Skip to main content

Administrator guide

The administrator console is where organizations, groups, and users are managed; credits allocated and requests decided; sessions monitored; and — for the platform administrator — the GPU catalog, policy, clusters, nodes, storage, images and system settings configured.

Start with Roles and scope: what each administrative role sees is the key to every page that follows.

Tenant administration​

The day-to-day of an organization or group administrator.

  • Dashboard — what is running in your scope.
  • Organizations — the top of the tree, its administrators and budget.
  • Groups — teams, members, group administrators.
  • Users — the account list and what membership decides.
  • Credits — the hierarchy and what is billed.
  • Session monitoring — the live view, and what ends a session by itself.
    • Intervening — force-terminate, bulk clean-up, the queue.
  • Audit log — filters, export, and the investigations that work.

Platform administration​

Super administrator only.

  • Resources and policy — catalog versus quota.
  • Clusters — registering, session domains, deregistering.
    • Nodes — liveness, cordon, drain, delete.
    • Node pools — dedicated hardware and the spill question.
  • GPU devices — aliases, faulty cards, allocation versus utilization.
  • Storage — pools, every volume, running out of space.
  • Images — importing, building, keeping the catalog small.
  • Metrics — DCGM and host metrics, and what they tell you.
  • System settings — branding, sign-up policy, GPU placement.

Deploying and running the platform itself — installation, storage, monitoring, upgrades, backups — is covered under Operations, starting with Getting started.