Choose where the monitor lives
Place the monitor where agents can reach its HTTPS URL. Agents need outbound access to that endpoint; a server address description in the dashboard does not replace the agent’s -url argument.
Your browser connects to the monitor rather than an individual agent. This separates display from collection and gives you one interface for desktop, tablet and phone.
Create a unique identity per node
Add a server in Servers, then store its issued ID and token on that machine. Repeat for the next server instead of copying the first token. Each agent needs its own state directory.
Choose names that distinguish locations and workloads. The OS is detected automatically. Avoid running multiple agents with one identity: their reports would belong to one node record.
Verify telemetry and update behavior
Select a server and check its latest report, resources, processes and logs. The selected server persists across reloads while its record exists.
Browser display intervals differ from agent collection. The default agent interval is 500 ms; -interval accepts 500 ms through 5 s. Real-time display cannot recover unavailable metrics or eliminate network latency.
Configure modules and backups
Module assignments belong to individual servers. One node can use a Unix socket for Docker, another a Windows named pipe, and a third no Docker module. Verify module results separately on each node.
Back up the monitor’s persistent data directory and protect the backup like the original data. Keep releases and configuration before upgrading. When rotating a token, update the appropriate agent token file and restart that agent.
Troubleshoot one offline node
Check the affected agent’s service, URL, ID, token, TLS and outbound connectivity first. The dashboard marks a node offline after 15 seconds without reports. That is a report freshness signal, not a diagnosis of the underlying problem.
Compare manual execution with service execution and review permissions. Filters and refresh intervals cannot fix a revoked token or stopped service.
These instructions are based on the release documentation linked below. Commands containing example.com or SERVER_ID are templates: substitute values from your own dashboard. Check the matching release documentation when upgrading.