So I did a tutorial based on tensorflow-servings and Kubernetes. All steps are working fine except the docker image pushing to the cluster.
this is the tutorial that i have tried. https://www.tensorflow.org/tfx/serving/serving_kubernetes
And when I'm trying to push the docker image it gives an error like this,
I have tried to create the cluster with scopes also. But the result is same as above.
The command I use to create a cluster with scopes:
gcloud container clusters create resnet-serving-cluster --num-nodes 5 --scopes=storage-rw
So what is the wrong with this? Have I done something wrong???
There may be a variety of reasons.
1) I'd recommend to start with check if full API access has been granted.
2) Update gcloud components gcloud components update
3) Use gsutil to make sure you have permission to write to the bucket:
$ gsutil acl get gs://<my-bucket>
You are trying to push your image into your private registry on gcloud. Please verify if you can access your private registry:
gcloud container images list-tags gcr.io/"your-project"/"image"
all information about gcloud private registry you can find here:
Additional helpful information you can find here
Please notice that:
By default, project Owners and Editors have push and pull permissions for that project's Container Registry bucket.
Project Viewers have pull permission only.
Ok found the answer. My project ID and registry name are not equal. I re-tag the docker image with new registry name providing my project id and push it. It works.