# Add a server you already have

> Not every server lives in a cloud account. You might have your own hardware in an office or a rack, or a machine at a provider InfraNest does not connect to…

Source: https://infranest.io/docs/add-a-server-you-already-have
Last updated: 2026-09-19

---

Not every server lives in a cloud account — you might have your own hardware in an office or a rack, or a machine at a provider InfraNest doesn't connect to yet. This article shows you how to add one of those and monitor it just like any other server.

## Overview

- Add any server you already have, even if InfraNest has no cloud account for it
- It's tracked in your monitoring list, marked **Self-managed**
- Install the agent afterward to get full monitoring: CPU, memory, disk, services, processes and alerts
- Power actions, backups, snapshots and cost tracking don't apply, since there's no provider account behind it

## Add it

1. Go to **Monitoring** and choose **Add monitor**.
2. Pick **A server** from the monitor types.

<!-- screenshot: add-a-server-you-already-have -->

<!-- docs-screenshots:start:add-a-server-you-already-have -->

![Add a server you already have](/media/196cda29-deae-4037-baa2-1e6a092248e3)

<!-- docs-screenshots:end:add-a-server-you-already-have -->

3. Choose it from the list — or, if it is not there, select **It is not in this list** and give it a name.
4. Choose how to collect its numbers. For a server with no cloud account behind it there is only one answer, and the page says why.
5. Press **Turn monitoring on**.

A name is the only thing we need. You can add an IP address and a location too, purely so you recognise it in the list — we do not connect to either.

The server appears in your monitoring list straight away, marked **Self-managed**.

It is deliberately not on the **Servers** page. That page is for machines in a connected cloud account — powering them on and off, backups, snapshots, volumes, cost — and none of that exists for a server we have no account for. A count on that page links back here, so it is never lost.

## Remove it again

A server you added by hand exists nowhere but InfraNest, so deleting it here is the whole of it — nothing happens to the machine, because there is no account for us to reach it through.

1. From **Monitoring**, open **Servers**, then open the ⋯ menu on its row and choose **Delete server**.
2. Or open the server and use the ⋯ menu at the top.
3. Or, on the **Servers** page, open it and use **Power → Delete server** — there you type the name to confirm, like any other delete.

Deleting removes its record, everything measured about it and its alert rules. If you would rather keep the history, **Stop tracking** hides it and keeps all of it — see [Remove a server from InfraNest](remove-a-server).

> [!WARNING]
> Deleting a self-managed server removes its full history and alert rules permanently. If you might want the history later, use **Stop tracking** instead.

## Then install the agent

On its own, an added server is just a name. The agent is what makes it useful — install it and you get everything InfraNest measures on any other machine:

- CPU, memory, disk space and load
- Which services are running, and why any of them failed
- The heaviest processes
- What is filling the disk
- Alert rules, and a warning if the server stops reporting at all

See [Install the server agent, step by step](install-the-agent-step-by-step). The command is the same one you would use anywhere else.

## What a self-managed server cannot do

There is no provider account behind it, so there is nothing for us to ask on your behalf. That means no:

- power actions — start, stop and reboot
- backups or snapshots
- rebuilds or rescue mode
- costs

Those controls are greyed out rather than hidden, so it is clear they exist and simply do not apply here.

Everything the agent measures works exactly as it does on a cloud server. The agent has never cared where a machine runs.

## Tips

> [!TIP]
> If your server isn't in the list InfraNest offers, just choose **It is not in this list** and give it a name — that's all it takes to start tracking it.

## Why not just pick a provider from the list?

Because it would be wrong everywhere afterwards. The server would carry another company's logo, its cost would be counted against an account it is not in, the monitoring page would claim we were collecting from a provider API that will never answer, and the power buttons would offer to reboot something we cannot reach.

**Self-managed** says the true thing instead: we know about this machine, and we do not manage it.
