Monitoring VPS Resources (CPU, RAM, Disk)

When a VPS feels slow, the cause is almost always one of four things: CPU, memory, disk or network. Linux already ships the tools to tell which one, and a few small extras make them easier to read. This guide shows the commands to run on your own VPS, how to read their output, and how to set up a simple alert before a full disk or a memory leak takes your site down.

1. A thirty-second health check

Log in over SSH and run these four commands:

uptime          # load averages for the last 1, 5 and 15 minutes
free -h         # memory and swap
df -h           # disk space per filesystem
nproc           # number of vCPUs, to compare with the load

If the 5- and 15-minute load is well below your vCPU count, memory available is comfortable, and every filesystem is under about 80%, the server is not short of resources and the problem is more likely in your application.

2. CPU and load

The load average counts processes running or waiting to run. Compare it with your vCPU count: a load of 2.0 on a 2-vCPU VPS means fully busy; a load that stays above the vCPU count means work is queuing.

For a live view, use top (built in) or htop (friendlier):

sudo apt install htop        # Ubuntu or Debian
sudo dnf install htop        # AlmaLinux or Rocky Linux (enable EPEL first if it is not found)
htop

In top, press 1 for per-core usage, P to sort by CPU and M to sort by memory. The CPU line matters more than the load figure:

3. Memory and swap

free -h

Read the available column, not free. Linux uses spare memory as file cache (buff/cache) and hands it back when an application needs it, so low "free" memory is normal. Low "available" memory is not.

To find what is using memory, and to see whether the kernel has had to kill anything:

ps aux --sort=-%mem | head -n 11     # top 10 processes by memory
sudo journalctl -k | grep -i "out of memory"

An "Out of memory: Killed process" line means the server ran out of RAM and the kernel stopped a process, often your database. Either reduce memory use (fewer PHP-FPM workers, a smaller database buffer) or move to a plan with more RAM.

Watch swap activity rather than swap size:

vmstat 2 5

The si and so columns show memory moving to and from swap. Occasional small numbers are fine; steady non-zero values mean the server is short of RAM.

4. Disk space and disk speed

Space first:

df -h                                        # space per filesystem
df -i                                        # inodes: a disk can be "full" of small files
sudo du -xh / --max-depth=1 | sort -hr | head -15

The usual culprits are logs in /var/log, old backups, database files and Docker images (docker system df shows those). A database or mail server stops working when the disk fills, so act at 80% and treat 90% as urgent.

Then speed. Install sysstat for iostat and iotop for per-process disk use:

sudo apt install sysstat iotop       # or: sudo dnf install sysstat iotop
iostat -x 2 5
sudo iotop -o

In iostat, %util near 100 with a high r_await or w_await means the disk is the bottleneck. iotop -o shows only the processes doing disk work right now.

5. Network and connections

sudo apt install nload iftop         # or install them with dnf (EPEL)
nload                                # live bandwidth in and out
sudo iftop -i eth0                   # traffic per connection; check your interface name with: ip -br link
ss -s                                # connection totals
ss -tan state established | wc -l    # open TCP connections

A sudden jump in connections from a few IP addresses can be a bot or an attack rather than real visitors. Your web server logs and a firewall with rate limits help there; see the SSH security hardening checklist for the basics.

6. History with sysstat, and pressure stall information

The commands above show the present. sysstat also records history, so you can see what happened at 3 a.m.:

sudo systemctl enable --now sysstat     # start collecting (Ubuntu/Debian may also need ENABLED="true" in /etc/default/sysstat)
sar -u        # CPU today
sar -r        # memory today
sar -d        # disks today

Modern kernels also report pressure stall information, which tells you how much time work was delayed waiting for CPU, memory or disk:

cat /proc/pressure/cpu /proc/pressure/memory /proc/pressure/io

some avg10=0.00 means nothing was waiting in the last ten seconds. Values that stay in double figures point straight at the resource you are short of.

For an all-in-one screen, glances combines CPU, memory, disk, network and processes. For dashboards and alerting across servers, see Linux server monitoring software compared.

7. A simple alert script

This script checks disk, memory and load, and posts a message to a webhook URL of your choice, such as a Slack, Discord or Telegram bot endpoint. It needs only curl.

sudo nano /usr/local/bin/resource-alert.sh

#!/bin/sh
WEBHOOK="https://example.com/your-webhook"   # replace with your own endpoint
HOST=$(hostname)

DISK=$(df --output=pcent / | tail -1 | tr -dc '0-9')
AVAIL_MB=$(free -m | awk '/^Mem:/ {print $7}')
LOAD=$(cut -d ' ' -f2 /proc/loadavg)        # 5-minute load
CPUS=$(nproc)

MSG=""
[ "$DISK" -gt 85 ] && MSG="$MSG Disk / at ${DISK}%."
[ "$AVAIL_MB" -lt 200 ] && MSG="$MSG Only ${AVAIL_MB} MB memory available."
awk -v l="$LOAD" -v c="$CPUS" 'BEGIN { exit !(l > c) }' && MSG="$MSG Load $LOAD on $CPUS vCPU."

[ -n "$MSG" ] && curl -fsS -m 10 -H 'Content-Type: application/json' \
  -d "{\"text\":\"$HOST:$MSG\"}" "$WEBHOOK" >/dev/null

sudo chmod 700 /usr/local/bin/resource-alert.sh
sudo crontab -e
# add this line to run it every 5 minutes:
*/5 * * * * /usr/local/bin/resource-alert.sh

Adjust the thresholds to your server. If you would rather receive email, send it through an external mail service; whether outbound mail on port 25 works from your VPS is something to confirm with support first.

8. When to upgrade, and how

Tune before you upgrade: a slow database query, too many PHP-FPM workers or an unrotated log often explains the numbers. Consider a bigger plan when, over several days of normal traffic:

• the 15-minute load stays above your vCPU count;
• memory available stays under about 10% of RAM, or the kernel kills processes;
• steady swap activity shows in vmstat;
• disk use keeps climbing past 80% after clean-up.

To move to a bigger plan, open a ticket naming the VPS and the plan you want; support confirms the price difference first. Cloning and resizing your VPS explains what happens and how to grow the disk afterwards.

9. Domain India VPS plans

A Domain India VPS is self-managed: you get root access and look after the operating system, security updates, monitoring and backups yourself. There is no VPS page with graphs in the client area, so the tools in this guide are how you watch your server. Start, stop, force restart or anything you can't do from inside the server goes through a ticket; to reboot, use sudo reboot over SSH, never poweroff, which leaves the VPS switched off. See managing your VPS.

Domain India list prices on 19 September 2026, excluding 18% GST, per month: VPS Starter ₹553 (1 vCPU, 2 GB RAM, 64 GB NVMe), VPS Basic ₹1,105 (2 vCPU, 4 GB RAM, 128 GB NVMe), VPS Standard ₹2,211 (4 vCPU, 8 GB RAM, 256 GB NVMe), VPS Pro ₹3,869 (6 vCPU, 16 GB RAM, 512 GB NVMe) and VPS Enterprise ₹6,632 (8 vCPU, 32 GB RAM, 1 TB NVMe). VPS plans are non-refundable once provisioned.

Frequently asked questions

Ready for more room, or a server of your own? Compare VPS plans, read VPS vs shared hosting, or open a ticket to ask about an upgrade.