Built to not lose anything¶
Kacky is a record book. If it loses your times, it's broken. So a lot of the work that never shows up on screen went into making sure it doesn't. Here's what's protecting your data.
Only trustworthy times count¶
Ranks and world records are built only from trusted online records. Times that came in offline, or were imported from Kacky's older systems, are kept out of the rankings on purpose, so nothing unverifiable can ever claim a record it didn't earn. This is the single most important rule on the platform, and it has its own page.
No finish is ever lost¶
Your finish isn't a fragile message that's gone if nobody's listening. It's a durable one that waits in a queue until the backend reads it.
- If the backend is down or mid-deploy when you cross the line, your finish waits, and gets processed when the backend comes back. The system catches up on its own.
- If a finish ever does slip through, it's still recoverable. The controllers keep a log of every finish, so an admin can find it and add it back by hand.
The full mechanism is on How the game talks to the platform.
The whole database, backed up three ways, every night¶
The database holds all of Kacky's history, so it's copied every night, to three different places:
- Into the dev environment, so testing always runs against fresh, real data.
- To off-site storage on Hetzner (our sponsor), a safe copy well away from the live server.
- As a physical copy on an admin's own machine, a backup that isn't in the cloud at all.
Three copies, three places, refreshed daily. Losing Kacky's history for good would take all three failing at once. See The backend and the data.
Servers that look after themselves¶
The backend watches for each server's regular check-in. If a server goes quiet for too long, it's marked offline automatically, so the site never shows a dead server as live. And if a machine ever dies for real, it isn't a crisis: the whole fleet is described in code, so a replacement is a seven-minute rebuild, not a lost weekend.
You never think about any of this¶
And that's the point. The record of every run you've ever done is meant to outlive any single crash, deploy, or dead server. The reliability is the part you're not supposed to notice.