Series on Tom Meurs

Linux desktop

Thu, 11 Jun 2026 20:28:00 +0100

Eight posts, in the order that keeps you from painting yourself into a corner. It opens at the install itself, one distribution and a USB stick, then walks through the first week of doing real work on the result, then spends a post on the things that go wrong and what to do when nothing fixes them, then asks why you would switch at all, and ends with three posts on the parts you keep when you change machines, the defaults, the multiplexer and the dotfiles. Arch against NixOS comes last on purpose, because it is not a first decision.

Homelab from scratch

Mon, 18 May 2026 19:30:00 +0100

Six posts that go from an empty shelf to a cluster with a backup you have actually restored from. The order matters: the hardware post comes before the network, the network before the hypervisor, and the backup post is last because it is the one everybody skips.

Start here if self-hosting appeals to you and you own no server yet. It assumes a terminal and nothing else.

Observability stack

Sat, 02 May 2026 20:15:00 +0100

Six posts in the order you build the stack. The three signals first, each with the storage question that comes with it, then what you do with them, and finally remote write, which is where a second cluster stops being a copy of the first.

Read this if you run Prometheus and are not sure what the next component is supposed to be.

Kubernetes in production

Tue, 28 Apr 2026 19:45:00 +0100

Nine posts, and the longest path on the site. It starts by separating simulation, emulation, virtualisation and containers, because most Kubernetes confusion is really confusion about one of those. Then the question of whether you need a cluster at all, then sizing, then the spine: etcd, graceful degradation, what unbreakable actually costs, chaos engineering, and finally a cluster that runs offline at the edge.

Start here if you have heard of Kubernetes and nothing else. By the end you will know which parts fall over first and what that looks like at 2am.

Key management

Sat, 18 Apr 2026 10:47:00 +0200

Four posts about keys you hold yourself. It starts from nothing, so it would qualify as an entry path, and it is not on the front door because the reader who needs it usually arrives by searching for gpg.

Read this if your secrets live in a password manager somebody else operates, and that has started to bother you.

Immutable infrastructure

Mon, 23 Mar 2026 19:28:00 +0100

Three posts about servers that cannot drift, because there is nothing on them to drift. Talos first as the concrete case, then the comparison with NixOS for the same job, then the part nobody expects: a machine you can rebuild from a file is also a machine you can prove something about.

Read this if you patch nodes by hand and have stopped trusting that two of them are the same.

ArgoCD in production

Fri, 20 Jun 2025 10:00:00 +0200

Six posts. ArgoCD against Flux sits at step two rather than step one deliberately: the comparison is the first thing everybody searches for and the wrong thing to read first, because you cannot judge it until you have run one of them.

Read this before you put GitOps in front of a cluster that matters. The last post is the one that pays for the other five.