cf-sewe / cplace-ops-kubernetes-on-hetzner-workshop

Kubernetes on Hetzner Workshop

Workshop: cplace on Kubernetes on Hetzner Cloud

We are conducting a workshop setting up cplace on a Kubernetes cluster running in the Hetzner Cloud.

The goals are:

1. Building up Kubernetes cluster architecture and operation know-how within the cplace Operations Team.
2. Building a PoC for running Kubernetes on the Hetzner Cloud, serving as a foundation for our road-map initiatives for 2023.
3. Running cplace on Kubernetes without the abstractions of the cplace Operator.

That allows us to understand the required Kubernetes components better and allows to build up know-how on how to support cplace in a Kubernetes environment.

Workshop Contents

Basic Kubernetes Deployment

We will deploy Kubernetes on Hetzner using the following technologies:

• Hetzner Cloud
  • Virtual Machines
  • Loadbalancer
  • Storage
• Talos OS
• Cilium (CNI)
• Piraeus/Linstor/DRBD (CSI)
• Traefik (Ingress/Gateway)

MySQL

A MySQL cluster deployed to Kubernetes will be used as the cplace RDB. We have selected the Percona Operator for MySQL based on Percona XtraDB Cluster, as it provides a great out-of-the-box experience, reducing operational efforts significantly.

Vitess?

Elasticsearch

An Elasticsearch cluster deployed to Kubernetes will be used as the cplace NoSQL database. We have selected the Elasticsearch operator to simplify the deployment and operation of the ES cluster in K8S.

cplace Deployment from Kubernetes Manifests

xxx

Kustomize

xxx

Deployment

xxx

Required Production Changes

The workshop intentionally applies some simplifications, to ensure the team can duplicate the results quickly.

• Grafana/Prometheus/Loki stack
• Kubernetes logging to Loki
• Talos kernel logging to Loki

Glossary

RDB Relational Database CNI Container Network Interface CSI Container Storage Interface CAS Container Attached Storage K8S Kubernetes (K with 8 more letters and an S) hcloud Hetzner Cloud kubectl The kubernetes CLI tool

Internal Links

Link	Description
https://traefik.training.cplace.cloud/dashboard/	Traefik reverse proxy dashboard
https://hubble.training.cplace.cloud/dashboard/	Cilium Hubble dashboard
https://dashboard.training.cplace.cloud/dashboard/	Kubernetes Dashboard

Useful External Links

• Comparing Ceph, LINSTOR, Mayastor, and Vitastor storage performance in Kubernetes
• Nginx vs Traefik: how slower one can be better?
• Why Use LocalPV with NVMe for Your Workload?