Is there any optimization we can do when using preemptible instances running jobs.

[zwzhang0107]

/kind question

Question:

Most cloud providers have provide preemptible instance with better cost efficiency(e.g. Alibaba Cloud ECI).

In old days before k8s, spark added better handling for node shutdown, for better use of EC2 spot node. Recently, there are also some discussions in ASPLOS'24.

I'm wondering is there any work (maybe already done?) to make jobs better use of preemptible instances. I will be honored to join if there is any plan.

[juliusvonkohout]

I think that is for kubeflow/manifests and yes it is possible.

You can join the CNCF Slack and access our meetings at the Kubeflow Community website. Our channel on the CNCF Slack is here #kubeflow-platform. You can also find there our biweekly meetings, including the commentable Agenda.

/close

[google-oss-prow]

@juliusvonkohout: Closing this issue.

In response to this:

I think that is for kubeflow/manifests and yes it is possible.

You can join the CNCF Slack and access our meetings at the Kubeflow Community website. Our channel on the CNCF Slack is here #kubeflow-platform. You can also find there our biweekly meetings, including the commentable Agenda.

/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.