If your VPS logs show “server reached pm.max_children setting”, this tutorial will show you how to calculate the correct value safely using actual memory usage from your server.
Table of Contents
Find the PHP-FPM Pool Used by the Website
Before tuning anything, you must identify which PHP-FPM pool the site is using.
The easiest way is to inspect the Nginx configuration for the site.
Run:
nginx -T | grep -n "example.com"
Example output:
configuration file /etc/nginx/sites-available/example.com.conf:
fastcgi_pass unix:/var/run/example.com.sock;
This shows that the site is using the PHP-FPM socket:
/var/run/example.com.sock
The pool configuration will typically be located at:
/etc/php/8.3/fpm/pool.d/example.com.conf
Check Available Server Memory
Next, determine how much memory is available on the server.
Run:
free -m
Example output:
Mem: 32094 total
This VPS has about 32 GB of RAM available. However, you should never allocate all of it to PHP workers.
Other services also require memory:
| Service | Typical Memory Usage |
|---|---|
| Operating system + cache | 2-4 GB |
| MySQL / MariaDB | 4-6 GB |
| Nginx | <1 GB |
| Other PHP pools | ~1 GB |
So a safe PHP-FPM memory budget for this pool might be:
3 GB – 4 GB
Measure Memory Usage of PHP Workers
Next, determine how much memory each PHP worker consumes.
Run:
ps -eo pid,user,rss,args | grep "php-fpm: pool example.com" | grep -v grep
Example output:
The RSS column shows memory usage in kilobytes.
Convert to megabytes:
185000 KB ≈ 180 MB
So each worker consumes approximately:
~180 MB
Calculate the Correct pm.max_children
Now we can calculate the safe pool size.
Formula:
max_children = memory_budget / memory_per_worker
Example:
Memory budget = 3 GB = 3072 MB
Worker memory = 180 MB
Calculation:
3072 / 180 ≈ 17
Safe configuration:
pm.max_children = 16
This allows up to 16 concurrent PHP workers without exhausting server memory.
Update PHP-FPM Pool Configuration
Edit the pool configuration file:
nano /etc/php/8.3/fpm/pool.d/example.com.conf
Example configuration:
pm = dynamic
pm.max_children = 16
pm.start_servers = 4
pm.min_spare_servers = 3
pm.max_spare_servers = 8
pm.max_requests = 500
Restart PHP-FPM:
php-fpm8.3 -t
systemctl restart php8.3-fpm
Monitor the Pool After Changes
After adjusting the pool, monitor the logs to ensure the limit is no longer reached.
Check PHP-FPM logs:
grep -i max_children /var/log/php8.3-fpm.log
Example warning before tuning:
If these warnings disappear, the pool size is now correctly tuned.
You should also monitor Nginx error logs for upstream timeouts:
grep "upstream timed out" /var/log/nginx/error.log
Important Notes
Increasing pm.max_children does not fix slow application code.
If PHP requests are slow (heavy Laravel queries, inefficient database calls, etc.), workers remain busy longer and the pool will still saturate.
Proper tuning simply ensures the server has enough workers to handle traffic bursts safely.
Summary
Instead of guessing pm.max_children, calculate it using real data:
Identify the PHP-FPM pool used by the site
Check available server memory
Measure memory usage of PHP workers
Allocate a safe PHP memory budget
Calculate the pool size using the formula
This method prevents both:
server crashes caused by memory exhaustion
