A Self-Hosting Checklist Before Pointing a Domain at Anything

Every few months I stand up a new service, and every few months I rediscover the same truth: the deploy is the easy 10%. This is the preflight checklist I now run before any domain points at a new box.

The basics

1. DNS with a low TTL first. Set the record at TTL 300 so mistakes propagate out quickly; raise it after launch week.
2. TLS from day one. Automated issuance and renewal, with a calendar reminder as the backstop. A cert that silently expires is how outages start.
3. HTTP → HTTPS redirect, HSTS once you are sure.

The parts everyone skips

• Backups that restore. An untested backup is a rumor. Schedule a quarterly restore drill — to a clean machine, not your laptop.
• Logs that survive reboots. Ship them off-box. The log you need is on the machine that just died.
• A status page you control, separate from the thing that can fall over.
• Monitoring with an owner. Alerts nobody receives are decoration.

The rollback story

Before going live, answer in one sentence: if this deploy is bad, how do I get back to the last good state, and how long does it take? If the answer is "redeploy and hope," you do not have a rollback story — you have a wish.

None of this is glamorous. That is the point: glamour belongs in the product, not in the plumbing.