Problem
A fleet of 1,000+ Linux VMs needs one place to see inventory, application health, certificate expiry, software versions and patch status.
Architecture
Dependencies (text)
- collector -> fleet [SSH]
- health -> fleet [HTTP and service registry]
- collector -> db [writes inventory]
- health -> db [HEALTHY, UNHEALTHY, STALE]
- portal -> db [reads]
- portal -> servicenow [syncs]
- mapper -> portal [reads context]
- mapper -> depyaml [generates]
- renderer -> depyaml [renders]
Python and FastAPI with server-rendered pages and SQLite, collecting data over SSH.
What it does
- Health checks every 30 minutes on application hosts, with retries and clear HEALTHY, UNHEALTHY and STALE states.
- Views for inventory, certificates, patch status and end-of-life hosts, expected versus open ports, software inventory and a service map.
- An admin area to refresh certificates on demand.
- ServiceNow synchronization.
- In progress: AI-assisted application dependency mapping that generates YAML service definitions and renders them as architecture diagrams automatically.
Decisions
- Complement the monitoring stack, do not replace it.
- YAML as the contract for dependencies: readable, reviewable and versioned.
- Safe by default. Configuration through environment variables and a dry-run mode for the health batch.
Result
- Covers 1,000+ VMs.
- Extended during my co-op and rebuilt as a full-time engineer.