A mechanism to list the Workloads that are admitted and not finished

[alculquicondor]

What would you like to be added:

Some way of querying the list of Workloads that are still "running", that is, they are admitted and don't have a Finished condition. Querying by ClusterQueue would be ideal too.

Some options (not necessarily exclusive to each other):

• The kueue CLI
• A visibility endpoint.
• Is there a way to do this just with a kubectl jsonpath?

Why is this needed:

For basic diagnostics of the state of the cluster.

Completion requirements:

This enhancement requires the following artifacts:

• Design doc
• API change?
• Docs update

The artifacts should be linked in subsequent comments.

[tenzen-y]

Do you expect that we extend the on-demand visibility server or save information on any CustomResource like ClusterQueue?

[alculquicondor]

I don't think we should save it. This is more for on-demand queries.

[tenzen-y]

I don't think we should save it. This is more for on-demand queries.

It makes sense.

[astefanutti]

Tangentially, do you think adding a running_worloads metric could be useful too?

[tenzen-y]

Tangentially, do you think adding a running_worloads metric could be useful too?

Does it mean Prometheus metric?

[astefanutti]

Tangentially, do you think adding a running_worloads metric could be useful too?

Does it mean Prometheus metric?

Yes, a Prometheus metric, that would complement the existing pending_workloads metric.

[alculquicondor]

we already have it

kueue/pkg/metrics/metrics.go

Lines 120 to 126 in 9afd7e6

	AdmittedActiveWorkloads = prometheus.NewGaugeVec(
	prometheus.GaugeOpts{
	Subsystem: constants.KueueName,
	Name: "admitted_active_workloads",
	Help: "The number of admitted Workloads that are active (unsuspended and not finished), per 'cluster_queue'",
	}, []string{"cluster_queue"},
	)

[astefanutti]

we already have it

kueue/pkg/metrics/metrics.go

Lines 120 to 126 in 9afd7e6

	AdmittedActiveWorkloads = prometheus.NewGaugeVec(
	prometheus.GaugeOpts{
	Subsystem: constants.KueueName,
	Name: "admitted_active_workloads",
	Help: "The number of admitted Workloads that are active (unsuspended and not finished), per 'cluster_queue'",
	}, []string{"cluster_queue"},
	)

Ah right, this is the one, thanks!

[alculquicondor]

Would you like to propose something for this area? Maybe along the lines of a new visibility endpoint?

[astefanutti]

Yes. Do you want it to be initially proposed as a KEP?

[alculquicondor]

Given that it's an API change, a KEP would be good.

[tenzen-y]

Feel free to assign yourself with /assign.
Since you joined this org, the command should work well :)

[KunWuLuan]

If no one working for this, I can help.

[astefanutti]

@KunWuLuan I haven't got the chance to work on it. Feel free to assign it to you.

[KunWuLuan]

Do we still need the new endpoints if we already have the KueueCtl?

[alculquicondor]

Maybe less so? One slight advantage would be to filter on the server side.

But maybe it would add unnecessary load to the kueue binary?

[k8s-triage-robot]

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

This bot triages un-triaged issues 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 as fresh with /remove-lifecycle stale
• Close this issue with /close
• Offer to help out with Issue Triage

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

/lifecycle stale

[tenzen-y]

/remove-lifecycle stale

[alculquicondor]

This is supported using kueuectl https://kueue.sigs.k8s.io/docs/reference/kubectl-kueue/commands/kueuectl_list/kueuectl_list_workload/

Should we close this?

[tenzen-y]

This is supported using kueuectl https://kueue.sigs.k8s.io/docs/reference/kubectl-kueue/commands/kueuectl_list/kueuectl_list_workload/

Should we close this?

Oh, you're right. Let me close this.
/close

[k8s-ci-robot]

@tenzen-y: Closing this issue.

In response to this:

This is supported using kueuectl https://kueue.sigs.k8s.io/docs/reference/kubectl-kueue/commands/kueuectl_list/kueuectl_list_workload/

Should we close this?

Oh, you're right. Let me close 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-sigs/prow repository.