Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Failed creating kubeflow cluster because VPC is not identified #337

Open
sidiksoleman opened this issue Dec 10, 2021 · 2 comments
Open

Failed creating kubeflow cluster because VPC is not identified #337

sidiksoleman opened this issue Dec 10, 2021 · 2 comments
Assignees

Comments

@sidiksoleman
Copy link

We used VPC for our gcp project. Thus, we modified the cluster.yaml for both management cluster and kubeflow cluster.

In management cluster, we add two specification, i.e.
network:
subnetwork:

I tried to use the specification here: https://cloud.google.com/config-connector/docs/reference/resource-docs/container/containercluster#vpc_native_container_cluster, but it is not working for deploying management cluster. While the above one is working well and the management cluster was successfully deployed.

However, when I use the same yaml format for deploying kubeflow cluster, the above format cannot be recognized. Thus, I used the format from GCP, i.e.
networkRef:
name:
subnetworkRef:
name:
But, I encountered another problem, i.e. ComputerNetwork is not found.

Did I miss something? Why we can successfully deployed management cluster using different format than the one mentioned in the GCP page. However, although the format is correct, but why it cannot find the VPC while it was not a problem while deploying management cluster.

Thank you.

Kubeflow version: 1.4.0
OS: ubuntu 18.04
Google Cloud SDK: 366.0.0
beta: 2021.12.03
bq: 2.0.72
core: 2021.12.03
gsutil: 5.5
kpt: 1.0.0-beta.7
kubectl: 1.20.8
kustomize: 4.4.0

@zijianjoy
Copy link
Collaborator

Hello @sidiksoleman , we currently do not support VPC with Kubeflow cluster. Welcome contribution if you find out the solution.

@schrodervictor
Copy link

It seems this is still open. After successfully creating the management cluster, I wanted to deploy the managed kubeflow cluster in a VPC (network) and subnet (subnetwork) that I had previously created in my project. It didn't work.

It turns out that networkRef.name and subnetworkRef.name only work for resources managed by the Config Connector (the name they expect are the respective proxy k8s resources). To reference a network and subnetwork not managed by Config Connector, we need to use networkRef.external and subnetworkRef.external, pointing to the fully qualified URL that identifies the target resources. Example:

# kubeflow/common/cluster/upstream/cluster.yaml
---
apiVersion: container.cnrm.cloud.google.com/v1beta1
kind: ContainerCluster
metadata:
  ...
spec:
  ...
  networkRef:
    external: https://www.googleapis.com/compute/v1/projects/${PROJECT_ID}/global/networks/${NETWORK_NAME}
  subnetworkRef:
    external: https://www.googleapis.com/compute/v1/projects/${PROJECT_ID}/regions/${REGION_NAME}/subnetworks/${SUBNETWORK_NAME}

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants