Low, predictable overhead
Bounded queues and configurable sampling budgets reduce optional telemetry before application workloads are affected.
A lightweight telemetry agent and collection system for heterogeneous computer fleets.
Discuss your fleet$ fwctl inspect HEL-VM-0421
status connected
platform linux/x86_64
agent 5.8.2
last_event 2026-10-01T19:42:18Z
queue_depth 0
queue_age 0s
sampling_budget 2.0%
collector hel-collector-03
session resumable
policy engineering-linux-v4
FleetWatch collects configured operating-system, hardware, process, and application metrics. What it does under pressure matters as much as what it collects.
Bounded queues and configurable sampling budgets reduce optional telemetry before application workloads are affected.
Agents journal locally when collectors cannot be reached. Upload resumes when connectivity returns.
No Java runtime, browser engine, database server, or local container environment is required. We consider this a feature.
Administrators specify what is collected, where it may be sent, and how long it may be retained.
The graphical administration console is optional. Nobody on the engineering team considers this unusual.
fw-agentEndpoint collection daemon
Monitored systemsfw-relayStore-and-forward relay for isolated networks and large sites
Site or network boundaryfw-collectorErlang/OTP ingestion service
Backend clusterfwctlCommand-line administration and diagnostics utility
Operator workstationfw-exportBatch export for customers who prefer to analyze their own data
Wherever requestedDisconnected systems are computers that happen not to have a route at the moment.
Include platforms, network constraints, retention needs, and the machines nobody is allowed to reboot.
sales@insight-technology.org