If the server CPU usage rate is too high, first check the processes in top or htop to see if any abnormal processes occupy high resources; secondly, use sar or mpstat to analyze historical trends; then review the application log to check code or query problems; then consider the impact of system level such as number of processes, swap memory, etc.; finally take measures such as restarting services, optimizing code or expanding capacity to solve them. 1. Run top/htop to identify high CPU processes; 2. Use sar/mpstat to analyze time trends; 3. Check log positioning application problems; 4. Troubleshoot system-level problems; 5. Perform repair operations.

If your server is showing high CPU usage, the first thing to do is identify what is using all that CPU power. High CPU usage doesn't always mean something's wrong — it could just be a spike due to normal workload. But if it's sustained or causing performance issues, you need to dig deeper.

Check Top Processes Using top or htop

Start by logging into the server and running top or htop (if installed). These tools show real-time CPU usage per process. Look for processes that are consistently near the top of the list.

• If you see something like php-fpm , mysql , or nginx taking up most of the CPU, that gives you a clue where to look.
• Sometimes a single rogue process or script might be looping or doing heavy computings — in that case, you can note its PID and investigate further with ps or strace .

Pro tip: Press P in top to sort by CPU usage directly.

Look at Resource Usage Over Time with sar or mpstat

Real-time tools are great, but they only show what's happening right now . To understand trends, use historical data tools like sar . It's part of the sysstat package on most Linux systems.

Run something like:

 sar -u 1 5

That shows CPU usage every second for five seconds. You can also check daily reports with just sar .

Another option is mpstat , which breaks down CPU usage across cores. This helps spot whether one core is maxed out while others sit idle.

Review Application Logs and Patterns

High CPU usage often points to an application-level issue. For example:

• A poorly optimized query in MySQL can cause repeated full table scans.
• A PHP script stuck in a loop or making too many external calls.
• A background job queue (like cron or a task runner) firing off too many workers.

Check logs in /var/log/ — especially syslog , messages , or service-specific logs like mysql/error.log or nginx/error.log . Look for repeated errors or slow queries.

If it's a web server, check access logs for spikes in traffic or unusual patterns like bots hitting search endpoints repeatedly.

Consider System-Level Issues

Sometimes the problem isn't an app but the system itself.

• Too many processes : Use ps aux | wc -l to count how many processes are running. An unusually high number may indicate a fork bomb or misconfigured services.
• Swapping : Run free -m and check if swap is being used. When memory runs low, the system swaps to disk, which can cause CPU to wait and appear busy.
• Kernel issues : Rare, but sometimes kernel threads ( kthreadd , kswapd0 ) show high usage. That usually means deeper problems like hardware issues or bugs.

Basic Fixes and Next Steps

Depending on what you find, here are some typical actions:

• Restart the problematic service ( systemctl restart mysql , for example).
• Kill a runaway process ( kill -9 PID if needed).
• Optimize slow-running code or queries.
• Add more resources (vertical scaling) or distributed load (horizontal scaling).
• Set up monitoring like netdata , Prometheus , or CloudWatch to catch future spikes early.

It's not always obvious why CPU usage jumps, but following these steps should help narrow things down quickly. Once you know what's driving the load, you can decide whether it's expected behavior or something to fix.

The above is the detailed content of How would you troubleshoot a server with high CPU usage?. For more information, please follow other related articles on the PHP Chinese website!