Skip to content

Presentation

A single agent, deployed on a server you control, that turns the state of your infrastructure into clean, queryable telemetry. It collects metrics, status changes and logs from operating systems, backup software, hyperconverged appliances, virtualization stacks and network gear — then exposes them through whichever monitoring or observability tool you already operate.

In one paragraph

You install a single binary on a host that has network access to the systems you want to monitor. A YAML file declares which probes the agent should run and where the values should go (PRTG / Nagios HTTP endpoints, Prometheus /metrics scrape page, a native Zabbix agent, OTLP push over gRPC or HTTP). The agent starts collecting on its own schedule and stays out of the way — restartable, observable, and runnable in air-gapped environments where no callback to the SenHub backend is allowed.

What the agent does

  • One collector for many sources. Each probe knows how to talk to one class of system (Veeam, Citrix, NetScaler, Redfish, Linux/Windows hosts, …). You add probes to the agent's YAML; you do not glue together vendor exporters.
  • One vocabulary across every output. Probes emit values into a shared in-memory cache. From there the agent serves PRTG / Nagios HTTP endpoints, exposes a Prometheus /metrics scrape page, acts as a native Zabbix active agent, and natively pushes OTLP metrics and logs, over gRPC or HTTP, to any OpenTelemetry receiver (collector, vmagent, Tempo, Grafana Cloud OTLP). Metric names, units and attributes match across all sinks, so a query that works in Grafana today keeps working in your PRTG sensor template tomorrow.
  • Tiered licensing. The free tier ships the whole universal collection tier — OS/host probes (CPU, memory, disk, network), log collection, network checks, and the application, database and broker probes (MySQL, PostgreSQL, Redis, Kafka, Docker, and more). The Pro and Enterprise tiers unlock the deep vendor integrations — Citrix, NetScaler, Redfish, Veeam, custom events, network synthetic checks.
  • Self-observability built in. Process resources, OTLP push health, collect-cycle counters and a system.processes.count metric are emitted on every output so the agent stays observable in the same panel as the systems it watches.

What the agent does NOT do

  • It does not ship dashboards by itself. Grafana dashboards shipped under docs/grafana/ are a separate catalog you can import; PRTG sensor templates are generated from the agent's channel definitions on the fly.
  • It does not host long-term storage. Metrics, logs and traces live in your own Prometheus / VictoriaMetrics / Tempo / Grafana Cloud backend.
  • It is not a router or proxy. Probes connect outbound to the systems they query; receivers like syslog accept inbound traffic but do not relay it elsewhere unless an OTLP push storage is configured.

Output paths at a glance

Path Protocol Use case
/api/{key}/prtg/metrics/{probe} HTTP(S), JSON Native PRTG sensor templates (Sensor URLs tab of the console)
/api/{key}/nagios/metrics/{probe} HTTP(S), text NRPE / Nagios performance line
/api/{key}/prometheus/metrics HTTP(S), text exposition Prometheus / VictoriaMetrics scrape
Zabbix active agent Zabbix protocol, TLS or PSK Push to a Zabbix server or proxy on 10051, autoregistration and discovery
OTLP client gRPC or HTTP, TLS or mTLS Push to OTel collector, vmagent, Tempo, Grafana Cloud OTLP

The three HTTP paths are served by the agent's http output, over HTTP on port 8080 by default, or HTTPS when TLS is enabled; {key} is the agent key. All five read from the same in-memory metric cache so the data is consistent regardless of which sinks you enable.

Where to next