Blog
Notes from the on-call rotation
Field notes on incident response, escalation design and running SkyLogs in production.
Alert fatigue is an ownership problem
Reducing noise is only half the work. Until every signal has an owner, quiet dashboards just hide the same risk.
2026-07-28 · 6 min
- On-call
Designing escalation chains that actually get answered
Timeouts, fallbacks and manager steps: how to build a chain your team trusts at 3am.
2026-07-09 · 8 min
- Deployment
Self-hosting SkyLogs on Kubernetes
A walkthrough of the Helm chart, persistence, and safe upgrade paths.
2026-06-22 · 11 min
- Concepts
Alert vs. incident: why the distinction matters
An alert is a signal. An incident is a response with an owner, a clock and an end.
2026-06-03 · 5 min
- Notifications
Routing notifications by severity, not by habit
Phone for critical, Slack for warning, email for info — and why the mapping matters.
2026-05-19 · 7 min
- Practice
Measuring on-call load without punishing people
Fair rotations start with visible data: pages per shift, night interrupts, handoffs.
2026-05-02 · 9 min
Prefer to see it working?
The live demo walks through a full incident, from alert to resolution.