Install the agent

Requirements

• •A Linux server on amd64 or arm64 with systemd. The installer stops with a clear message if either is missing.
• •Root or sudo. The agent installs a system service. On shared hosting without root the command fails; use an uptime monitor for those sites instead.
• •curl on the server.
• •Outbound HTTPS (port 443) to `api.getopservo.com`. No inbound port is opened.
• •OpenSSL 3.x or python3, used to verify the Ed25519 signature on the download. Older distributions such as Ubuntu 20.04 ship OpenSSL 1.1 but have python3, so no extra step is needed. If neither is present the installer refuses to continue rather than install an unverified binary.

Add the server in the dashboard

1. 1.Open your workspace and click Add a new server.
2. 2.Keep Linux server selected. Optionally type a name; leave it blank to use the host’s own hostname. You can rename it any time.
3. 3.Click Generate install command. The dialog shows a command with a fresh enrollment token.
4. 4.Click Copy, then paste the command into a root shell on the server. If someone else manages the box, use Email these instructions to my developer.
5. 5.The dialog shows Waiting for the server to check in… and then Server connected and added with a View it link. The fleet grid updates on its own.

The token is single-use and expires after one hour. If it expires, generate a new command; nothing on the server needs cleaning up. If the dialog says the plan limit is reached, upgrade or remove a server you no longer monitor. See Plans and billing.

The command

curl -sSL https://api.getopservo.com/install.sh | sudo env OPSERVO_API_URL=https://api.getopservo.com OPSERVO_TOKEN=<token> bash

Replace <token> with the value from the dialog. The dashboard fills it in for you; you never type it by hand.

What the installer does

1. 1.Detects the distribution and CPU architecture.
2. 2.Downloads the signed update manifest and the matching agent binary from Opservo, verifies the Ed25519 signature and the SHA-256 checksum, and refuses to install if either fails.
3. 3.Registers the server with your workspace using the token, which is consumed in the same step.
4. 4.Writes /etc/infrapilot/config (root-only, mode 0600) with the API URL, the agent ID and its secret key.
5. 5.Installs the binary at /usr/local/bin/infrapilot-agent.
6. 6.Installs the systemd unit infrapilot-agent.service, enabled on boot and set to restart on failure.
7. 7.Starts the agent and prints the agent ID and a link to its page in the dashboard.

On a server with an Intel integrated GPU the installer also tries to install intel-gpu-tools so GPU load can be read. This is best effort and never fails the install.

Confirm it is running

systemctl status infrapilot-agent
journalctl -u infrapilot-agent -f

A healthy log shows the agent starting and sending heartbeats every 60 seconds. The first snapshot usually lands within a minute. If the service is not active, the journal output names the reason. See Troubleshooting.

Re-running is safe

Running the installer again on a server that already has /etc/infrapilot/config keeps the existing identity and only upgrades the binary. You do not need a token for that; the server page also offers Re-install / upgrade agent…. Normal updates happen by themselves once an hour. See Agent updates.

One server, one workspace

A server reports to exactly one workspace. If you run an Add server command from a different workspace on a server that is already registered, the installer does not create a duplicate. It pauses for up to 10 minutes while the dialog that generated the command asks you to choose Move it to this workspace or Leave it where it is. If you move it, its entry in the previous workspace is archived (and can be restored there) and the server starts reporting to the new one. If you leave it, nothing changes.

Running a command from the same workspace on an already-registered server simply completes the dialog with the existing entry.

Other ways to install

• •Auto Scaling groups and golden AMIs. Use a reusable fleet key in launch user data instead of a one-hour token. Clones of an installed image register as new servers by themselves. See Auto Scaling.
• •Kubernetes. Choose Kubernetes cluster in the Add server dialog and run the generated kubectl apply command. See Kubernetes.