v1.Extender.HTTPTimeout is a time.Duration

[mikedanese]

kubernetes/staging/src/k8s.io/kube-scheduler/config/v1/types.go

Lines 210 to 212 in 2ab6357

	// HTTPTimeout specifies the timeout duration for a call to the extender. Filter timeout fails the scheduling of the pod. Prioritize
	// timeout is ignored, k8s/other extenders priorities are used to select the node.
	HTTPTimeout time.Duration `json:"httpTimeout,omitempty"`

This requires specification as int64 nanoseconds so a one second timeout would need to be specified as 1000000000. It should be a metav1.Duration like other durations in the API but it's been like this since 1.2. Is anyone actually using this field?

[angao]

@alculquicondor WDYT? I can do this update if it is accepted.

[alculquicondor]

Any updates in here requires a new API version.
@mikedanese wasn't there a related issues already opened for this?
Can anybody work on a KEP?

[angao]

@alculquicondor I can write a KEP for this.

[gavinfish]

/cc

[mikedanese]

@alculquicondor, not sure if there is a releated issue. I noticed it working on openapi stuff.

[alculquicondor]

Sorry, I confused you with @damemi who had done some work in 1.18 to make our API compliant.

[damemi]

@mikedanese should it be a metav1.Duration? Or an integer value? I've noticed in the API conventions that using Duration in the API isn't advised in order to keep compatibility with non-go clients (https://github.com/kubernetes/community/blob/master/contributors/devel/sig-architecture/api-conventions.md#units)

[damemi]

(cross-ref to the open issue to improve the extender API: #88634)

[liggitt]

As a config API, this isn't served anywhere, so the only consumer is the go scheduler binary. Other config APIs use metav1.Duration or a plain integer with seconds in the field name to clarify the units.

[damemi]

Thanks @liggitt, I've been confused on what the distinction was

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[alculquicondor]

We just fixed this for KubeSchedulerConfiguration (going for v1beta1 in 1.19) #91625

Once this is GA, the Policy API (which uses this Extender type) will be removed #92143

[alculquicondor]

/close

[k8s-ci-robot]

@alculquicondor: Closing this issue.

In response to this:

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.