I seem to be missing a step in the deploy process. I get as far as deploy : WordPress Installed? and it fails. I’ve seen conflicting information in the forum regarding the installation of WP on the staging server. I can see I have everything but the web directory in src/www/staging.example.com
I’ve googled and searched as far as I can go. Any insight/tips on what I’m missing greatly appreciated.
What’s the error message?
Trellis doesn’t install WordPress on remote servers. But the deploy shouldn’t fail because of that.
The full message
TASK [deploy : WordPress Installed?] *******************************************
fatal: [XXX.XXX.XX.XXX]: FAILED! => {"changed": false, "cmd": ["wp", "core", "is-installed"], "delta": "0:00:00.268589", "end": "2016-02-01 17:29:50.279160", "failed": true, "failed_when_result": true, "rc": 1, "start": "2016-02-01 17:29:50.010571", "stderr": "WordPress database error Table 'chapel_dev.wp_blogs' doesn't exist for query SELECT * FROM wp_blogs WHERE domain = 'chapel.thedevyard.com' AND path = '/' made by include('phar:///usr/bin/wp/php/boot-phar.php'), include('phar:///usr/bin/wp/php/wp-cli.php'), WP_CLI\\Runner->start, WP_CLI\\Runner->load_wordpress, require('phar:///usr/bin/wp/php/wp-settings-cli.php'), require('wp-includes/ms-settings.php'), get_site_by_path\nWordPress database error Table 'chapel_dev.wp_sitemeta' doesn't exist for query SELECT meta_value FROM wp_sitemeta WHERE meta_key = 'site_name' AND site_id = 1 made by include('phar:///usr/bin/wp/php/boot-phar.php'), include('phar:///usr/bin/wp/php/wp-cli.php'), WP_CLI\\Runner->start, WP_CLI\\Runner->load_wordpress, require('phar:///usr/bin/wp/php/wp-settings-cli.php'), require('wp-includes/ms-settings.php'), WP_Network->__construct, WP_Network->_set_site_name, get_network_option\nWordPress database error Table 'chapel_dev.wp_sitemeta' doesn't exist for query SELECT meta_value FROM wp_sitemeta WHERE meta_key = 'siteurl' AND site_id = 1 made by include('phar:///usr/bin/wp/php/boot-phar.php'), include('phar:///usr/bin/wp/php/wp-cli.php'), WP_CLI\\Runner->start, WP_CLI\\Runner->load_wordpress, require('phar:///usr/bin/wp/php/wp-settings-cli.php'), wp_cookie_constants, get_site_option, get_network_option", "stdout": "", "stdout_lines": [], "warnings": []}
Aside from that, I’m confused about the “doesn’t install WordPress on remote servers.” Does that mean I need to create web directory and install there?
I had seen in the forums a suggestion to use
- if ! wp core is-installed; then wp core install --url="{{ project.env.wp_home }}" --title="{{project.site_title }}" --admin_user="{{ project.admin_user }}" --admin_password="{{ project.admin_password }}" --admin_email="{{ project.admin_email }}"; fi
in role/deploy/defaults/main.yml file, but that doesn’t seem to do anything.
EDIT Looking closer, I don’t have either the trellis or site folder. Do I need to install all of that on the server somehow? Must admit, the documentation on this part seems sparse.
I think you’re confusing the terms deploying and installing. Deploying a project basically means getting the code onto the server. Installing WordPress means creating a database and running their installer.
The workflow for a brand new server is:
server.yml
./deploy.sh
Now all that being said, you’re just running into an error which has nothing to do with this. That “WordPress Installed?” shouldn’t fail like that. I’m guessing there’s some mis-configuration here. Is this a WP multisite or no?
And no, trellis or site should not appear on the server. If you have repo_subtree_path set correctly then it just takes that sub-folder: https://github.com/roots/trellis/blob/11516d12e42355d0304eb4fb8c15be7ca8fcf6d4/group_vars/production/wordpress_sites.yml#L9
I guess I am. When I think of “deploy” from a git repo, I’m assuming all of the code is copied to the server, which in this case, it’s not?
To the question of whether this is a multi-site, yes. Everything works locally, but I will check the wordpress_sites.yml files to compare.
I appreciate your help in walking me through this.
