Hi all, a newbie question. We are trying to setup metaflow in AWS with Batch/Fargate. Below is my understanding of the overall flow of submitting a job locally

1) upload code to datastore in s3. (user's account is used to upload code to s3)
2) create job definition (user's account is used to create the batch job definition)
3) batch ecs task downloads code from datastore (using the METAFLOW_ECS_S3_ACCESS_IAM_ROLE)
4) batch job run the code in ECS(Fargate or EC2) (using the METAFLOW_ECS_FARGATE_EXECUTION_ROLE)

Does it look right to you? Thank you

Correct - except that item 4 happens before item 3. the code is downloaded inside a running batch job.

Thank you @square-wire-39606 Is there a way to skip step 1 and/or 2?

the code is only uploaded if there are any changes to it