Operate

Place compute, work the queue, and debug Kubernetes.

The process

  1. 1
    Check compute
    Review compute nodes and their kube status.
  2. 2
    Manage capacity
    Reserve or release compute spaces and placements.
  3. 3
    Work the queue
    Inspect and cancel queued compute items.
  4. 4
    Scale GPU VMs
    Spawn or despawn the Azure GPU node pool to match load.
  5. 5
    Debug k8s
    Pods, jobs, PVCs/PVs — logs, describe, kill.

Tools & resources

Compute Nodes
Compute Spaces
Compute Queue
GPU VMs
Kubernetes
process-first view · pages shared with Legacy