Skip to content

lt-infra

needs verification

Purpose

lt-infra defines LT's Hetzner infrastructure and deployment configuration. Terraform owns cloud resources; Ansible owns provisioning and deployments.

What it does

It provides staging and production application and data hosts, a shared operations host, private networking, storage and the deployment runner. The technical team maintains the deployment and recovery procedures.

Production backups run nightly from the operations host to a separate Hetzner project. They capture database dumps and media recovery points. Redis can be recreated and Meilisearch rebuilt from content.

For maintainers: technology

Terraform, Ansible, Docker and Caddy on Hetzner Cloud. GitHub Container Registry (GHCR) holds application images. 1Password supplies secrets at runtime.

How it connects

The technical team first delivers changes to staging so they can be checked. Reviewed changes then move to production. The infrastructure provides the servers, storage and connections that make these applications available.

Caddy exposes the app services. PostgreSQL, Redis and Meilisearch live on the environment's data host. Hetzner Object Storage holds raw uploads and operational data; Cloudflare R2 serves finished media. The public website runs separately on Cloudflare Pages.

Who can do what

These are product personas, not infrastructure permissions. An LT Administrator does not automatically receive deployment access. “Partial” means separately authorized engineering operators may administer infrastructure.

ActionAnonymousResearcherCoordinatorAdministratorEditor
Use the hosted public servicesYesYesYesYesYes
Provision hosts or deploy imagesNoNoNoPartialNo
Access operational secretsNoNoNoPartialNo
Run backup and restore operationsNoNoNoPartialNo

The access-owner table is missing. Ask the team lead for the appropriate operational access. Only authorized technical staff make infrastructure changes.

Live production names are established by DNS configuration and the cutover report. See the full environment map.

Where to read more

The environment map and getting access. Infrastructure is managed by the technical team; you do not need to provision or deploy anything to use the applications.

Known gaps

Older production documentation still describes pre-cutover DNS and temporary core.prod.lt.org and cms.prod.lt.org aliases. The cutover report leaves alias cleanup pending. The brief records missing monitoring beyond health checks and backup heartbeats, and no access-owner table. Follow open questions rather than treating old aliases as recommended environments.

For maintainers: source references

These references require repository access. Everything needed to understand this page is explained above.

Latest Thinking · Welink internal guide