This message was deleted.

hmm this is a little tricky because of the declarative model, it would have to be done outside the parent stack with a second stack I think

sure. So i have a stack that setup up the infrastructure. The K8S cluster. Another stack sets up the ECRs and their configs. Then I want to create a "deployer" project which would dynamically set the stack via github actions. This one will deploy the projects to a stack. It would create the k8s resources (namespace, fargate profile, etc). Now, the tricky part for me is that the ELB can only have 100 rules. So if i predefined one ELB, then I'd be limited in the number of stacks. So my idea was to calculate at each deployment how many ELB instances I need and, if one more is needed, add it. Then if one can be removed, remove it before deployment.

are you using an ingress controller?

Yes, i am using aws load balancer controller

ahh I see, I was confused because ALBs have rules but ELBS do not. i think the best recommendation I have is to provision multiple ingress classes and round robin them

I meant alb, sorry :) provisioning multiple ones might not be ideal though. Not scalable either right ? We could easily over provision or under provision. I was hoping i could do some kind of for loop and provision the ingress with predictable names (eg ingress-index) and i was hoping that pulumi could figure out to either add or remove them.

How are you going to perform the application deployment? Is it going to be managed by pulumi?

@prehistoric-activity-61023 Yes, it will be managed by pulumi

I didn’t see the part when you wrote about separate stacks. I guess it’s not gonna be trivial to do that automatically inside pulumi.