Documenting deploy for ML-Ops, including issues/debugging log

[aaronsteers]

Log from walkthrough session on a new deployment with Slalom NY Market:

1. Started by cloning the repo locally.
2. Navigated to the repo subfolder: samples/ml-ops-on-aws
3. Because the account requires MFA and is integrated with Azure Active Directory, we had to use the following tool to create temporary AWS credentials: https://github.com/sportradar/aws-azure-login
4. We opened the file 01-environment.tf and modified the path to the AWS credentials file. Instead of the default, we used: aws_credentials_file = expandpath("~/.aws/credentials")
5. The terraform init command failed with an error message about inability to validate checksums. After reinstalling openssl, the issue was resolved and terraform init was successful.
6. Deployment using terraform apply was then partially successful. We received an error that the maximum number of VPCs had already been reached.
7. To resolve the error with AWS not allowing additional VPCs, we ran terraform destroy, then changed the region specified in infra-config.yml and then re-ran terraform apply.
8. This time, in a region that had not already reached its VPC limit, the terraform deploy was successful - with the one exception being the docker push to ECR.

Next steps:

1. To deploy to ECR, we could either (a) setup the ECR Credential Helper or (b) simply manually login to ECR and manually docker push _x_, where _x_ is the image referenced in the failed docker push command.