How the agent works
Four steps, all started from your server.
Install
Creates the agent service, a local config and a host token.
Collect
Reads metrics and the modules you enabled, on the host itself.
Push
Sends telemetry to Fivenines over HTTPS. No port to open.
Receive settings
Gets module settings and intervals back from Fivenines.
What the agent sends
Operational telemetry goes out. Nothing comes back in that could act on your server.
- Host identity: hostname, OS, kernel, architecture, IPs
- Resource metrics: CPU, memory, disks, network, temperatures
- Top process names by CPU and memory
- Optional modules, each off until you turn it on
Four hard limits
What the agent will never do, whatever is enabled.
No remote shell
The dashboard cannot SSH into your server or run commands through the agent.
No inbound listener
Nothing listens for connections. No agent port is exposed.
No arbitrary file reads
No source code, SSH keys or shell history. Only what a module points at.
No database rows
Database modules read operational stats, never the data in your tables.
Logs only if you ask for them
Log monitoring is off by default. When you turn it on, you choose exactly which services it may read.
Frequently asked questions
No. It pushes metrics over HTTPS, accepts no inbound connections and does not run commands from the dashboard.
The standard installer uses sudo to set up a system service. A user-level install works without sudo, with less coverage for modules that need elevated reads, such as S.M.A.R.T. and RAID.
Yes. Proxmox uses a read-oriented API token, PostgreSQL and MySQL use monitoring views, MongoDB needs the clusterMonitor role, and S.M.A.R.T. and RAID can be limited to specific sudoers commands.
Only through two opt-in log modules on Linux, for the units you list. Masking is best effort, so keep units that log sensitive payloads off the list.
Yes. The agent is open source on GitHub, so you can audit it before you install it.
Something else on your mind? Talk to us →