monitoring checklists

A practical, Bash-first guide to monitoring AI on Linux: why resources, drift, latency, and reproducibility matter, plus copy-paste checks to ship today—host/GPU samplers, HTTP SLO probes, data-drift sentinels, reproducible version metadata via systemd, and fast Prometheus+Grafana dashboards/alerts—with apt/dnf/zypper installs, cron-ready scripts, and a step-by-step rollout to make silent failures loud.