monitoring

Logging Field Guide

How to instrument a service so the people who have to read it later — on call, in an audit, during an incident — can actually find what they need.

The Dashboard Was Telling the Truth

A retail software team's perfectly-coded webhook module kept failing mysteriously every Monday—until they realized their metrics were measuring the wrong things and missing real-world problems.

How a Hypervisor Actually Shares CPU and I/O: A Practical Guide for KVM, LXC and Proxmox

vCPUs are just threads, cgroups don't replace the scheduler, and on NVMe the right I/O scheduler is usually 'none'. A ground-up tour of how Linux shares physical resources between virtual machines and containers — and which knobs are actually worth turning for small-to-medium workloads.