How to monitor dedicated server performance

Quick answer: On an unmanaged dedicated server, OS-level monitoring is yours to set up: start with built-in tools (htop, iostat, df), add an agent-based system like Netdata, Zabbix or Prometheus node_exporter for history and alerts, and watch disk health via SMART. Cloud2Y monitors the network and hardware layer on its side.

Overview

Good monitoring answers three questions: is the server healthy now, what changed before an incident, and when will resources run out. On dedicated hardware also watch component health (disks, RAID state, temperatures) — you are running on real disks that age.

Before you start

• SSH access to the server.
• A decision: lightweight single-node dashboard (Netdata) vs central monitoring for a fleet (Zabbix/Prometheus + Grafana).
• Somewhere to send alerts — email or a messenger webhook.

Step-by-step guide

1. Baseline with built-ins:

   htop            # CPU/RAM per process
   iostat -x 5     # disk load and latency
   df -h           # disk space
   ss -tulpn       # listening services

2. Install an agent (example — Netdata, one-node dashboard):

   wget -O /tmp/netdata-kickstart.sh https://get.netdata.cloud/kickstart.sh
   sh /tmp/netdata-kickstart.sh

3. Watch disk health (SMART):

   apt install smartmontools
   smartctl -a /dev/sda

   On hardware RAID, query the controller (e.g. storcli / MegaCli) — disks behind the controller are not visible to plain smartctl.
4. Check RAID state regularly — software RAID: cat /proc/mdstat.
5. Set alerts for: disk >80% full, RAID degraded, load sustained above core count, RAM+swap pressure, SMART reallocated sectors growing.

Common issues

• Monitoring only CPU/RAM: most dedicated-server incidents are disk-related — always include SMART and RAID state.
• No history: a spike you cannot look up afterwards is a lost diagnosis — keep at least a few weeks of metrics.
• Alert fatigue: start with a handful of high-signal alerts; a muted channel is worse than none.

When to contact support

Open a ticket when metrics point at hardware: SMART errors, a degraded RAID array, ECC memory errors in logs, or abnormal temperatures — we will run diagnostics and replace failing components.

Frequently asked questions

Does Cloud2Y monitor my server for me?

Cloud2Y monitors the network and hardware environment on its side. Operating-system and application monitoring on an unmanaged dedicated server is set up and run by you.

What should I monitor first?

Start with disk space, disk health (SMART and RAID state), CPU load, RAM and swap pressure, and your key services — and configure alerts, not just dashboards you never open.

How do I check disks behind a hardware RAID controller?

Query the RAID controller with its own CLI, such as storcli or MegaCli. Plain smartctl does not see the physical disks hidden behind a hardware RAID controller.

Related articles

• How to request hardware diagnostics
• How to set up RAID on a dedicated server
• How to request a bandwidth upgrade
• Dedicated server SLA and support scope

Need a hand? Contact Cloud2Y support →