Introduction

In previous post Deploy Scalable and Reliable WordPress Site on LEMP(2), we successful set up Linux+Nginx+PHP+MySQL (LEMP) stack to hold WordPress site. However, Nginx and PHP services were enabled on the same server WEBo1.

In this lab, we will separate PHP to an external server PHP01 and leave WEB01 as Nginx web server only. It adds some flexibility scalability strategy. For performance enhancement details, please refer to Scaling PHP apps via dedicated PHP-FPM nodes, a test post I found online.

Deployment Steps

This lab involves PHP01 deployment and configuration change on WEB01 to forward PHP requests to PHP01. The topology is as below:

Step 1 – Configure PHP01 as php-fpm node

Boot another $5 ubuntu server from DigitalOcean, details available in Deploy Scalable and Reliable WordPress Site on LEMP(1).

#log onto PHP01
#update and install glusterfs client, php-fpm and php-mysql services on PHP01
sudo apt-get update
sudo apt-get -y install glusterfs-client php-fpm php-mysql
#make a folder called 'gluster' under root
sudo mkdir /gluster
#mount the 'file_store' volume on FS01 to '/gluster'on PHP01. '10.132.43.212' is FS01's private IP. 'glusterfs' is the filesystem type.
sudo mount -t glusterfs 10.132.43.212:/file_store /gluster

# add the partition to fstab so it mounts automatically at boot time.
sudo echo "10.132.43.212:/file_store /gluster glusterfs defaults,_netdev 0 0" >> /etc/fstab

Verify PHP01 can access the same WordPress folder we created on FS01 before, by executing the following command on PHP01.

sudo ls -la /gluster/www

Step 2 – WEB01 forward PHP requests to PHP01

Log onto WEB01, edit the nginx site default file by executing the following command:

sudo nano /etc/nginx/sites-enabled/default

Update the file as below; where ‘10.132.19.6’ is PHP01’s private IP, and 9000 is the port used by fastcgi. Comment the local fastcgi socket ‘unix:/run/php/php7.0-fpm.sock’.

	# pass the PHP scripts to FastCGI server listening on the php-fpm socket
        location ~ \.php$ {
                try_files $uri =404;
                #fastcgi_pass unix:/run/php/php7.0-fpm.sock;
                fastcgi_pass 10.132.19.6:9000;
                fastcgi_index index.php;
                fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name;
                include fastcgi_params;
                
        }

Restart Nginx service by executing the following command:

service nginx restart

Let’s now log onto the WordPress site, see whether it is still working as expected in Deploy Scalable and Reliable WordPress Site on LEMP(2).

Unfortunately, we get ‘502 Bad Gateway’ message this time.

Fortunately, Nginx provides us error log. Execute the following command on WEB01 to view the log.

cat /var/log/nginx/error.log

Nginx error log reveals the following:

2016/11/04 10:25:21 [error] 18867#18867: *1 connect() failed (111: Connection refused) while connecting to upstream, client:x.x.x.x, server:y.y.y.y, request: "GET / HTTP/1.1", upstream: "fastcgi://10.132.19.6:9000", host: "y.y.y.y"

OK…it appears WEB01 passed on the request to PHP01, but PHP01 refused the connection. Step 3 will help resolve the issue.

Step 3 – Allow PHP01 to Listen WEB01

On PHP01, edit PHP ‘www.conf’ file to allow listen on WEB01.

sudo nano /etc/php/7.0/fpm/pool.d/www.conf

Perform the following changes in ‘www.conf’.

#Add the following line to allow WEB01's private IP
listen.allowed_clients = 10.132.84.104

#Comment the following line by adding ";" in the front.
;listen = /run/php/php7.0-fpm.sock

#Add the following line to have php-fpm listen on port 9000
listen = 9000

Restart PHP on PHP01 and Nginx on WEB01.

#On PHP01 restart php service
service php7.0-fpm restart
#On WEB01 restart nginx service
service nginx restart

Let’s now access the WordPress site again. ‘Hello world!’ – it’s working!

The End