StatefulSet without Service

[pohly]

Is this a BUG REPORT or FEATURE REQUEST?:

/kind feature

What happened:

Developers have started to use StatefulSet without a Service definition as a way to deploy a pod on a cluster exactly once, for example here: https://github.com/kubernetes-sigs/gcp-compute-persistent-disk-csi-driver/blob/master/deploy/kubernetes/stable/controller.yaml

Conceptually that makes sense when the app running in the pod doesn't accept any connections from outside. In this example, it's the CSI sidecar containers which react to changes in the apiserver.

But is this a supported mode of operation for a StatefulSet? https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.12/#statefulset-v1-apps says that

serviceName is the name of the service that governs this StatefulSet. This service must exist before the StatefulSet, and is responsible for the network identity of the set. Pods get DNS/hostnames that follow the pattern: pod-specific-string.serviceName.default.svc.cluster.local where "pod-specific-string" is managed by the StatefulSet controller.

serviceName is mandatory. The example above gets around this by setting serviceName without defining a corresponding Service.

The concern (raised by @rootfs in ceph/ceph-csi#81) is that while it currently works, future revisions of the StatefulSet controller might not support this anymore.

What you expected to happen:

Clarify in the docs that the Service definition is not needed or (better?) make serviceName optional.

Environment:

• Kubernetes version (use kubectl version): tested on 1.11 and 1.12

[pohly]

/sig apps

[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

[fejta-bot]

Stale issues rot after 30d of inactivity.
Mark the issue as fresh with /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.

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 rotten

[pohly]

/remove-lifecycle stale

Still a valid question...

[kfox1111]

+1. I'd like to know this too. Having statefulsets without a service makes sense to me if they don't care about a stable identity, just that there is no more then 1.

[pohly]

/remove-lifecycle rotten

[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

[pohly]

/remove-lifecycle stale

[Noah-Huppert]

Would a headless service be the solution to this problem?

Kubernetes Docs:

Sometimes you don’t need load-balancing and a single Service IP. In this case, you can create what are termed “headless” Services, by explicitly specifying "None" for the cluster IP (.spec.clusterIP).

[pohly]

Noah Huppert <notifications@github.com> writes:

Would a headless service be the solution to this problem?

It's better, but still a useless object that could be avoided if we could get clarified that a StatefulSet is okay to use without a Service.

[kfox1111]

What is the status of this?

[marcelloromani]

@Noah-Huppert Yes, a headless service was the solution for me. From what I understand it's the only workaround at the moment.

[kfox1111]

I've launched statefulsets without services and it seems to work ok. We're just looking for a guarantee that this will continue to be safe going forward. The docs say its required, but the implementation does not enforce it last I looked.

[kow3ns]

We will always support this. You can close this issue. Please feel free to open an issue against the docs repository, or (if you are feeling generous) a PR against the same.

[kfox1111]

From sig-apps channel:
@kowens
@kfox1111 if you don’t need DNS for the Pod it is supported

@kowens
the reason the docs are written as is is for conceptually simplicity. You do need a Service if you want to generate a CNAME and SRV records for the Pods
it is the most common use case

So, they won't break serviceless statefulsets.

[fejta-bot]

Stale issues rot after 30d of inactivity.
Mark the issue as fresh with /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.

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

Send feedback to sig-contributor-experience at kubernetes/community.
/lifecycle rotten

[TBBle]

/remove-lifecycle rotten

#69608 (comment) still seems to be the state of this issue.

[k8s-triage-robot]

The Kubernetes project currently lacks enough contributors to adequately respond to all issues and PRs.

This bot triages issues and PRs according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue or PR as fresh with /remove-lifecycle stale
• Mark this issue or PR as rotten with /lifecycle rotten
• Close this issue or PR with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale

[TBBle]

At this point the two CSI examples from the original ticket have gone away (see kubernetes-sigs/gcp-compute-persistent-disk-csi-driver#521 and ceph/ceph-csi#414 which replaced the StatefulSet singletons with Deployments using leader-election).

Are there any practical/current use-cases for StatefulSet without a backing Service? The ticket's approaching three years old, and has not yet been triaged, suggesting there's not much appetite/motivation to advanced the proposed change in #69608 (comment).

[pohly]

This is still relevant, for example here: https://github.com/kubernetes-csi/csi-driver-host-path/blob/c480b671f63c142defd2180a6ca68f85327c331f/deploy/kubernetes-1.21/hostpath/csi-hostpath-plugin.yaml#L189-L199

The Service referenced there never gets created because this issue clarified that this is okay.

[TBBle]

/remove-lifecycle stale

[k8s-triage-robot]

The Kubernetes project currently lacks enough contributors to adequately respond to all issues and PRs.

This bot triages issues and PRs according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue or PR as fresh with /remove-lifecycle stale
• Mark this issue or PR as rotten with /lifecycle rotten
• Close this issue or PR with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale

[k8s-triage-robot]

The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs.

This bot triages issues and PRs according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue or PR as fresh with /remove-lifecycle rotten
• Close this issue or PR with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle rotten

[kfox1111]

/remove-lifecycle stale

[k8s-triage-robot]

The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs.

This bot triages issues and PRs according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Reopen this issue or PR with /reopen
• Mark this issue or PR as fresh with /remove-lifecycle rotten
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/close

[k8s-ci-robot]

@k8s-triage-robot: Closing this issue.

In response to this:

The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs.

This bot triages issues and PRs according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Reopen this issue or PR with /reopen
• Mark this issue or PR as fresh with /remove-lifecycle rotten
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/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.

[TBBle]

/reopen

#69608 (comment) intended to keep this ticket active, but named the wrong lifecycle, I assume by accident.

[k8s-ci-robot]

@TBBle: You can't reopen an issue/PR unless you authored it or you are a collaborator.

In response to this:

/reopen

#69608 (comment) intended to keep this ticket active, but named the wrong lifecycle, I assume by accident.

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.

[djschny]

/reopen

[k8s-ci-robot]

@djschny: You can't reopen an issue/PR unless you authored it or you are a collaborator.

In response to this:

/reopen

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.