What is Server monitoring?

Updated April 20, 2026

Server monitoring is a Hyperping product that installs a lightweight agent on your Linux and macOS hosts and streams system metrics back to your dashboard. Each install creates one server entry where you can see CPU and load averages, memory, filesystem usage, disk I/O, network throughput, and host metadata, without managing your own collector.

Platforms

Linux (systemd) · macOS (launchd)

Architectures

amd64 · arm64

Cadence

30 second scrape

Transport

OTLP/HTTP with on-disk queue

Enrollment

Install token (stable or bound)

Footprint

~50 MB RSS target

📘Sits alongside your HTTP monitors

HTTP monitors tell you a page stopped loading from the outside. The agent tells you the machine ran out of memory two minutes earlier. Use both together for full coverage.

What it collects

The agent embeds an OpenTelemetry collector that scrapes system metrics every 30 seconds and ships them to Hyperping over OTLP/HTTP. The first data point reaches the dashboard within a few seconds of enrollment.

Metric categories 6 groups

CPUpercent / load

Utilization percentage, 1, 5, and 15-minute load averages, logical core count.

Memorybytes / percent

Used, available, total, and utilization as a percentage.

Filesystembytes

Used and free space for every real mount point, rendered as a per-mount table in the dashboard. Utilization percentage is computed in the UI from these two values.

Disk I/Obytes

Bytes read and written per block device, with a direction attribute.

Networkbytes

Bytes in and out per interface. The dashboard derives TX and RX rates automatically.

Systemseconds / string

Uptime, boot time, hostname, OS and distribution, kernel, architecture, CPU model and vendor, agent version. Powers the dashboard's System Info panel.

📘Not ingested today

Per-process CPU and memory, per-interface packets and errors, per-device disk operation counts, and swap / paging are on the roadmap. The agent can emit them but the ingestor drops them. See Metrics collected for the exact accepted list.

Queue resilience, not a metric: the collector persists unsent batches to /var/lib/hyperping/queue so metrics survive ingest outages, reboots, and network blips.

For the full list of fields and units, see Metrics collected.

How enrollment works

Step 01

Generate a token

Add a server in the dashboard. Hyperping returns a single-use install token scoped to that server entry.

Step 02

Exchange for an agent token

The installer hands the install token to Hyperping and gets back a long-lived per-host agent token. From that point on the host authenticates as itself, not as the shared install token.

Step 03

Rotate by re-running

Running the installer with a new install token re-enrolls the agent and revokes the old credential. Same server UUID, clean rotation.

Next steps

Quickstart

Getting started →

Add a server, run one command, see metrics in about 30 seconds.

Reference

Install the agent →

Paths, service names, platform notes, and troubleshooting.

Schema

Metrics collected →

Every field and unit the agent reports, plus collection cadence.

Removal

Uninstall →

Stop the agent, purge state, and remove the server entry.

Getting started →