Appearance
Agents
The OpsMerge agent is the small native binary that runs on every endpoint you manage. It's how OpsMerge gathers telemetry, runs scripts, applies policies, and surfaces alerts.
Regional ingest hosts
Articles in this section use ingest-eu.opsmerge.cloud for examples — the EU cell's agent ingest host. If your tenant is in another region (e.g. US), the ingest hostname differs (ingest-us.opsmerge.cloud, and so on). The Add agent dialog in OpsMerge always shows the correct hostname for your tenant — copy from there.
What the agent does
- Telemetry. Hardware inventory, installed software, running processes, disk space, network interfaces, OS version. Reported on connect and on change.
- Monitoring. CPU, RAM, disk, service status, custom checks defined by your policies. See Monitoring & alerts.
- Scripts. Runs PowerShell (Windows), bash (Linux/macOS), and Python scripts pushed from OpsMerge. See Scripts & automation.
- Updates. Self-updates on a rolling schedule. See Updates & rollouts.
- Remote control. Brokers OpsMerge Connect sessions (RustDesk under the hood) when an operator initiates one. See Remote control with Connect.
How it runs
| Platform | Service | Run as | Binary path |
|---|---|---|---|
| Windows | RMMAgent (Windows Service) | LOCAL SYSTEM | C:\Program Files\RMM\rmm-agent.exe |
| macOS | cloud.opsmerge.agent (launchd) | root | /usr/local/bin/rmm-agent |
| Linux | rmm-agent.service (systemd) | root | /usr/local/bin/rmm-agent |
Config and state live in:
- Windows:
C:\ProgramData\rmm\(config, logs, working data) - macOS:
/etc/rmm/(config),/var/log/rmm-agent.log(logs) - Linux:
/etc/rmm/(config),/var/log/rmm-agent.log(logs)
How it talks to OpsMerge
Two channels:
- HTTPS to
ingest-eu.opsmerge.cloudon port 443 — used for binary downloads, version-check, and one-shot REST calls during registration. Not proxied via Cloudflare so curl/PowerShell installers work without bot-protection interference. - NATS over TLS to
ingest-eu.opsmerge.cloudon port 4222 — the live, low-latency channel for telemetry, scripts, and remote-control brokering. Persistent connection, reconnects automatically.
If port 4222 is blocked but 443 is allowed, the agent registers fine but never appears as "Online". That's the most common cause of "the install said it worked but the agent isn't there" — see Troubleshooting agents.
Installing
- Windows — PowerShell one-liner or signed MSI
- macOS — curl one-liner, launchd
- Linux — curl one-liner, systemd
- MSI deployment — for Group Policy, Intune, and other deployment tools
After installation
- Updates & rollouts — how the agent updates itself
- Troubleshooting agents — common issues and how to diagnose
- Policies — controlling what the agent monitors and what runs on the box
Resource cost
The agent is intentionally small:
- On disk: ~25 MB (binary + updater).
- RAM: typically 30–60 MB resident.
- CPU at idle: under 0.5% on most hardware.
- Network at idle: a few KB/min of telemetry, mostly heartbeats.
- Network during script runs: depends on the script's output; typically under 100 KB.
You can run the agent on anything from a Raspberry Pi to a Windows Server without it being noticeable.