This is a demo of how to define step functions using Serverless Framework.
We have a third party service that is unavailable from time to time then we need a way that will receive those files and upload them as soon as possible.
The solution is a REST service that schedules the uploading of a certain file in a S3 bucket to the third-party service as soon as the service gets available.
We have a REST service that will instantiate a Step Function workflow responsible for sending the file. If for some reason it's not possible to send a specific file for more than 3 days, we mark the workflow as failed and send the file upload requeset to a DeadLetterQueue in SQS, so we can analyse later the reason and eventually retry manually those cases.
For details about the workflow, see the spec at serverless.yml.
-
Configure your serverless with AWS credentials
-
Open serverless.yml and change the S3 bucket name to something different from 'upload-flaviostutz' because this must be unique among all AWS accounts in the same region
-
Deploy the stack
sls deploy --verbose-
Open AWS console, find S3 bucket 'upload-[name you used]-dev' and upload any file to folder 'tmp'
-
From the output of the deploy run, get the URL from the endpoint that ends with '/schedule-upload'
-
Schedule the asynchronous processing of this file by calling the endpoint
curl -X POST [URL from deploy ouput]/schedule-upload?file=[name of the file you uploaded to tmp]- Open StepFunction on AWS Console and check the executions of the steps related to the file