Comment by Abhinav Sharma on "How I audit 50 servers with one Ansible command"

Abhinav Sharma

Founder of Impact Gate. See who breaks before you change an API.

"An unreachable server is a result, not a crash" is the line that separates an audit from a health check - the box that doesn't answer is usually the one that matters most. changed_when: false is the other honest touch: nothing is worse than a read-only run that shows 50 "changed". Of the five problems you hid, which did the report rank highest, and did any hide below ones that only looked worse?

Opeyemi Adeniji

DevOps and SRE projects, built in public

8h ago

Thanks Abhinav, glad those two lines landed. The report ranked db-02 (UNREACHABLE) first, then the three CRIT hosts, then the WARN, then OK. So nothing hid below something worse across levels. Within a level, though, the order is only alphabetical, so web-07 (disk 92% full) sits below app-11 (app server down) because “app” sorts first, not because the report judged it less severe. That’s a limitation. A better version would rank inside each level by number of issues or by weight (service down above a nearly full disk). Good question, it’s going on my list for the next release.

Alphabetical within a level is a fine v1 - at least it's predictable. The weighting gets interesting fast though: "app server down" outranks "disk 92%" until the disk hits 100 and takes the app down with it. You're really ranking time-to-page, not severity. Curious what you land on.

Search Hashnode