CODEWITHSUNDEEP
tensorflowimage-processinggoogle-cloud-platformgoogle-compute-enginetensorflow2.0
GOB
GOB

Reputation: 143

Google Cloud Platform failure to complete job - permissions error?

I am trying to use the GCP to train a computer vision project. I am using the AI platform.

When I submit my job, it fails with the following error:

message: ....does not have storage.objects.list access to NAME OF BUCKET REMOVED FOR SECURITY "domain": "global", "reason": "forbidden"

Any suggestions of where to start to fix this? Does it matter that my project location and bucket location are different?

Thanks!

Upvotes: 1

Views: 199

Answers (1)

Serhii
Serhii

Reputation: 4471

This problem caused by insufficient IAM permissions. More information you can find in the documentation like Cloud IAM roles for Cloud Storage section Predefined roles.

To solve this issue you should grant to your service account Storage Admin role (roles/storage.admin) or you can use other IAM roles to grant permissions storage.objects.* (in some cases permissions storage.buckets.* could be required) like Environment and Storage Object Administrator role (roles/composer.environmentAndStorageObjectAdmin) you've mentioned above.

Upvotes: 1

Related Questions