Skip to content

open source

What we maintain.

We build on public foundations, and part of those foundations we maintain ourselves. All Apache 2.0, all readable before you even talk to us: it is the most honest way we know to let you judge how we work.

the operator

understudy

Best-effort high availability for single-replica workloads: it keeps them alive through drains, upgrades, consolidations and spot interruptions, without paying for a second replica around the clock.

In a real spot interruption, measured on EKS, it lost 1 request out of 764: less than an ordinary rolling update.

Go · Apache 2.0 · CRD and admission webhook · the repository

the modules

Terraform for Scaleway

Five modules for the Kubernetes platform on Scaleway — network, gateway, cluster, node pool, API server ACL — published on the Terraform Registry and installable in one line.

  • vpc

    Creates a Scaleway VPC and its private networks.

    registrysource

  • public-gateway

    Creates a Public Gateway, its flexible IP, and its attachments to private networks.

    registrysource

  • kapsule-cluster

    Creates a Scaleway Kubernetes Kapsule cluster.

    registrysource

  • kapsule-pool

    Creates a node pool for a Kapsule cluster.

    registrysource

  • kapsule-acl

    Restricts which addresses may reach the cluster API server.

    registrysource

We wrote them because we needed them: on the Kubernetes side we could not find maintained modules we would bet a production cluster on. The documentation covers Scaleway API quirks, the ones you would otherwise discover at apply time.

Why a European cloud

The point is not the flag. It is that the method does not change: landing zone, GitOps and FinOps work the same way on AWS, GCP, Azure and Scaleway — the Terraform provider changes, not how we work. If keeping data in Europe is a requirement of yours, it is an option we genuinely cover, with the code in the open. If it is not, what remains is the proof that we can carry the same setup elsewhere.

How they are built

  • CI on every pull request: formatting, validation, lint, examples that initialize
  • documentation generated from the code, with the build failing when it drifts
  • semver and changelogs: no silent breaking changes
  • complete examples that run, not fragments to paste

// If that reads as fussiness, it is the same standard we hold your Terraform to.

A technical call, not a demo.

Half an hour, and the person on the other end is the one who writes the Terraform. Bring your cloud bill or your most annoying problem; if we can’t help, we’ll tell you right away.

no credit card · no sales reps · no pushy follow-ups