Hi, I have been asked to deploy Prefect by our Data team. Since we operate in AWS I would like to use Aurora Postgres for the database and Fargate for the containers. Is there some documentation on this deployment?

Hi @Craig we don’t yet officially support customized Server deployments outside of a single node setup but if someone from the community has some experience with a similar setup to the one you are looking into I encourage them to chime in! We will be sure to update the community as this area of Server deployment develops over time.

I set up a standard Aurora Postgres instance and pointed my locally run compose server (on an EC2) with a config.toml like:

debug = "true"
backend = "server"
[server]
  [server.ui]
    graphql_url = "<http://10.72.112.29:4200/graphql>"
  [server.database]
    host = "<http://prefect-gdesmarais.cluster-cyx6dhev9iel.us-east-1.rds.amazonaws.com|prefect-gdesmarais.cluster-cyx6dhev9iel.us-east-1.rds.amazonaws.com>"
    port = 5432
    username = "myaurorausername"
    password = "myaurorapassword"
    connection_url = "<postgresql://postgres:myaurorausernane@prefect-gdesmarais.cluster-cyx6dhev9iel.us-east-1.rds.amazonaws.com:5432/prefect_gdesmarais>"
[flows]
  checkpointing = "true"
  [flows.defaults]
    [flows.defaults.storage]
      default_class = "prefect.engine.results.PrefectResult"
[tasks]
  checkpointing = "true"

Nothing private in there... The

host

and

connection_url

are built from the writer endpoint name in the DB Identifier with the

Regional

role. Hope this helps.