Admin console
Where a tenant ADMIN manages their tenant's members, tokens, audit log, training operations and image variants
Required permission: SettingsADMIN
The admin console manages the one tenant currently selected in the tenant selector. You need the tenant's ADMIN role (settings permission) to enter; global administrators can enter for every tenant. Enter through Admin console (관리자 콘솔) in the avatar menu or at the bottom of the left menu.
Admin home
My permissions (내 권한) at the very top shows the tenant code and your role. Below it, the Readiness of this tenant (이 테넌트의 준비 상태) card checks the three things you need before submitting the first training run. Each line links to the screen where you can fix it.
| Line | Ready when | If not |
|---|---|---|
| Published runtime (게시된 런타임) | At least one training runtime variant that has finished publishing or points to an external image | Manage runtime variants (런타임 변형 관리) → Image variants |
| GPU wiring (GPU 배선) | A shared pool or dedicated slots are wired | If unwired, training and serving run on CPU only. Ask a global administrator for GPU wiring |
| Member count (멤버 수) | There are members | Manage members (멤버 관리) → Members |
The cards below are the same list as the left menu, filtered by your role and the tenant's feature settings (for example, a tenant with the training feature turned off has no training runner or dataset cache).
Menu
| Menu | What it does | Page |
|---|---|---|
| Members | Register accounts, grant roles, permission matrix | Members |
| MLflow tokens · Container tokens | Issue and revoke MLflow tracking tokens and docker login tokens | Tokens |
| Audit log | This tenant's records of administrative actions and CSV export | Audit log |
| Training runner · Dataset cache | Training Job state and force kill, staging cache cleanup | Training operations |
| Training runtime · Dataset stager · Serving builder | Add, publish and set default image variants | Image variants |