Alert design, on-call, SLAs and SLOs, uptime maths, incidents and runbooks.
Network Monitoring, Uptime & Alerting
A green ping check only proves a kernel answered. What ICMP, TCP, HTTP and synthetic checks each catch, and a layered plan for one web service.
Continue reading Ping is not monitoring: ICMP, TCP and synthetic checks compared ›
Network UPS Tools from the USB cable to a tested shutdown: drivers, upsd, upsmon, what upsc tells you, and which box should go down last.
Continue reading UPS monitoring with NUT: clean shutdowns for a home rack ›
A good runbook is the calm colleague who is awake when you are not. A six-part template, writing tips, and a full example for a core uplink.
Continue reading Runbooks an on-call engineer can use at 3 a.m. ›
99.9 percent of a 30-day month is 43.2 minutes. The window, the maintenance clause and the place you measure from decide the rest.
Continue reading What 99.9 percent really allows: uptime maths and how SLAs are measured ›
A page should mean someone has to act now. Symptom alerts, severity tiers, hysteresis, dependencies and three bad alerts rewritten.
Continue reading Alert fatigue: writing alerts people still read six months later ›
