Website checks show what a visitor can reach; server metrics help explain what is happening inside the machine. Server monitoring generally requires an approved agent or integration and sufficient administrative access. A shared-hosting login does not automatically permit installing server-wide software.
IN THIS ARTICLE
01 Confirm permission and scope
Check your monitoring subscription and server-management agreement. For a managed VPS or dedicated server, ask us whether the agent is already installed or should be deployed by support. Identify the operating system and avoid running an installation command intended for another platform.
- Open the server-monitoring area and choose the supported add-server process. Give the server a meaningful label and keep production and testing environments distinct.
- Obtain the current installation instructions directly from the monitoring dashboard or official documentation. Review the command and its destination before running it with privileged access.
- Install or enable the integration using the authorised administrator. Protect registration tokens and API keys; do not include them in public notes or screenshots.
- Confirm the server appears and reports fresh data. Check that its hostname, operating system and resource totals match the intended machine.
- Review CPU, memory, filesystem and network information. Enable only the additional service metrics relevant to your workload and supported by the integration.
- Configure actionable thresholds and recipients. An early capacity warning is more useful than an alert that first fires when the disk is completely full.
02 Combine internal and external evidence
Keep a public website check alongside server metrics. A healthy agent cannot prove that DNS, TLS or a firewall allows customers to reach the site. Likewise, a working cached page can conceal a struggling database or background worker.
What should I check when server metrics stop arriving?
Check the last-report time, server availability and agent status through the supported management method. Review whether a firewall, operating-system update or credential change affected the connection. Avoid installing duplicate agents before confirming the existing configuration.
After replacing or retiring a server, remove the old monitor and revoke credentials that are no longer needed. Confirm new alert routing and labels so your team does not mistake a decommissioned machine for an active production incident.