search

GoogleCloudPlatform / kubeflow-distribution

Blueprints for Deploying Kubeflow on Google Cloud Platform and Anthos
Apache License 2.0
79 stars 63 forks source link

[Feature] Support private GKE cluster on Google Cloud #267

Open zijianjoy opened 3 years ago

zijianjoy commented 3 years ago

This is a survey form for collecting interest on the feature: Private Kubeflow Cluster

Description

GKE allows you to create private GKE cluster http://cloud/kubernetes-engine/docs/how-to/private-clusters which doesn't expose external IP address to the public. It requires customization and deployment steps change in kubeflow/gcp-blueprints.

Action

If this feature fits in to your use case, please upvote this GitHub issue. We will prioritize this work accordingly, also welcome contributor from the community.

kungfuchicken commented 3 years ago

tracking.

markcberman commented 3 years ago

Are there any instructions for deploying latest stable KF version to a private GKE cluster. My company's firewall rules block access to the public internet.

4j4y commented 3 years ago

Any update on this?

bsikander commented 2 years ago

I think we should clarify here and in the documentation that the managed version of kubeflow pipelines (AI Platform Pipelines) on GCP, do work with private GKE cluster. Only custom deployment of kubeflow on private GKE requires manual steps. For new comers, this is totally not clear.

zijianjoy commented 2 years ago

@bsikander That is a good point. However, AI Platform Pipelines is the managed version of Kubeflow Pipelines standalone, not Full fledged Kubeflow. Therefore AI Platform Pipelines guidance lives in https://www.kubeflow.org/docs/components/pipelines/installation/overview/#google-cloud-ai-platform-pipelines. Feel free to contribute documentation there to explain that AI Platform Pipelines works with private GKE cluster. The Distribution section is more about Full fledged Kubeflow instructions.

bhack commented 10 months ago

Any update on this?