Season 7: It Worked on My Laptop · Episode 4 of 8
We Went Viral
You play: Junior AI Developer
Included with Core programme and Certification prep
The situation
A fictional celebrity, Nova Reyes, has agreed to appear on Code Nexus Live on Friday. The current plan for Code Nexus Assist, which answers creators' and viewers' questions in the chat, is one container. It is the same image you built in the last three episodes, running as a single process on a single machine. Nobody expects that to hold, and Priya would rather find out on a Tuesday than during the stream.
What you'll learn
- Kubernetes fundamentals: what a cluster is made of (control plane and nodes, API server, etcd, scheduler, controllers, kubelet), declarative resources, Deployments, ReplicaSets and Pods, reconciliation, resource requests and limits, and how the scheduler places pods.
- Deployment, StatefulSet, DaemonSet, Job and CronJob; ConfigMap and Secret with envFrom; Role, ClusterRole, RoleBinding, ServiceAccount, ResourceQuota and LimitRange; nodeSelector and affinity; containerd, CRI-O, the CRI and the OCI specifications.
Who you work with
Kabelo Mokoena
Platform and DevOps Engineer
Priya Naidoo
Cloud Architect
8-BIT
Code Nexus Internal AI Assistant
Scenes
- 1.Friday, and one container
- 2.Who does what in a cluster
- 3.Which workload kind?
- 4.Read the manifest
- 5.Scale it and watch it reconcile
- 6.Half of them are Pending
- 7.Fix it before Friday
- 8.Cluster fundamentals check
What you leave with
A cluster is a control plane (the API server, etcd, the scheduler and the controller manager) plus worker nodes running a kubelet, a container runtime and kube-proxy. You declared a Deployment and the controllers made it real and kept it real: scaling changed the desired count, and deleting a pod made the ReplicaSet create a replacement without anyone asking. Six pods wanted and three Pending was the scheduler saying no, because each pod reserved a whole node's CPU; the fix was a realistic request, not a toleration or removing the requests.