Netquirk Monitoring · mon.netquirk.com

Uptime checks and host telemetry, in one signed-in workspace.

Netquirk Monitoring runs synthetic probes from three worker regions (us-east, eu-west, ap-south) and collects host metrics through a single static Go binary, the mon-agent. Probes cover ICMP, TCP, UDP and HTTP; the agent publishes per-core CPU, RAM, per-disk I/O, per-interface network, LVM and Btrfs volume metrics.

Probes from us-east · eu-west · ap-south Push-metric profiles · agent / statsd / prometheus

What it is

A monitoring surface for the rest of the Netquirk stack.

Synthetic probes from three regions

Each probe type runs from all three worker regions in parallel; the public dashboard surfaces per-region latency and success so a single CDN outage does not look like a single-region blip.

Push-metric profiles

The push endpoint accepts three named profiles: agent (the mon-agent Go binary), statsd (line-oriented text over TCP), and prometheus (text exposition). Each profile declares the keys it surfaces; unknown keys are accepted or rejected per profile.

Public status pages

A status page exposes the operators' chosen monitors to the public. Each page is reachable on its own URL under /status/<label>; the rendered page is purely server-side and survives a CDN origin hiccup.

Auth and signing keys

Sign in with your Netquirk account through OIDC; the JSON API uses API keys prefixed nq_ over bearer-token authentication. Both surfaces share the same monitor database — there is one source of truth, not two.

Probe types

Six probe kinds, all backed by the same dispatcher.

Each monitor has a single kind. The kind chooses what the worker sends and what counts as a pass.

icmp — round-trip time and packet-loss threshold against an IPv4 or IPv6 target.
tcp — three-way handshake against a host:port; timeout and optional pre-ACK banner match.
udp — payload send + optional reply regex; useful for DNS and other stateless services.
http — full HTTP/HTTPS probe with method, expected status class, optional body regex, optional JSONPath assertions, and TLS expiry window.
metrics — counts and stores metrics pushed to the push endpoint, evaluated against a profile-defined set of allowed keys and per-key thresholds.
agent — alias for metrics with the agent profile; identifies monitors backed by the Go agent.

Install the agent

One shell command, run on each Linux host.

The installer is the canonical source on the netquirk/mon-agent GitHub repository. It places a single static binary, registers a systemd unit, and starts pushing metrics on the interval you pass to -interval.

$ curl -fsSL https://raw.githubusercontent.com/netquirk/mon-agent/main/install.sh \
 | sudo bash -s -- -id "<monitor-uuid>" -interval 30

Replace <monitor-uuid> with the monitor id shown on the monitor's "raw agent metrics" page. The interval is in seconds; thirty is a sensible default, smaller values cost more disk per day per host.

JSON API

Everything the dashboard does, scripted.

The JSON API is documented at /api/docs and lives under /api/v1/. Authentication is bearer-token with an API key prefixed nq_, scoped per monitor. Programmatic callers can list, create, update, pause and delete monitors; pull response-time samples from the metric store; and reach the public read-only endpoints (/healthz, /metrics, status badges) without any token at all.

Read the API reference Hit /api/v1/health

The dashboard lives behind the same Netquirk account that signs you in to netquirk.com. There is no separate signup form to fill in.