This message was deleted.
# atlantis-community
s
This message was deleted.
c
That can happen if you change multiple roots, so I assume that can also happen with multiple namespaces in the same root as well.
a
Am I correct in thinking the only way to resolve that would be new commits to the same branch to get the apply to be successful? As I am looking into failure cases where our main branch does not reflect the true state of the applications
b
This can happen, but then I would say just fix whatever didn't succeed properly. A best practice is to only target multiple workspaces as long as they provision the same infrastructure across multiple environments for example.
I would try to avoid using different roots and different workspaces from within a single repo in a single PR.
a
I see so, the emphasis here is on damage control so it would be the same infrastructure so it’s not impacting multiple none related workspaces and create a single PR per workspace
So for reference to the docs this isn’t best practice but it’s supported? here
Copy code
version: 3
projects:
- dir: project1
  workspace: staging
- dir: project1
  workspace: production
Just trying to figure out the best way to keep PR’s mergable. I might have re organise some of the projects
b
the docs do refer the best practice
It's preferred to use a single root, and multiple workspaces
It's not preferred to use multiple roots, and multiple workspaces
as then 2 (or more) complete different applies are happening in a single PR
which can cause quite some annoying situations if 1 of them breaks
Copy code
version: 3
projects:
- dir: project1
  workspace: staging
- dir: project1
  workspace: production
Copy code
version: 3
projects:
- dir: foo1
  workspace: foo1
- dir: foo2
  workspace: foo2
a
Thats great thanks!
b
🙂!
r
@Bruno Schaatsbergen any chance this is possible without having an
atlantis.yaml
in each repo? relating to this question https://atlantis-community.slack.com/archives/C5MGGAV0C/p1695216034862679