Dustin C. Hatch
ab458df415
By default, Kubernetes waits for each pod in a StatefulSet to become "ready" before starting the next one. If there is a problem starting that pod, e.g. data corruption, then the others will never start. This sort of defeats the purpose of having multiple replicas. Fortunately, we can configure the pod management policy to start all the pods at once, regardless of the status of any individual pod. This way, if there is a problem with the first pod, the others will still come up and serve whatever data they have.
Dustin's Kubernetes Cluster
This repository contains resources for deploying and managing my on-premises Kubernetes cluster
Cluster Setup
The cluster primarily consists of libvirt/QEMU+KVM virtual machines. The Control Plane nodes are VMs, as are the x86_64 worker nodes. Eventually, I would like to add Raspberry Pi or Pine64 machines as aarch64 nodes.
All machines run Fedora, using only Fedora builds of the Kubernetes components
(kubeadm, kubectl, and kubeadm).
See Cluster Setup for details.
Jenkins Agents
One of the main use cases for the Kubernetes cluster is to provide dynamic agents for Jenkins. Using the Kubernetes Plugin, Jenkins will automatically launch worker nodes as Kubernetes pods.
See Jenkins Kubernetes Integration for details.
Persistent Storage
Persistent storage for pods is provided by Longhorn. Longhorn runs within the cluster and provisions storage on worker nodes to make available to pods over iSCSI.
See Persistent Storage Using Longorn for details.