https://pulumi.com logo
Join Slack
Powered by
I’m working with Pulumi and the S3 backend. I have...
# general
a
I’m working with Pulumi and the S3 backend. I have two Pulumi projects - 
infrastructure
and 
application
. The 
infrastructure
project has a single stack: 
internal
. The 
application
project has a single stack: 
dev-aws
. When I do 
pulumi stack ls
in either project, I see both stacks:
Copy code
$ pulumi stack ls
NAME       LAST UPDATE  RESOURCE COUNT
dev-aws    n/a          n/a
internal*  n/a          n/a
That is not what I expect. I can also select (e.g.) the 
dev-aws
stack when I’m in the 
infrastructure
project. This is clearly a mistake, but it will create a new 
dev-aws
stack within the 
infrastructure
project when I run 
pulumi up
from the 
infrastructure
project with the 
dev-aws
stack selected. Why is it doing that? Is there some other recommended way to manage multiple projects with the S3 backend?
I’m going to try specifying the backend in the Pulumi.yaml file for each project and see if that works the way I’d expect.
l
You have two choices: you can specify different S3 backends via the 
backend: url:
property (they can be different keys in the same bucket), or discriminiate stacks with projects names.
If you use different backends, you cannot access outputs of either stack from the other.
If you use the same backend then you have one bag of stacks. So you need your stacks to describe which project they belong to, in order to avoid confusion.
Only the Pulumi backend has built-in support for "this stack belongs to that project".
Also, it's the only one that supports organizations.
b
what tenwit said ^^
the general pattern is that people will prefix their stacks with a project name, like 
infrastructure.dev
a
Ok, thanks for the explanation. I definitely want to access the infrastructure outputs in the application stacks, so I guess I’ll need to use the naming conventions suggestion.
p
33 Views
Open in Slack
PreviousNext
Powered by