JOY WANGUI
Kui · writing
// articles

Notes from production.

Write-ups of real infrastructure work: incidents, release engineering, monitoring and the decisions behind them. Company and client details are left out.

27 Oct 2026Triaging an Apache 503 from the edge inward, and the monitoring that would have caught itA staging site sat on a 503 for about 39 hours because nothing was watching it. I rebuilt the failure in a Docker lab, added layered Prometheus probes, and got time-to-detect down to about a minute.