-
Notifications
You must be signed in to change notification settings - Fork 10
Description
What happened?
Pulumi 3.61.0 (ish) introduced org/project/stack
type naming conventions for stacks in self-managed backends.
This is causing the EKS installer to fail since, say, stack 02-cluster-services
is looking for a stack named, say, 01-cluster-configuration
but the system is returning organization/01-cluster-configuration/01-cluster-configuration
and so the installer failes.
The other installers probably have the same problem - haven't tested yet to confirm.
WORKAROUND
Set environment variable:
PULUMI_SELF_MANAGED_STATE_LEGACY_LAYOUT=1
Specifically when deploying a new self-hosted installation do the following to avoid the issue:
- CREATE a new self-managed backend for the self-hosted installation.
- BEFORE running any of the installers or
pulumi
commands,export PULUMI_SELF_MANAGED_STATE_LEGACY_LAYOUT=1
NOTE: If you ran into this problem, the backend is tainted and you'll need to clean up the resources, and then start afresh with a new backend.
Expected Behavior
Things need to work with the new naming convention
Steps to reproduce
Try launching an eks cluster using a new s3 bucket backend.
(NOTE: pre-existing self-managed backends will work since the legacy naming convention will continue to be used.)
Output of pulumi about
stuff
Additional context
No response
Contributing
Vote on this issue by adding a 👍 reaction.
To contribute a fix for this issue, leave a comment (and link to your pull request, if you've opened one already).