Laravel System Resources Laravel Package

Getting Started

Install the package and let Laravel auto-discover the service provider:

composer require stepanenko3/laravel-system-resources
php artisan vendor:publish --tag=system-resources-config

Look first at config/system-resources.php — that's where you control which probes run (CPU load, memory, disk) and whether the HTTP route is registered. Then check the package's routes/ directory (or php artisan route:list | grep -i resource) to see the exact URI and middleware applied out of the box.

First use case: hit the JSON endpoint from your ops dashboard or an uptime monitor to get a cheap server-health snapshot without installing Prometheus:

curl -s https://yourapp.test/system-resources | jq

If the route is disabled by default, resolve the underlying class directly instead:

use Stepanenko3\SystemResources\SystemResources;

return SystemResources::make()->all(); // or ->cpu(), ->memory(), ->disk()

Implementation Patterns

Dashboard polling. Point a small Vue/React widget at the endpoint on a 10–30s interval. Because each call may shell out for metrics, wrap it in Cache::remember('sys-resources', 15, ...) at the controller level rather than hitting the package on every request.

Scheduled health logging. Use a scheduled command to persist metrics for trend lines, which is far more useful than a live snapshot:

Schedule::call(function () {
    SystemResourceLog::create(SystemResources::make()->all());
})->everyFiveMinutes();

Guarding the route. Always place the endpoint behind auth + an admin gate, or restrict by IP in config/system-resources.php. A public metrics endpoint leaks hostnames, disk paths, and load characteristics — treat it as internal-only and prefer registering it on an internal-only route group.

Alerting integration. Return the array into your existing notification pipeline (Slack, Mail, Healthchecks) and threshold on disk.used_percent or memory.used_percent rather than raw bytes — absolute numbers don't port between hosts.

Multi-server fan-out. If you run several nodes, the package only reports the host it executes on. Either expose the endpoint per-node behind a shared secret and aggregate centrally, or run the scheduled-logger pattern and ship rows to a central store.

Blade/Direct access. For a quick internal admin page, inject the class into a controller and pass the array to the view; no need to consume your own HTTP endpoint.

Gotchas and Tips

• v1.1.1 is a patch release with no documented changes in the changelog beyond the compare diff. Treat it as a safe in-place bump from v1.1.0, but read the actual commit diff before upgrading if you depend on exact output keys.
• Platform assumptions. System metric collection on Linux typically shells out to commands or reads /proc; Windows and containerized environments frequently return zeros, null, or throw. Validate output on your actual runtime image before wiring it into alerts.
• exec() may be disabled. Shared hosting and hardened PHP-FPM configs often have disable_functions=exec,shell_exec. If metrics come back empty, that's the first thing to check — not the package config.
• Permission and container quirks. In Docker, disk numbers reflect the container's overlay filesystem, not the host. Mounted volumes and cgroup memory limits can make memory.total misleading.
• Config caching. After editing config/system-resources.php, run php artisan config:clear (or config:cache in deploy) — a stale cached config is the usual cause of "my change did nothing."
• Cost of polling. A system-resources call per page-load per user adds up. Cache aggressively; metrics don't need sub-second freshness.
• Extension point. If the built-in probes miss something (GPU, queue depth, Redis memory), subclass or decorate the resource class in a small service provider rather than forking the package — keeps future patch upgrades (like this one) painless.
• Don't assert on exact values in tests. Mock or bind a fake implementation of the resource class in your test container; real CPU/disk values are non-deterministic and will flake CI.