Search for a command to run...
As a passionate DevOps Engineer with 3+ years of experience, I specialize in building robust, scalable, and secure infrastructures. My expertise spans Kubernetes, Jenkins, Docker, AWS, Ansible, Flask, Apache, Nginx, Kibana, Uyuni, Percona PMM, MySQL, and more.
No comments yet. Be the first to comment.
📋 Prerequisites Install base tools and keys: sudo apt update && sudo apt install -y \ curl gnupg2 ca-certificates lsb-release debian-archive-keyring Add NGINX.org signing key: curl https://nginx.org/keys/nginx_signing.key | gpg --dearmor | sudo t...
This article explains how to install and manage Uyuni on openSUSE Leap Micro 5.5 and above. The purpose of setting up Uyuni is primarily for patch management, configuration management, and integrating Ansible control. Uyuni serves as a centralized ma...
In this guide, we'll learn how to export Kibana detection rules using the Kibana Saved Objects API and store them in an S3 bucket. This approach ensures you have backups of your Kibana rules and can restore them whenever needed. Why Do You Need This?...
In today's world, it's crucial to monitor your application logs and get alerted about suspicious or critical activities. If you are using Kibana for log monitoring and want to get real-time alerts on Microsoft Teams or Slack, this guide is for you. T...
I know we have been in a situation where the PHP application goes down even if there is no CPU and memory usage on both the application and database on the server.
This document explains the common bottleneck and blockage issues that occur when using PHP-FPM with Apache and Nginx, especially in environments involving MySQL databases. We will also explore effective solutions using multiple PHP-FPM sockets, Apache load balancer methods, and Nginx upstream configurations.
PHP-FPM (FastCGI Process Manager) manages incoming PHP requests by using worker processes. When all available workers are occupied, incoming requests get queued, leading to performance degradation.
By default, Apache uses mod_php a single PHP-FPM socket for handling PHP requests.
This can cause bottlenecks if all requests are directed to a single socket.
Nginx uses a single fastcgi_pass directive to forward requests to PHP-FPM.
If the server block configuration is limited to a single socket, the request handling becomes inefficient.
PHP-FPM bottlenecks can be compounded by MySQL-related issues, especially when:
Table Locks: When tables are locked for writing, other queries trying to access the table are blocked until the lock is released.
Connection Pile-Up: If PHP requests wait on long-running MySQL queries, PHP-FPM workers are held up, causing a backlog of requests.
Insufficient Database Connections: The MySQL server might have too few connections allowed (max_connections), causing PHP requests to hang.
To mitigate bottlenecks, we create multiple PHP-FPM pool sockets and configure Apache to use them as a load balancer.
www.conf . You just need to modify the listen directive and [site]./etc/php/8.1/fpm/pool.d/www1.conf)[site1] # needs to be changed
user = www-data
group = www-data
listen = /run/php/php8.1-fpm1.sock # needs to be changed
listen.owner = www-data
listen.group = www-data
listen.mode = 0660
pm = dynamic
pm.max_children = 50
pm.start_servers = 5
pm.min_spare_servers = 5
pm.max_spare_servers = 10
same as above, create 2-3 pools as per you needs/etc/php/8.1/fpm/pool.d/www2.conf
[site2] # needs to be changed
user = www-data
group = www-data
listen = /run/php/php8.1-fpm2.sock # needs to be changed
listen.owner = www-data
listen.group = www-data
listen.mode = 0660
pm = dynamic
pm.max_children = 50
pm.start_servers = 5
pm.min_spare_servers = 5
pm.max_spare_servers = 10
save it and restart the php-fpm service
systemctl restart php-fpmsudo apt install php-pfm/etc/apache2/sites-available/your-site.conf)<VirtualHost *:80>
ServerName your-site.com
<FilesMatch ".+\.php$">
SetHandler "proxy:unix:/run/php/php8.1-fpm2.sock|fcgi://localhost"
</FilesMatch>
ProxyPassMatch ^/(.*\.php(/.*)?)$ unix:/run/php/php8.1-fpm1.sock|fcgi://localhost
</VirtualHost>
you need to enable below module before enabling load balancing in apache
sudo a2enmod proxy
sudo a2enmod proxy_fcgi
sudo a2enmod proxy_balancer
sudo a2enmod lbmethod_byrequests
To distribute PHP-FPM requests evenly:
<Proxy balancer://myphpfpmcluster>
BalancerMember unix:/run/php/php8.1-fpm2.sock|fcgi://localhost
BalancerMember unix:/run/php/php8.1-fpm1.sock|fcgi://localhost
ProxySet lbmethod=byrequests
</Proxy>
<VirtualHost *:80>
ServerName your-site.com
ProxyPassMatch ^/(.*\.php(/.*)?)$ balancer://myphpfpmcluster
</VirtualHost>
For better load distribution with Nginx, configure upstream with least_conn method:
upstream php_backend {
least_conn;
server unix:/run/php//run/php/php8.1-fpm1.sock;
server unix:/run/php//run/php/php8.1-fpm2.sock;
}
server {
listen 80;
server_name your-site.com;
location ~ \.php$ {
include fastcgi_params;
fastcgi_pass php_backend;
## add other php-fpm or proxy congigs here
}
}
once the configurations are complete. Please restart Apache/nginx. and visit the website.
Apache Load Balancer (mod_proxy_balancer):
Uses BalancerMember to distribute traffic among PHP-FPM sockets.
lbmethod=byrequests ensures a round-robin distribution of requests.
Nginx Upstream (Least Connections):
least_conn; ensures requests are routed to the least busy backend socket.PHP-FPM Pool Configuration:
pm = dynamic allows the number of PHP-FPM workers to scale based on demand.Improving MySQL configurations can also reduce bottlenecks:
wait_timeout: Controls the timeout for connections waiting for activity. Increase or decrease based on application needs.
interactive_timeout: Similar to wait_timeout but applied for interactive clients. Adjust for long-running connections.
max_execution_time: Defines the maximum execution time of SELECT statements. This can prevent slow queries from hanging PHP-FPM workers.
Example Configuration (/etc/mysql/my.cnf):
[mysqld]
wait_timeout = 120
interactive_timeout = 60
max_execution_time = 3000
max_connections = 200
By implementing multiple PHP-FPM pool sockets and using load balancing mechanisms in Apache and Nginx, we effectively mitigate PHP-FPM bottlenecks. Additionally, addressing MySQL-related issues like table locks and connection pile-up can further enhance performance.