Server Inventory

As cluster build-out has progressed, devices and their resource configuration have been standardised. Detail of the original development environments has since been refactored out of this documentation.

Cluster nodes

Cluster nodes are HP t630 Thin Clients, for which the specifications are available from HP online: https://support.hp.com/za-en/document/c05240287.

The devices were purchased second hand, and typically had limited RAM and disk installed (e.g. 32GB disk, and 8GB DDR4 RAM).

RAM and disk were installed, giving the following common node configuration:

HP t360 specs

  • HP t630 Thinclient
  • OS: Ubuntu Server 22.04 minimised
  • AMD Quad Core CPU @ 2.0 Ghz
  • 24Gb DDR4 RAM
  • 128 GB M.2 SATA SSDs
  • 256 GB M.2 SATA SSDs

northern zone solution

http request router

In order to support multiple k8s clusters on the northern network served by a single static IP assigned by the ISP, without any cluster depending on another, an HP t520 device is added to the solution to serve a reverse proxy (apache) function. Routing of https requests at the proxy is based on domain name. In addition, the domain name IP address management function (ddclient) and tls certificate management for the reverse proxy (certbot) are provided on this device:

  • rudolfensis: t520; apache, ddclient, certbot

Kubernetes nodes

The northern zone solution has the following node configuration:

  • naledi: t630; k8s, ceph
  • habilis: t630; k8s, ceph
  • antecessor: t630; k8s, ceph
  • neanderthal: t630; k8s, ceph
  • erectus: t630; k8s, ceph
  • floresiensis: t630; k8s, ceph
  • norham01: t630; k8s, ceph
  • norham02: t630; k8s, ceph
  • norham03: t630; k8s, ceph
  • norham04: t630; k8s, ceph

Cluster resource totals

  • Total t630 CPU: 40 cores
  • Total t630 RAM: 200 GB
  • Total t630 Raw Disk: 2.5 TB

southern zone solution

  • denisova: t630; k8s, ceph, ddclient, ularu
  • rudolfensis: t630; k8s, ceph
  • ergaster: t630; k8s, ceph