My homelab is about to run the production database for a real website. What would you fix before Friday?
retained text
I’m launching a free keyword research tool on Friday, and the main database lives on an Ubuntu box in my room. The setup: \- Ubuntu 24.04 with Postgres 16 installed natively (no Docker), data on an internal HDD \- Nightly backups to two USB drives, 14 kept on each \- A Cloudflare Tunnel locked down with Access service tokens. No ports are open on my router. \- The public site runs on Cloudflare and reads a synced copy of the data. If my internet or the box goes down, the site stays up and new data waits in a queue until it’s back. It’s already had a rough week. My internet dropped twice in 12 hours, and the second time it killed an overnight import job halfway through. That job retries now. Then I moved Postgres to the HDD, and my install script tried to query the database before Postgres had finished starting up, so I added a wait loop. A sync job also ran into its 30-minute systemd timeout while pushing 120k changed rows. I have no UPS, and my monitoring is me checking on it. If this were your box, what would you change before real users show up?
document fetched 2026-11-18 00:40 · canonical https://reddit.com/r/homelab/comments/1xyponj/my_homelab_is_about_to_run_the_production