Container Registry limits and behaviour
This page summarises operational limits and behavioural constraints for the container registry.
Limits
| Topic | Detail |
|---|---|
| Namespace rename | Not supported — create a new namespace and re-push images |
| Manual repository create | Not supported — repositories are created on first push |
| Regions | Each namespace is tied to one region; use separate namespaces per region |
| Duplicate namespace | One namespace name per organisation (or project) per region |
| Labels / annotations | Up to 20 each on namespace metadata |
| API base path | /v1/container-registries |
Namespace naming
Namespace names must be:
- Lowercase ASCII
- No spaces
- Maximum 255 characters
The name is set at creation and cannot be changed afterwards. You can reuse the same namespace name in different regions.
Regional behaviour
Each namespace is bound to exactly one cloud region. The registry hostname, data locality, and backend storage all follow that region assignment.
To serve images from multiple regions, create separate namespaces — one per region — and push images to each as needed.
Visibility defaults
Namespaces behave as private until namespace configuration is created. Changing visibility does not move existing images — it only changes who may pull without credentials.
Retention policy scheduling
Retention policies run automatically on a schedule (approximately hourly). You can also trigger an immediate run via the API (POST …/retention-policy/run).
Related documentation
- Concepts — Namespaces, repositories, and scoping
- Namespace configuration — Visibility and retention rules
- Repositories — Delete operations and lifecycle
- Regions — Regional deployment model