Currently the EC2 instance only puts objects to the S3 bucket at the end of its lifecycle to upload results.
If the EC2 instance does not have adequate permissions to upload to the bucket, this means the entire experiment was run but results could not be uploaded.
I think it would be better to "touch" a file on start-up and identify any possible permissions errors before installing the AutoML framework and running the experiment.
Currently the EC2 instance only puts objects to the S3 bucket at the end of its lifecycle to upload results. If the EC2 instance does not have adequate permissions to upload to the bucket, this means the entire experiment was run but results could not be uploaded.
I think it would be better to "
touch" a file on start-up and identify any possible permissions errors before installing the AutoML framework and running the experiment.