Kubernetes Details¶

Under Construction

This page is still being written

Kubernetes cluster nodes provisioned by RKE2 have 3 roles: Manager, Worker, or Combo (Manager+Worker).

RKE2 Node Roles

Manager: runs kubernetes control and etcd processes in a "management plane" to coordinate, schedule, and manage workloads
Worker: runs workloads and provisions storage capacity through Longhorn
Combo: runs both manager and worker roles

Different clusters (admin, prod, staging, etc...) use a mix of different roles.

Cluster Details¶

Production¶

prod-cluster diagram kubernetes-prod-cluster diagram dark

manager nodes do not run workloads, they are dedicated manager nodes
there are three nodes to ensure replication and high availability
worker nodes run workloads and contribute storage to the persistent disk provisioner, Longhorn.

Staging¶

staging-cluster diagram staging-cluster diagram dark

all nodes run combo (manager+worker) roles

Dev_cp and Dev_vsm¶

all nodes run combo (manager+worker) roles

Rancher Admin¶

Rancher is a UI interface for Kubernetes that runs as a workload in a dedicated admin cluster. Rancher helps us visually understand what is happening in our Kubernetes clusters.

prod-admin* nodes run the Rancher application in a highly available configuration.
all nodes run combo (manager+worker) roles

Common Cluster Components¶

Rancher "Projects"¶

Projects are Rancher constructs that help organize namespaces. See Rancher Docs - Project Administration.

`nginx` Ingress¶

Rancher's default Ingress is powered by nginx. See Rancher Docs - Adding Ingresses for more info.

Longhorn¶

Longhorn is deployed on all worker nodes

Longhorn is a persistent disk provisioner for Kubernetes; it allows pods to request storage that remains available independent of the pod lifecycle. Regardless of the cluster, any node with a worker role (including the combo role) has an additional block device that is dedicated for use by Longhorn.

GitLab Runner¶

GitLab Runner is an application that works with the GitLab CI/CD pipeline coordinator to run the jobs specified in a CI/CD pipeline. GitLab Runner has been deployed in the Utilities project on each of the clusters so that our pipelines can deploy to our clusters.