Lab 05: Kubernetes with BCM and running a workload
180 minutes suggested, up to 360 in one session. Price: 5 credits per minute. Includes Check my work.
Hosted, this lab covers Parts B, C. The other Parts need your own machine (Home Lab Kit) and are shown collapsed in the lab.
The GPU tier is not available on this site yet.
What you will be able to do
- Deploy Kubernetes on a BCM cluster with
cm-kubernetes-setup, and find what it created in cmsh. - Give a BCM user rights in one namespace with the two-step
--add-user/--modify-usersyntax, and test the limits of those rights. - Install K3s and the NVIDIA GPU Operator on a node whose driver and toolkit are already installed, and prove that the node advertises
nvidia.com/gpu. - Run a GPU Job, a GPU Deployment and a Service, with resource requests and limits, and prove each one from cluster evidence.
- Steer pods with labels, node selectors, taints and tolerations, and guard a namespace with RBAC and a ResourceQuota.
- Diagnose the six faults that most often stop a Challenge 2 workload: a cordon, a taint, a GPU over-request, a quota, a missing pull secret and a broken Service.