Trellis assumes your WordPress configuration already has multisite set up. If not, ensure the following values are placed somewhere before calling
Config::apply() in Bedrock's
config/application.php and before provisioning your server:
/* Multisite */ Config::define('WP_ALLOW_MULTISITE', true); Config::define('MULTISITE', true); Config::define('SUBDOMAIN_INSTALL', false); // Set to true if using subdomains Config::define('DOMAIN_CURRENT_SITE', env('DOMAIN_CURRENT_SITE')); Config::define('PATH_CURRENT_SITE', env('PATH_CURRENT_SITE') ?: '/'); Config::define('SITE_ID_CURRENT_SITE', env('SITE_ID_CURRENT_SITE') ?: 1); Config::define('BLOG_ID_CURRENT_SITE', env('BLOG_ID_CURRENT_SITE') ?: 1);
You'll also need to update the multisite settings under your environment directory (
multisite: enabled: true subdomains: false # Set to true if you're using a subdomain multisite install
You may also want to define the
env dictionary for more multisite specific settings such as
env: domain_current_site: store1.example.com
env will be merged in with Trellis' defaults so you don't need to worry about re-defining all of the properties.
Here's an example of a complete entry set up for multisite:
# group_vars/production/wordpress_sites.yml wordpress_sites: example.com: site_hosts: - canonical: example.com local_path: ../site # path targeting local Bedrock site directory (relative to Ansible root) admin_email: firstname.lastname@example.org multisite: enabled: true subdomains: true ssl: enabled: false cache: enabled: false env: domain_current_site: store1.example.com
After provisioning your remote server and deploying your sites, you'll need to install WordPress as a final step in your staging and production environments. SSH into your server as the
web user with
ssh web@<domain> and in the
/srv/www/<domain>/current/ directories run the following WP-CLI command to install WordPress:
$ wp core multisite-install --title="site title" --admin_user="username" --admin_password="password" --admin_email="email@example.com"
You may notice that your network's main site URLs contain
/wp/ before the post's or page's pathnames. This is a problem in WP core which occurs when WordPress is located in a subdirectory, as is the case with Bedrock. See issue Bedrock issue #250 for details, along with the site URL fix plugin in the Multisite Fixes plugin collection for a solution.
If you use Let's Encrypt as your SSL provider and your multisite install uses subdomains, currently you have to generate individual certificates for each of your subdomains, but this may change soon as Let's Encrypt will begin issuing wildcard certificates in January of 2018. You can generate SSL certificates for your subdomains if you know these subdomains in advance while provisioning your server. To do this, define multiple
canonical entries under
site_hosts in your corresponding
wordpress_sites.yml file like this:
site_hosts: - canonical: example.com redirects: - www.example.com - canonical: subdomain.example.com redirects: - www.subdomain.example.com
# Subdomains locally
For subdomains in development, you'll need DNS entries for every subdomain/host. The Landrush Vagrant plugin is how you can do this. Install it via:
$ vagrant plugin install landrush
Landrush spins up a small DNS server that allows us to use wildcard subdomains, a requirement for subdomain multisite installs.
Some users may have external DNS issues when using Landrush. If you encounter this, add this to your
config.landrush.guest_redirect_dns = false
See issue #511 for more details.
# Debugging Landrush
If something goes wrong with Landrush such as not being able to resolve a website from the guest:
$ vagrant landrush list
And remove any extraneous entries and try again.