Hi,
We've defined a pipeline in Kubeflow SDK and successfully deployed it to Vertex AI Pipelines where it runs through.
When deploying the same .yaml pipeline using the Kubeflow Pipelines dashboard. We get an unresolveable error: "Cannot get MLMD Objects from Metadata store."
One example online resolved the issue by rolling back to v1, but no other information on the error seems to be available.
Is it perhaps a permissions error that we are missing? We have added to the service account Vertex AI Admin, Kubernetes Admin and all other permissions required to write out to buckets.
Any feedback or help would be much appreciated.
LinkedIn
Twitter
