Closed montenegrodr closed 2 years ago
Issue Label Bot is not confident enough to auto-label this issue. See dashboard for more details.
Any updates? Got the same error on a similar code, after accessing the metadata-grpc-deployment through an external IP. Kubeflow was deployed using GCP's UI.
@jlewi Any update on this? This is breaking for the demo notebook.
@jlewi Same with me.
@montenegrodr did you find any workaround?
@jlewi Same with me.
@montenegrodr did you find any workaround?
no
@montenegrodr It's due to this issue.
@montenegrodr
Try using Kubeflow v1.1.0 and it should work just fine. It worked for me.
am using kubeflow v1.2.0 and still seeing the issue,,
AFAIK Kubeflow metadata is deleted from Kubeflow v1.2.0 in favor of ML metadata since the first one is abandoned and the second one is supported by KFP. Please see: https://github.com/kubeflow/metadata/pull/242#issuecomment-716343533 https://github.com/kubeflow/metadata/issues/250 https://github.com/kubeflow/metadata/issues/225 Please correct me if Kubeflow metadata should work with Kubeflow v1.2.0.
AFAIK Kubeflow metadata is deleted from Kubeflow v1.2.0 in favor of ML metadata since the first one is abandoned and the second one is supported by KFP.
That is correct @umka1332. The SDK in this repo is unmaintained, and so we decided to move to MLMD gRPC service for now.
/kind bug
What steps did you take and what happened: [A clear and concise description of what the bug is.]
Created a step in my kubeflow pipeline only for persisting metadata. But the step fails when trying to create an
Execution
with:Full stack:
Pipeline Step:
What did you expect to happen:
Create an execution and store the metadata.
Anything else you would like to add: [Miscellaneous information that will assist in solving the issue.]
Running in a GCP managed service.
Environment:
kubectl version
):/etc/os-release
):