Reducing Server Load and Setting Up Swap Memory on Linux
Practical techniques to reduce CPU and RAM pressure on a Linux server, including creating persistent swap space to handle memory spikes without crashes.
Why Servers Run Out of Steam
Low-resource servers — VPS instances, Oracle Cloud Always Free, small Raspberry Pi nodes — share a common failure pattern: they run fine at idle, then crash or slow to a crawl the moment a process spikes memory or CPU. Two things cause this:
- RAM exhaustion — the kernel starts killing processes (OOM killer) when physical memory is full and there’s no fallback.
- Unnecessary CPU load — background processes, logging, and services eating cycles they don’t need to.
Both are fixable with straightforward configuration.
Part 1 — Swap Space
Swap is disk space the kernel uses as overflow when RAM fills up. It’s slower than RAM, but it prevents OOM kills and buys the system time to recover from spikes.
Check Current Swap
1
2
free -h
swapon --show
If the swap row is empty or shows 0, you have none configured.
Create a Swap File
1. Allocate the File
1
sudo fallocate -l 2G /swapfile
fallocate is instant — it reserves the space without writing zeros. Use 2G for servers with 1–2 GB RAM. Adjust as needed:
| RAM | Recommended Swap |
|---|---|
| 1 GB | 2 GB |
| 2 GB | 2–4 GB |
| 4 GB+ | Equal to RAM or less |
On some filesystems (like
btrfs),fallocatemay fail. Useddinstead:
2. Secure the File
1
sudo chmod 600 /swapfile
Only root should be able to read or write the swap file. This is a security requirement — world-readable swap can leak sensitive data from other processes.
3. Format and Enable
1
2
sudo mkswap /swapfile
sudo swapon /swapfile
Verify it’s active:
1
free -h
You should see swap space in the output.
Make It Persistent Across Reboots
Without this step, swap disappears after every reboot:
1
echo '/swapfile none swap sw 0 0' | sudo tee -a /etc/fstab
Verify the entry was added correctly:
1
tail -1 /etc/fstab
Expected output:
1
/swapfile none swap sw 0 0
Double-check
/etc/fstabafter editing. A malformed entry can prevent the server from booting. Runsudo mount -ato catch errors before rebooting.
Tune Swap Aggressiveness (swappiness)
swappiness controls how eagerly the kernel moves data to swap. The default is 60, which is tuned for desktops. On a server, a lower value keeps more data in RAM and only swaps under real pressure:
1
2
3
4
5
6
7
8
9
# Check current value
cat /proc/sys/vm/swappiness
# Apply immediately (lost on reboot)
sudo sysctl vm.swappiness=10
# Make it permanent
echo 'vm.swappiness=10' | sudo tee -a /etc/sysctl.conf
sudo sysctl -p
A value of 10 works well for most server workloads. Use 1 for databases that are sensitive to swap latency.
Part 2 — Reducing RAM Usage
Identify What’s Using Memory
1
2
# Sorted by memory usage, top 15 processes
ps aux --sort=-%mem | head -15
Or with htop — press F6 and sort by MEM%.
Disable Unused Services
List all running services:
1
systemctl list-units --type=service --state=running
Disable any service you don’t need:
1
2
3
sudo systemctl disable --now snapd
sudo systemctl disable --now ModemManager
sudo systemctl disable --now avahi-daemon
Common candidates on a minimal server:
| Service | Safe to disable if… |
|---|---|
snapd | You don’t use snap packages |
avahi-daemon | You don’t need mDNS/Bonjour |
ModemManager | No mobile broadband hardware |
bluetooth | No Bluetooth hardware |
cups | No printer attached |
Reduce Journald Log Retention
The system journal can grow large over time. Cap it:
1
2
sudo journalctl --vacuum-size=200M
sudo journalctl --vacuum-time=7d
Make the limits permanent:
1
sudo nano /etc/systemd/journald.conf
Set:
1
2
3
[Journal]
SystemMaxUse=200M
MaxRetentionSec=1week
Then restart the journal service:
1
sudo systemctl restart systemd-journald
Part 3 — Reducing CPU Load
Find CPU-Heavy Processes
1
2
3
4
5
# One-shot snapshot, sorted by CPU
ps aux --sort=-%cpu | head -15
# Live view
top
Adjust Process Priority with nice
Lower the priority of background tasks so they yield CPU to foreground workloads:
1
2
3
4
5
# Run a new process at low priority
nice -n 19 your-command
# Renice an already-running process (use its PID)
sudo renice -n 10 -p PID
nice values range from -20 (highest priority) to 19 (lowest). For background jobs like backups or compression, 10–19 is appropriate.
Limit a Service’s CPU with systemd
For any systemd service, you can cap its CPU share without touching the process itself:
1
sudo systemctl edit your-service
1
2
[Service]
CPUQuota=20%
This limits the service to 20% of one CPU core, regardless of load.
Part 4 — Monitor After Changes
After applying these changes, watch the server’s behaviour over a few minutes:
1
2
3
4
5
6
7
8
# Live memory and swap usage
watch -n 2 free -h
# Overall system load
vmstat 2 10
# Check if swap is actually being used
swapon --show
If swap usage stays near zero under normal load, your RAM headroom is healthy and swap is doing its job as a safety net — exactly as intended.
Quick Reference
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
# Create and enable 2G swap
sudo fallocate -l 2G /swapfile
sudo chmod 600 /swapfile
sudo mkswap /swapfile
sudo swapon /swapfile
echo '/swapfile none swap sw 0 0' | sudo tee -a /etc/fstab
# Set swappiness
echo 'vm.swappiness=10' | sudo tee -a /etc/sysctl.conf && sudo sysctl -p
# Check memory and swap
free -h && swapon --show
# Top memory consumers
ps aux --sort=-%mem | head -10
# Top CPU consumers
ps aux --sort=-%cpu | head -10
# Clean journal logs
sudo journalctl --vacuum-size=200M --vacuum-time=7d
Conclusion
Swap space is not a replacement for RAM — it’s a safety net that prevents crashes when memory spikes unexpectedly. Combined with disabling unused services, tuning swappiness, and capping noisy processes, these changes can significantly extend how long a low-resource server runs stably without intervention.