PHP Auto-Configuration

Auto-Configurations for PHP Application Servers

The platform automatically adjusts selected PHP application-server directives when the allocated RAM and CPU resources change. This helps the server configuration follow the current Cloudlet limit and use available capacity more efficiently.

Automatic optimisation can be disabled when custom values are required. Manual settings should be applied carefully because unsuitable values can reduce performance or make the server unstable.

On this page

Auto-Configuration Example

For example, consider an Apache PHP application server with a limit of 16 Cloudlets. Its prefork-module settings are calculated according to the resources available to the container.

Apache PHP auto-configuration with 16 Cloudlets
Apache PHP prefork settings with a 16-Cloudlet resource limit.

After increasing the server limit to 32 Cloudlets and refreshing /etc/httpd/conf/httpd.conf, the related directives are recalculated automatically.

Apache PHP auto-configuration with 32 Cloudlets
The directive values increase automatically after the Cloudlet limit is raised.
The same resource-aware approach is used by the other supported PHP application-server stacks.

Disable Automatic Optimisation

Automatic optimisation must be disabled before manually changing auto-managed directives. Otherwise, custom values can be replaced during a container restart or update.

Useful variables: PHP_MEMORY_LIMIT and PHP_MAX_EXECUTION_TIME can adjust the PHP memory limit and maximum execution time without disabling automatic optimisation.

Smart Auto-Configuration

For LiteSpeed-based stacks, including LiteSpeed Web Server and LLSMP, set the JELASTIC_AUTOCONFIG environment variable to false or a disabled value.

Disable PHP Smart Auto-Configuration
Disable Smart Auto-Configuration with the JELASTIC_AUTOCONFIG variable.

Legacy Implementation

Apache PHP and NGINX PHP use an older optimisation mechanism. Remove the appropriate marker from the relevant configuration file before applying custom values:

  • Apache PHP: remove # Jelastic autoconfiguration mark from /etc/httpd/conf/httpd.conf.
  • NGINX PHP: remove ; Jelastic autoconfiguration mark from /etc/php-fpm.conf.
Apache PHP auto-configuration mark
Remove the Apache PHP optimisation marker before applying manual settings.
NGINX PHP auto-configuration mark
Remove the NGINX PHP optimisation marker from php-fpm.conf.

Save the modified configuration file after removing the marker.

Configuration warning: Manual directive values should be used only when their effect is fully understood. Incorrect settings can make the PHP application server unstable.

Apache PHP Settings

Apache PHP handles requests with the prefork Multi-Processing Module. The related directives are stored in /etc/httpd/conf/httpd.conf.

Apache PHP httpd.conf file
Apache PHP request-processing settings are stored in httpd.conf.

The platform manages these prefork directives:

StartServersNumber of child server processes created during startup.
MinSpareServers / MaxSpareServersMinimum and maximum number of idle processes kept ready.
ServerLimitMaximum permitted value for MaxRequestWorkers.
MaxRequestWorkersMaximum number of simultaneous requests.
MaxConnectionsPerChildNumber of requests a child process can serve before it is replaced.
Apache PHP prefork-module configuration
Apache PHP prefork-module directives managed by the platform.

ServerLimit and MaxRequestWorkers are calculated using the lower value from RAM-based and CPU-based limits:

MaxRequestWorkers = ServerLimit =
min({containerRam} / 30 MiB, {coresCount} × 5)
  • The RAM-based limit divides the container memory by 30 MiB.
  • The CPU-based limit multiplies the available physical-core count by five.
  • The lower result prevents the worker limit from exceeding available hardware capacity.
A very high ServerLimit can reserve unnecessary shared memory, while a value that is too low can restrict server performance. MaxRequestWorkers must balance expected concurrent traffic against available RAM.

NGINX PHP Settings

NGINX PHP uses PHP FastCGI Process Manager. Its main settings are stored in /etc/php-fpm.conf, which is normally available in the file manager’s Favorites list.

NGINX PHP php-fpm.conf file
Open php-fpm.conf to review NGINX PHP process-manager settings.

The default process-management mode is ondemand. New worker processes are created when required and idle processes are removed.

  • pm.max_children — maximum number of child processes. It follows the CPU capacity available to the container and is not set below two.
  • pm.process_idle_timeout — time before an idle process is terminated; the documented default is 60 seconds.
NGINX PHP FastCGI Process Manager settings
NGINX PHP uses ondemand FPM process management.
The PHPFPM_MAX_CHILDREN environment variable can define pm.max_children explicitly without disabling the complete optimisation mechanism.

PHP-FPM communication uses a UNIX-domain socket instead of a TCP connection, reducing routing and networking overhead for processes running on the same host.

LiteSpeed and LLSMP

LiteSpeed stacks automatically optimise the worker-process count according to allocated RAM and CPU cores. The JELASTIC_AUTOCONFIG environment variable controls whether this optimisation is enabled.

  • true — enables auto-configuration and is the default.
  • false — disables auto-configuration for manual tuning.

What’s next?