AB 4, Linux tool, solo, 25 stars

Self-healing monitor

A Bash script that checks CPU, memory and disk every five minutes, restarts the Docker daemon when it dies and charts the history on a small dashboard.

Observed: Dec 2025 to Feb 2026
Role: Everything. 28 commits.
Made with: Bash
cron
Docker
Chart.js
GitHub Actions
ShellCheck
Links: Repository

Context

Services crash at 3 AM when nobody is watching. I wanted the machine to notice, fix the obvious case on its own and keep a record of what happened.

Idea

Cron and plain files, no agent and no database. The script writes metrics.json, a CSV history and an alert log, and the dashboard only reads them.

Craft

A warning at 80% and a critical alert at 90%, with a five minute cooldown so one bad spike doesn’t send twenty alerts. CI runs ShellCheck on every pull request.

Outcome

It installs with one crontab line, ships with a Dockerfile, and its log shows the daemon going down and coming back on its own.

Plate 1The monitor dashboard: CPU 12%, memory 54%, disk 13%, a 15 minute history chart and a log reading “docker daemon down” then “docker daemon restarted”

Plate 2The repo’s Bash CI runs on GitHub: “fixed the shell check” and “fixed the shebang” fail ShellCheck, then “fixed the json file” passes

Plate 3The system-monitor repository on GitHub: analyse.sh, dashboard.html, a Dockerfile and the CI workflow, 28 commits

Selected stars

Real commits from this figure, with their messages as I wrote them. Each opens on GitHub.

Next figure, AB 5

Upstream