An update can be available, installed, or actually running. How to make monit watch all three with small shell checks, and how to prove each check really fires.
Monitoring
Knowing about a problem before visitors do: sites that stop answering, certificates near expiry, full disks, failed jobs and updates that wait.
Monitoring means checking a system on a schedule and raising an alert when something drifts from normal. The usual subjects are whether a web server such as nginx still answers, whether a certificate is close to expiry, whether a disk is filling up and whether a scheduled job actually ran.
I would rather hear about a problem from a check than from a customer. I watch my servers with monit, a small open-source tool that runs its checks in cycles and can act on the exit code of any script. In November 2026 I improved how it tells me about pending updates.
Updates taught me a trap. An update can be available, installed or actually running, and only the last state counts for security. A new kernel waits for a reboot, and a long-running service keeps an old library in memory until it restarts, so a check has to look at what runs, not only at what is installed.
Two habits help. I make every check fail once on purpose before I trust it, because a check that has never fired is a guess. And I let monitoring alert rather than repair: restarting a database is a decision for a person, which is where I draw the line in automation too.
2 blog posts
Longer articles about Monitoring
Practical lessons from building two public, read-only MCP servers: the smallest protocol that works, the mistakes that are easy to make, and what helped people find them.
1 update
Everything I posted about Monitoring, newest first
PersonalNew post: Let monit tell you when an update is waiting
New
An update can be available, installed, or actually running. How to make monit watch all three with small shell checks, and how to prove each check really fires.