aws-samples / amazon-sagemaker-tensorflow-object-detection-api

Train and deploy models using TensorFlow 2 with the Object Detection API on Amazon SageMaker
MIT No Attribution
45 stars 34 forks source link

Missing image #14

Closed bonzoq closed 3 years ago

bonzoq commented 3 years ago

I get the following error:

Error for Processing job tf2-object-detection-2021-06-10-13-17-37-173: Failed. Reason: ClientError: API error (404): manifest for 757901089955.dkr.ecr.us-west-2.amazonaws.com/tfrecord-processing:20210610131649 not found: manifest unknown: Requested image not found

Does this mean that an image is no longer available?

bonzoq commented 3 years ago

Before that error I get this warning:

Waiting g denied: User: arn:aws:sts::757901089955:assumed-role/AmazonSageMaker-ExecutionRole-20210607T223583/SageMaker is not authorized to perform: ecr:InitiateLayerUpload on resource: arn:aws:ecr:us-west-2:757901089955:repository/tfrecord-processing

sofianhamiti commented 3 years ago

Can you verify the permissions of the IAM role used in your notebook instance. This seems to be: AmazonSageMaker-ExecutionRole-20210607T223583 Does it have permissions for pushing/pulling images to ECR?

bonzoq commented 3 years ago

This notebook has a role with the policy AmazonSageMakerFullAccess, which turns out to be insufficient. After adding EC2InstanceProfileForImageBuilderECRContainerBuilds policy the problem has been solved. Thank you!