If you have set up a LEMP (Linux, NGINX, MySQL/MariaDB, and PHP) stack, you are likely utilizing FastCGI proxying in NGINX (acting as an HTTP server) for handling PHP scripts. PHP-FPM, which stands for FastCGI Process Manager, is a popular and high-performance implementation of PHP FastCGI.

Here are some helpful tutorials for installing a LEMP Stack on Linux systems.

• How to Install LEMP Stack with PhpMyAdmin in Ubuntu 20.04
• How to Install LEMP Server on CentOS 8
• How to Install LEMP on Debian 10 Server

Not long ago, all of our PHP websites hosted on one of our LEMP servers started becoming sluggish and ultimately stopped responding when attempting to log into the server. Upon investigation, we found that the system was critically low on RAM: PHP-FPM had used up a large portion of the available memory, as shown in the screenshot below (using glances – a system monitoring utility).

<code>$ glances</code>

In this guide, we will explain how to stop PHP-FPM from using excessive or all of your system memory (RAM) on Linux. By the end of this article, you’ll know how to cut PHP-FPM memory usage by 50% or more.

Minimize PHP-FPM Memory Consumption

After some online research, we found that adjusting the PHP-FPM process manager settings and related parameters in the pool configuration file would help reduce memory usage.

The default pool is named www, and its configuration file can be found at /etc/php-fpm.d/www.conf (on CentOS/RHEL/Fedora) or /etc/php/7.4/fpm/pool.d/www.conf (on Ubuntu/Debian/Mint).

<code>$ sudo vim /etc/php-fpm.d/www.conf             [On CentOS/RHEL/Fedora]
$ sudo vim /etc/php/7.4/fpm/pool.d/www.conf    [On Ubuntu/Debian/Mint]</code>

Locate the following directives and adjust their values based on your needs. For any that are commented out, remove the comment symbol to enable them.

<code>pm = ondemand
pm.max_children = 80
pm.process_idle_timeout = 10s
pm.max_requests = 200</code>

Here’s a quick explanation of each of these directives and their values. The pm directive controls how the process manager manages child processes. The default is dynamic, which means the number of child processes changes dynamically based on other settings, such as pm.max_children, which sets the maximum number of simultaneous child processes.

The best option in this scenario is the ondemand method. With this setting, no child processes are created at startup; instead, they are launched only when needed. Processes are created only when incoming requests require them, based on pm.max_children and pm.process_idle_timeout, which specifies how many seconds an idle process should remain active before being terminated.

Finally, the pm.max_requests directive should be set as well. This defines the number of requests each child process should handle before it is restarted. It’s also useful for mitigating memory leaks in third-party extensions.

Reference: A better way to run PHP-FPM.

After applying the above changes, I observed that RAM usage returned to normal on our server. Do you have any thoughts or questions regarding this topic? Feel free to reach out via the feedback form below.

The above is the detailed content of How To Prevent PHP-FPM From Consuming Too Much RAM in Linux. For more information, please follow other related articles on the PHP Chinese website!