https://pulumi.com logo
#google-cloud
Title
# google-cloud
h

hallowed-cat-56281

10/22/2021, 6:09 AM
Hey there, 👋 I am facing some an interesting problem, we have been deploying our GCP infra through Pulumi for a year and it worked well overall, but we start noticing random failures more and more such as “no route found”. While not solving the problem, what this highlighted is that I probably should consider breaking up the monolith to reduce the feedback loop, and keep things separated and easier to navigate and deploy. We currently have one project (
acme
) and two stacks (
staging
and
prod
, on two different GCP accounts). This project is getting relatively big (250 resources), and I was considering creating multiple projects (each with two stacks) - e.g.
acme-storage
,
acme-compute
Here are my questions: 1. Is this a good approach or am I missing anything? 2. I tried creating one small project (e.g.
acme-storage
) but I struggle when trying to export the existing state of the bigger project to import parts of it into the new project and I end-up with resources that need to be “replaced” when doing a
pulumi preview
. Is it just a matter of not cleaning up the exported stack file enough/properly before import? Should I even be doing this? Seems dirty.
2 Views