Hi! When it comes to managing side-effects when using Pulumi, is there a suggested pattern? For example, we're deploying some EC2 instances to AWS via a Pulumi script, we have two sets of instances, and we need to configure the first set before making the next set. The configuration can be handled via an API, so is there an ability in Pulumi to 'pause' the deployment until we have done our configuration? Is this was the automation API was made for? I have thought about splitting my resources into separate Pulumi projects, but then I am not so sure how to pass the IDs of resources between these projects. Are Pulumi programs intended to be 'pure', i.e. only describing Pulumi resources in functional ways without any side-effects?

Hi Ben. Yes, this is the kind of thing the Automation API is good at. You can do the same kind of thing with multiple projects and stack references, though: https://www.pulumi.com/docs/intro/concepts/stack/#stackreferences

Thanks!