-
Notifications
You must be signed in to change notification settings - Fork 1.6k
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
Kubeflow sdk - error in client.list_experiments() #6120
Comments
IIRC, Kubeflow v1.3 enables multi-user isolation by default. @Bobgy, is that right? In that case, pipelines/sdk/python/kfp/_client.py Lines 452 to 461 in 24c551d
|
@chensun I think that is correct yes. I did try passing a namespace, but still get the error. I did also try passing the namespace 'kubeflow-example-user-com' which is what is visible from the UI. Neither works. I'm not sure whether I'm passing it incorrectly?
|
Hi @TanjaDuPlessis, I wonder if you may missed any setup required for multi tenancy. /cc @Bobgy |
@TanjaDuPlessis refer to https://www.kubeflow.org/docs/components/pipelines/multi-user/#in-cluster-api-request-authentication, it's one of the current caveats of multi-user mode. We are now supporting this usage through #5138. |
@chensun this is expected behavior, because KFP API server is authorizing requests. |
hello @Bobgy , if it's ecpected does that mean we can't use it ? |
@rexad yes, and it's now documented in https://www.kubeflow.org/docs/components/pipelines/sdk/connect-api/#connect-to-kubeflow-pipelines-from-the-same-cluster |
I'm following the example in docs (https://www.kubeflow.org/docs/components/pipelines/sdk/connect-api/) but am getting an error when trying to access the experiments or runs (when trying to list and/or create them)
I've port-forwarded ml-pipeline-ui:
kubectl port-forward svc/ml-pipeline-ui 3000:80 --namespace kubeflow
Trying to list the experiments results in an error:
The error I get from client.list_experiments():
However, listing the pipelines works:
I have verified that the port-forwarding worked - I am able to access the UI at http://localhost:3000/ and can see the pipelines. I am working with Kubeflow v1.3. I have also tried adding the additional namespace argument as suggested for multi-user (https://www.kubeflow.org/docs/components/pipelines/multi-user/#when-using-the-sdk). Any ideas on where I am going wrong?
The text was updated successfully, but these errors were encountered: