Promote kubelet graceful node shutdown to beta

- Change the feature gate from alpha to beta and enable it by default

- Update a few of the unit tests due to feature gate being enabled by
  default

- Small refactor in `nodeshutdown_manager` which adds `featureEnabled`
  function (which checks that feature gate and that
  `kubeletConfig.ShutdownGracePeriod > 0`).

- Use `featureEnabled()` to exit early from shutdown manager in the case
  that the feature is disabled

- Update kubelet config defaulting to be explicit that
  `ShutdownGracePeriod` and `ShutdownGracePeriodCriticalPods` default to
  zero and update the godoc comments.

- Update defaults and add featureGate tag in api config godoc.

With this feature now in beta and the feature gate enabled by default,
to enable graceful shutdown all that will be required is to configure
`ShutdownGracePeriod` and `ShutdownGracePeriodCriticalPods` in the
kubelet config. If not configured, they will be defaulted to zero, and
graceful shutdown will effectively be disabled.
This commit is contained in:
David Porter
2021-03-02 15:41:50 -08:00
parent 076bd6c401
commit 893f5fd4f0
7 changed files with 71 additions and 12 deletions

View File

@@ -379,11 +379,15 @@ type KubeletConfiguration struct {
// EnableSystemLogHandler enables /logs handler.
EnableSystemLogHandler bool
// ShutdownGracePeriod specifies the total duration that the node should delay the shutdown and total grace period for pod termination during a node shutdown.
// Defaults to 30 seconds, requires GracefulNodeShutdown feature gate to be enabled.
// Defaults to 0 seconds.
// +featureGate=GracefulNodeShutdown
// +optional
ShutdownGracePeriod metav1.Duration
// ShutdownGracePeriodCriticalPods specifies the duration used to terminate critical pods during a node shutdown. This should be less than ShutdownGracePeriod.
// Defaults to 10 seconds, requires GracefulNodeShutdown feature gate to be enabled.
// Defaults to 0 seconds.
// For example, if ShutdownGracePeriod=30s, and ShutdownGracePeriodCriticalPods=10s, during a node shutdown the first 20 seconds would be reserved for gracefully terminating normal pods, and the last 10 seconds would be reserved for terminating critical pods.
// +featureGate=GracefulNodeShutdown
// +optional
ShutdownGracePeriodCriticalPods metav1.Duration
// ReservedMemory specifies a comma-separated list of memory reservations for NUMA nodes.
// The parameter makes sense only in the context of the memory manager feature. The memory manager will not allocate reserved memory for container workloads.

View File

@@ -100,7 +100,6 @@ func TestValidateKubeletConfiguration(t *testing.T) {
ShutdownGracePeriodCriticalPods: metav1.Duration{Duration: 0},
FeatureGates: map[string]bool{
"CustomCPUCFSQuotaPeriod": true,
"GracefulNodeShutdown": true,
},
}
if allErrors := ValidateKubeletConfiguration(successCase2); allErrors != nil {
@@ -133,7 +132,7 @@ func TestValidateKubeletConfiguration(t *testing.T) {
NodeLeaseDurationSeconds: -1,
CPUCFSQuotaPeriod: metav1.Duration{Duration: 100 * time.Millisecond},
ShutdownGracePeriod: metav1.Duration{Duration: 30 * time.Second},
ShutdownGracePeriodCriticalPods: metav1.Duration{Duration: 10 * time.Second},
ShutdownGracePeriodCriticalPods: metav1.Duration{Duration: 60 * time.Second},
}
const numErrsErrorCase1 = 28
if allErrors := ValidateKubeletConfiguration(errorCase1); len(allErrors.(utilerrors.Aggregate).Errors()) != numErrsErrorCase1 {