I follow the tutorial for building kubeflow on GCP.
At the last step, I got stuck at "Check the permissions for your training component".
After setting these secretName and secretMountPath.
kustomize edit add configmap mnist-map-training --from-literal=secretName=user-gcp-sa
kustomize edit add configmap mnist-map-training --from-literal=secretMountPath=/var/secrets
and run
kustomize build . |kubectl apply -f -
I got the error:
Error: field specified in var '{GOOGLE_APPLICATION_CREDENTIALS ~G_v1_ConfigMap {data.GOOGLE_APPLICATION_CREDENTIALS}}' not found in corresponding resource error: no objects passed to apply
I cannot find my GOOGLE_APPLICATION_CREDENTIALS at /var/secrets in my local machine, but I think kubeflow will automatic create for me based on this document.
Or maybe because I use "Authenticating with username and password" for authenticating kubeflow?
Kubeflow kustomize is under dev and expect to be mature in next release v0.6.0.