Hello, I am building an HTTP API using the Automation API for managing dynamic Pulumi stacks based on HTTP request parameters. The API provides actions like

create

,

update

, and

delete

for managing the stacks. My question is, what is the correct workflow for updating all stacks when I need to reconcile a new version of the Pulumi code for existing stacks? Do I need to get all stack identifiers and run an update on each one individually?

Happy to help out with the automation API route, but have you considered using deployments? Deployments makes this a lot easier because it allows you to monitor long running async updates (creating a EKS or GKE cluster for instance might take 5 mins which is too long for sync HTTP request)

Thanks. We haven’t tried

Pulumi Deployments

we will look to it. Regarding long running async updates we are planning to use workflow engine to manage these “async” HTTP requests. In our current case only single workflow step is in pulumi, other steps (which not based on pulumi) still need workflow engine.

We don't have it scheduled, but we do want to add webhook support to deployments (web hooks in response to events like deploy complete, fail, etc). That might significantly reduce the amount of code you need to write for your workflow engine. If you're seriously considering deployments as a part of the solution, we could look at trying to get the webhooks work scheduled soon.