Delete a Tenant
StableHow tenant deletion works and why it should be treated as an asynchronous destructive operation.
Updated: March 5, 2026
Deleting a tenant is a destructive operation: together with it, a set of databases and related durable state may also be deleted.
Basic rule
If the tenant still contains databases, use the cascade scenario:
DELETE /v2/tenants/:tenant?cascade=true
In the UI, a more cautious path is usually used: delete the databases first, then delete the tenant itself.
What happens after the request
Deletion does not always finish instantly. Usually, the platform:
- moves the resource into an intermediate state;
- stops active compute if it is still running;
- cleans up storage entities;
- removes the tenant record from the catalog.
That is why the tenant may still remain visible in the system for some time after the request.
When deletion will fail
Typical reasons:
- the token does not have
can_delete; - the resource is already being deleted;
- the tenant still contains databases and cascade mode was not specified.
Practical advice
Before deleting a tenant, make sure that:
- the required data is no longer used by applications;
- dependencies have been checked;
- the team understands the consequences.
This is especially important for production resources: deleting a tenant is easier than later restoring the usage logic around it.
