AgentWatch Blog
Practical notes on server monitoring, uptime and alerting for small ops teams.
CPU, RAM and disk alert thresholds: what numbers make sense?
Recommended CPU, RAM and disk alert thresholds by workload: web server, database, CI runner. Why each number differs, plus hysteresis and consecutive samples to kill false alarms.
Read article →
How to write a Docker HEALTHCHECK properly
A guide to HEALTHCHECK in Dockerfile and docker-compose: what interval, timeout, retries and start_period mean, working examples, and common mistakes like health endpoints that call external services.
Read article →
How to monitor a Linux server without running Zabbix
An honest comparison of 3 ways to monitor a Linux server: self-hosting Zabbix or Prometheus, writing cron scripts, or using a hosted service with an agent. Pros, cons and who each path fits.
Read article →
Server down? A 7-step incident checklist that works
A 7-step checklist for handling a server outage, with real Linux commands: systemctl, journalctl, df -h, free -m, docker ps, ss -tlnp. Follow the order to recover fast without destroying evidence.
Read article →
What does 99.9% uptime mean in real downtime?
Exact conversion table for 99% / 99.5% / 99.9% / 99.95% / 99.99% uptime into allowed downtime per day, month and year, plus how to calculate it yourself and pick the right target.
Read article →
What is a status page, and when does a small business need one?
A status page is a public page showing the live health of your service. What goes on one, the real benefits for a small business, and 4 signs that you already need it.
Read article →