Argus

We notice it first

Argus watches the systems we run for you — servers, services and the application itself. We see bottlenecks before they turn into outages, and in an emergency we already know what needs doing.

in development · first systems connected

01 — What Argus watches

Four layers, one picture

Infrastructure

CPU, RAM, disk and I/O; MySQL, Redis, OpenSearch and the PHP-FPM workers; availability, response times and TLS certificate expiry. Every reading has a threshold that triggers something.

The application itself

Is the Magento cron running or has it been stuck for days? Are the indexers current? Are the queue consumers keeping up or are messages piling up? How high is the full page cache hit rate, and did the last deployment complete cleanly? Precisely the questions that decide revenue in a shop.

Versions and upgrade path

A running inventory of every version — shop system, Composer packages, PHP, MySQL, operating system — compared against current releases and end-of-support dates. The result is a prioritised list: security patch, feature release, or something that can wait.

Configuration and sizing

Argus reads the configuration that is actually running — FPM pools, OPcache, MySQL variables, Redis, OpenSearch heap — and holds it against measured load. Out of that come concrete recommendations: more workers, a different cache, a larger buffer. Or the welcome variant, that a smaller server will do.

02 — What comes of it

Readings become a decision

One overview

Every system we look after at a glance, with a detailed view per system.

Alerts reach us

Not you. Ideally you hear about a problem once it has been solved.

Recommendations with a status

Open, implemented, deliberately rejected. Rejection is documented too.

Regular assessment

Logs and readings are pre-processed and condensed into one recommendation per system. Pattern recognition and language models help sort the picture — what follows from it is our decision.

Why we built it

Standard monitoring reports CPU load and ping times. What interests us is Magento's cron groups, the backlog of a message queue and the end of support for the installed PHP version — the questions that actually matter in day-to-day shop operations. So we built the tool that answers them.

Laravel · Filament · Prometheus and Grafana as the metrics layer · agentless access over SSH — nothing of ours runs on the monitored systems