Forget Deleted Resources
Infrastream keeps a record of each resource it manages. If someone destroys a resource outside the platform, the record stays. Each later run tries to reconcile a resource that no longer exists. This guide shows how to remove those records. It also shows the one case where removal is a serious mistake.
What you see
The resource shows as stalled, not failed. The error names a provider, not the resource:
postgres ClientManager not found in context —
resource not registered through a postgres provider
The stall does not clear on its own. A new run does not help.
The platform needs the provider to delete a resource. For example, it needs a Postgres connection to drop a Postgres role. If the database is gone, so is the connection.
The platform releases a record on its own only with proof that the resource is gone. One proof is a report that the parent project no longer exists. The absence of a provider is not proof. It also fits a live resource that a configuration change orphaned. You give the platform the evidence that it lacks.
Check before you confirm
A project that lost billing is not a deleted project. The project and its resources still exist. They keep billing. Infrastream cannot read them. If you forget those records, live infrastructure becomes invisible to the only system that tracks it.
| Item | Resource deleted | Billing unlinked |
|---|---|---|
| Does it still exist? | No | Yes |
| Error | ClientManager not found in context | PermissionDenied … check billing account |
| Shown as | Stalled | Failed |
| Offers "Forget this entry" | Yes | No |
| What to do | Forget the record | Restore billing, or remove the resource from your manifests |
If you are not sure, ask the project:
gcloud projects describe <project-id>
gcloud beta billing projects describe <project-id> # look for billingEnabled
A project that answers is not gone. Do not forget its resources.
Steps
- Open the Infrastructure tab and select the stalled run. It must be the latest apply run. On an older run, the option is off.
- Select the stalled resource and choose Forget this state entry.
- Read the resource path and confirm that the resource is gone.
- Select I have checked, and this resource no longer exists.
- Select Forget this entry.
Only an administrator of the environment that owns the resource can do this. A production resource needs a production administrator.
Verify
The stall disappears from the run, and the stalled count updates. The next run does not try the resource again.
The platform keeps a permanent entry in the run history that shows what was forgotten and by whom. A forgotten record does not delete or change anything in your cloud account.
Expect two passes. When you forget a database, its roles and grants do not disappear with it. The next run reaches them directly and offers them one by one. The confirmation message tells you how many dependents to expect.