[terraform] Hetzner / Linode / Vultr / GCP / Azure sibling modules

[EmperorMew]

We ship Terraform modules for DigitalOcean and AWS Lightsail today (pay-hydra/terraform/{digitalocean,aws-lightsail}/). Port the same pattern to:

• Hetzner Cloud (~$5/mo CX22)
• Linode (~$5/mo nanode)
• Vultr (~$5/mo HF)
• GCP (e2-micro)
• Azure (Standard_B1s)

Each is ~60 lines — copy digitalocean/main.tf, swap provider block + resource, keep the shared cloud-init.yaml.

Blocker first

The current cloud-init installs Node via NodeSource's Debian apt repo. Hetzner / Linode / Vultr often default to Debian-family so those work straight away. GCP and Azure may need a RHEL-family fork of the install step. Audit note: the "cross-distro" claim in cloud-init.yaml line 2 is not actually true today — only fork first if you're sure the base image is Debian-family.

From Terraform audit (2026-04-20), item 3.11. Take any one — file a separate PR per cloud.

[EmperorMew]

Reopening — only the Hetzner module shipped in commit 5a1391f. Still need siblings for Linode, Vultr, GCP, and Azure. The Hetzner module is the cleanest template to fork from: https://github.com/voidly-ai/voidly-pay/tree/main/pay-hydra/terraform/hetzner

Remaining checklist (each is its own PR):

• Linode — linode_instance resource, hcloud-style pattern, ~60 lines
• Vultr — vultr_instance resource
• GCP — google_compute_instance — note: cloud-init on non-Debian needs a fork of the NodeSource install step
• Azure — azurerm_linux_virtual_machine — same note

Good first PR if you want to touch real Terraform in a sandbox.

[EmperorMew]

Update — 2026-04-23: Linode and Vultr landed in commit 972e77e. Remaining modules:

• GCP — google_compute_instance — cloud-init needs a RHEL-family fork of the NodeSource install block if using a Google-curated image, or it can stay as-is if operators pick ubuntu-2404-lts family.
• Azure — azurerm_linux_virtual_machine — same note.

Hetzner + Linode + Vultr are now in pay-hydra/terraform/ — good templates to fork from. Terraform-lint CI matrix covers all 5 existing clouds.

[snakefood3232]

I've read your request to port the Terraform module pattern to Hetzner, Linode, Vultr, GCP, and Azure, and I'd handle this by adapting the existing DigitalOcean module to each provider, swapping the provider block and resource, while ensuring cloud-init works for each platform's base image. I can have a PR up for one of these within 2 days. I've previously created Terraform modules for similar multi-cloud deployments.

[snakefood3232]

I see this issue involves porting Terraform modules to additional providers, focusing on Hetzner Cloud and ensuring cloud-init compatibility. I'd handle this by first copying the existing digitalocean/main.tf, updating the provider and resource blocks for Hetzner, and testing with the existing cloud-init setup. Given my experience with Terraform module development, particularly with provider integrations, I can ensure the configuration aligns with Debian-family distributions. Can have a PR up within 48 hours. I've built cloud deployment strategies for similar environments before, smoothing transitions across diverse providers.