7 +201,7 @@ func (sc *stateCheckpoint) SetCPUSet(podUID string, containerName string, cset c
sc.cache.SetCPUSet(podUID, containerName, cset)
err := sc.storeState()
if err != nil {
klog.InfoS("Store state to checkpoint error", "err", err)
klog.ErrorS(err, "Failed to store state to checkpoint", "podUID", podUID, "containerName", containerName)
}
}
@ -212,7 +212,7 @@ func (sc *stateCheckpoint) SetDefaultCPUSet(cset cpuset.CPUSet) {
sc.cache.SetDefaultCPUSet(cset)
err := sc.storeState()
if err != nil {
klog.InfoS("Store state to checkpoint error", "err", err)
klog.ErrorS(err, "Failed to store state to checkpoint")
}
}
@ -223,7 +223,7 @@ func (sc *stateCheckpoint) SetCPUAssignments(a ContainerCPUAssignments) {
sc.cache.SetCPUAssignments(a)
err := sc.storeState()
if err != nil {
klog.InfoS("Store state to checkpoint error", "err", err)
klog.ErrorS(err, "Failed to store state to checkpoint")
}
}
@ -234,7 +234,7 @@ func (sc *stateCheckpoint) Delete(podUID string, containerName string) {
sc.cache.Delete(podUID, containerName)
err := sc.storeState()
if err != nil {
klog.InfoS("Store state to checkpoint error", "err", err)
klog.ErrorS(err, "Failed to store state to checkpoint", "podUID", podUID, "containerName", containerName)
}
}
@ -245,6 +245,6 @@ func (sc *stateCheckpoint) ClearState() {
sc.cache.ClearState()
err := sc.storeState()
if err != nil {
klog.InfoS("Store state to checkpoint error", "err", err)
klog.ErrorS(err, "Failed to store state to checkpoint")
}
}