How to diagnose and fix VPS performance issues. Covers CPU, memory, disk, and network troubleshooting with practical command-line tools and solutions.
Your VPS feels slow. Here's how to find out why and fix it.
The Diagnostic Flow
| Symptom | First Check | Tool |
|---|
| Everything slow | CPU and memory | htop |
| Website slow | Disk I/O | iotop |
| Connection drops | Network | ss, nethogs |
| Out of memory | Memory usage | free -h |
| Disk full | Disk space | df -h |
htop: Your Primary Tool
sudo apt install htop
htop
Reading htop
| Section | What to Look For |
|---|
| CPU bars | All cores near 100% = CPU bottleneck |
| Memory bar | Over 90% = memory pressure |
| Swap bar | Any usage = memory shortage |
| Load average | Above CPU count = overloaded |
| Process list | Which process uses the most CPU/RAM |
Load Average Guide
| Load Average vs CPU Cores | Status |
|---|
| Load < cores | Healthy |
| Load = cores | Fully used |
| Load > cores | Overloaded |
| Load > 2x cores | Severely overloaded |
Example: 2-core VPS with load 1.5 is busy but healthy. Load 4.0 is overloaded.
CPU Troubleshooting
Finding the Culprit
# Top CPU processes
ps aux --sort=-%cpu | head -10
Common CPU Hogs
| Process | Cause | Solution |
|---|
| mysql/mariadb | Unoptimized queries | Add indexes, optimize queries |
| php-fpm | Heavy PHP processing | Enable OPcache, add caching |
| node | Unoptimized code or memory leak | Profile and fix code |
| apache2 | Too many concurrent connections | Tune MaxClients |
| python | Data processing task | Optimize or schedule off-peak |
Memory Troubleshooting
# Memory overview
free -h
# Per-process memory
ps aux --sort=-%mem | head -10
When Memory is Low
| Action | Impact |
|---|
| Restart PHP-FPM | Clears memory leaks |
| Reduce MySQL buffer pool | Frees memory |
| Kill unused processes | Immediate relief |
| Add swap | Temporary fix |
| Upgrade VPS | Permanent fix |
MySQL Memory Optimization
MySQL often uses the most memory:
# /etc/mysql/mariadb.conf.d/50-server.cnf
innodb_buffer_pool_size = 256M # Reduce if too high
key_buffer_size = 32M
max_connections = 50 # Reduce from default 151
Disk I/O Troubleshooting
# Install iotop
sudo apt install iotop
sudo iotop
Disk Space
# Filesystem usage
df -h
# Find large directories
du -sh /var/* | sort -rh | head -10
# Find large files
find / -type f -size +100M -exec ls -lh {} \; 2>/dev/null
Common Disk Space Fixes
| Culprit | Location | Solution |
|---|
| Log files | /var/log/ | Configure logrotate, clean old logs |
| Old backups | /var/backups/ | Delete old backups |
| Package cache | /var/cache/apt/ | sudo apt clean |
| Old kernels | /boot/ | sudo apt autoremove |
Network Troubleshooting
# Active connections
ss -tuln
# Bandwidth per process
sudo apt install nethogs
sudo nethogs
# Network statistics
sudo apt install vnstat
vnstat
Quick Diagnostic Script
#!/bin/bash
echo "=== System Overview ==="
uptime
echo ""
echo "=== Memory ==="
free -h
echo ""
echo "=== Disk ==="
df -h /
echo ""
echo "=== Top CPU Processes ==="
ps aux --sort=-%cpu | head -5
echo ""
echo "=== Top Memory Processes ==="
ps aux --sort=-%mem | head -5
echo ""
echo "=== Network Connections ==="
ss -tuln | grep LISTEN
Save as diagnose.sh and run when something feels off.
When to Upgrade
| Sign | Current Plan Issue |
|---|
| Consistent load > cores | Need more CPU |
| OOM kills in logs | Need more RAM |
| Disk constantly > 80% | Need more storage |
| Bandwidth limits hit | Need more bandwidth |
Space-Node's VPS plans scale from small to large. Start with what you need and upgrade when monitoring tells you it's time, not before.
