System console
Where global administrators manage resources that span tenants
Required permission: Global admin
The system console is where you manage things that belong to no tenant: the tenant list, all users, shared images, system diagnostics, the build cache, the weights mirror, the monitoring stack (Prometheus and Alertmanager) and the Kubernetes cluster. These screens give the same answer whichever tenant you are looking at, so the top bar shows the System (시스템) badge instead of the tenant selector. Only global administrators can enter.
System home
This is where you land when you press System console (시스템 콘솔) in the avatar menu. The same list as the left menu is laid out as cards, and some cards carry a status badge.
| Card | Badge |
|---|---|
| System diagnostics | Overall diagnostics verdict — Diagnostics OK (진단 정상) / Diagnostics need attention (진단 주의 필요) / Diagnostics down (진단 장애) |
| GPU slots | "GPU pods waiting N (GPU 대기 파드 N)" if pods are waiting for a slot |
| Tenants | "Deletions in progress N (삭제 진행 N)" if a deletion is in progress, "Deletions partially failed N (삭제 일부 실패 N)" if a deletion partially failed |
If a badge cannot be read, only the badge is left out; the link still works.
Menu
| Menu | What it does | Page |
|---|---|---|
| Tenants | Create tenants, toggle features, deactivate, delete | Tenants |
| Users | All users and their role in each tenant; activate, deactivate or delete accounts | Users |
| Global audit log | Records of administrative actions across all tenants and the system | Global audit log |
| Shared images | Runtimes, stagers and builders baked once and handed out to several tenants | Shared images |
| System diagnostics | Connection state of dependent components and configuration verdicts | System diagnostics |
| Build cache | Package proxy state and cleanup of the build layer cache | Build cache |
| Weights mirror | Keeps one copy of training starting weights on the internal network | Weights mirror |
| Prometheus · Alertmanager | Raw screens of the monitoring stack | Monitoring stack |
| Cluster · Nodes · Namespaces · Workloads · Storage | Kubernetes control panel | Cluster |
| GPU slots | GPU occupancy and idle slots, and wiring as shared pool / dedicated / unwired | GPU slots |