Without merging these two organizations, it's really hard to see how this ends well. Cross-org collaboration within a conglomerate is real "knives out" territory, no one wants to cede power and they all want the integration to be "equal" in some sense. Terraform and Ansible's billing models are wildly different as well, not sure how you marry them together (are CloudPaks still a thing?). At least Ansible Tower is somewhat useful compared to TFE as a workflow tool, maybe that will be the integration point.
olgeni · 16h ago
Will Terraform become dog-slow too? :|
leoh · 15h ago
Probably not; on the contrary, a successful integration could make Ansible quite fast.
No, Ansible or Terraform by themselves are terrible. Together they will collapse into a singularity of awful which will consume the world into a goo of self-replicating YAML.
Pet_Ant · 14h ago
Just started learning Terraform. What would you recommend other than doing it byhand or provider dependent tooling?
mannyv · 12h ago
Terraform is hard because it can be a different way of thinking about deployments and your infrastructure. And since it's declarative you can't do conditionals in normal ways.
What I used to do is create things in my target environment then try to recreate them in terraform (without the import, because import didn't really work so well). Then do an apply/plan and see what changes were listed.
What I would do today is ask ChatGPT for terraform for specific things, then see what it outputs, compare it with the registry, then do the apply.
Terraform is half the battle, the other half is figuring out how the specific provider represents their stuff...which is why ChatGPT is helpful.
Oh, and also look at your provider's examples. You presumably know how things work in your provider, so looking at their terraform registry will help you figure out how they model stuff in terraform so you can model your stuff in terraform.
dp-hackernews · 12h ago
Or use Pulumi instead and get the best if both worlds.
ahmeni · 9h ago
The only thing worse than Terraform is not using Terraform. It's a pain but always worth using.
What I used to do is create things in my target environment then try to recreate them in terraform (without the import, because import didn't really work so well). Then do an apply/plan and see what changes were listed.
What I would do today is ask ChatGPT for terraform for specific things, then see what it outputs, compare it with the registry, then do the apply.
Terraform is half the battle, the other half is figuring out how the specific provider represents their stuff...which is why ChatGPT is helpful.
Oh, and also look at your provider's examples. You presumably know how things work in your provider, so looking at their terraform registry will help you figure out how they model stuff in terraform so you can model your stuff in terraform.