Database is not code
that’s helpful 
Looks like you may have run into a bug with multisite (won’t be the first or last 
)
Here’s the failing task: https://github.com/roots/trellis/blob/11516d12e42355d0304eb4fb8c15be7ca8fcf6d4/roles/deploy/hooks/finalize-after.yml#L2-L8
Looks like multisite requires a --network option from WP-CLI. (See http://wp-cli.org/commands/core/is-installed/)
You can try manually adding it for now to see if it fixes the problem.
OK, I changed it to command: wp core is-installed --network in finalize-after.yml
Error:
fatal: [162.243.70.107]: FAILED! => {"changed": false, "cmd": ["wp", "core", "is-installed", "--network"], "delta": "0:00:00.242924", "end": "2016-02-01 19:22:40.956159", "failed": true, "failed_when_result": true, "rc": 1, "start": "2016-02-01 19:22:40.713235", "stderr": "Error: Can’t select database", "stdout": "", "stdout_lines": [], "warnings": []}
I think you also need the --url option for multisite:
–url= Pretend request came from given URL. In multisite, this argument is how the target site is specified.
So you can just try the first/main site URL.
So at this point, I’ve gotten past the previous issues, but now am hanging up with the domain_current_site: getting “site not found”.
I am using the example project as a template, using a subdomain, and have tried every variation of my staging.roots-example-project.com in the domain_current_site:. i.e., http://staging.roots-example-project.com, staging.roots-example-project.com, roots-example-project.com
Is this an issue with trying to do a multisite on a subdomain?
EDIT: full error message
TASK [deploy : Update WP theme paths] ******************************************
fatal: [XXX.XXX.XX.XXX]: FAILED! => {"changed": true, "cmd": ["wp", "eval", "wp_clean_themes_cache(); switch_theme(get_stylesheet());"], "delta": "0:00:00.097746", "end": "2016-02-01 21:16:01.744476", "failed": true, "rc": 1, "start": "2016-02-01 21:16:01.646730", "stderr": "Error: Site example.com/ not found.", "stdout": "", "stdout_lines": [], "warnings": []}
Just a follow up, in case some one has any insight.
I commented out that particular part of roles/deploy/hooks/finalize-after.yml but got the same error when trying trying to update the db in a multisite instance.
I ultimately went about testing if I could deploy a non-multisite instance without using a subdomain and was successful.
So somewhere along the line, I’m getting that Error: Site example.com/ not found when trying to update the database in multisite. I did see a suggestion to add the url to the wp-cli.yml file but not sure how that would affect local dev, or if it would do anything. I did try adding --url=example.com to wp core update-db --network to no avail.
Pretty stumped, but appreciate the responses thus far.
I run into the same bug on the production server with multisite/subdomain setting. Ansible fails at the task Wordpress installed?. This is strange, because the task works without any problems on the development virtual machine. So as already mentioned by @miklb it doesn’t fail because of a missing argument, like --network or --url.
I tried to run the command manually on the server with the --debug flag, but it failed:
root@example:/srv/www/example.com/current# wp core --debug is-installed --allow-root
Debug: No readable global config found (0.008s)
Debug: Using project config: /srv/www/example.com/releases/20160205091556/wp-cli.yml (0.008s)
Debug: ABSPATH defined: /srv/www/example.com/releases/20160205091556/web/wp/ (0.025s)
Debug: Begin WordPress load (0.026s)
Debug: wp-config.php path: /srv/www/example.com/releases/20160205091556/web/wp-config.php (0.026s)
Debug: Set URL: example.com/ (0.028s)
WordPress database error Table 'div_prod.wp_blogs' doesn't exist for query SELECT * FROM wp_blogs WHERE domain = 'example.de' AND path = '/' made by include('phar:///usr/bin/wp/php/boot-phar.php'), include('phar:///usr/bin/wp/php/wp-cli.php'), WP_CLI\Runner->start, WP_CLI\Runner->load_wordpress, require('phar:///usr/bin/wp/php/wp-settings-cli.php'), require('wp-includes/ms-settings.php'), get_site_by_path
PHP Notice: Undefined property: stdClass::$public in /srv/www/example.com/releases/20160205091556/web/wp/wp-includes/ms-settings.php on line 199
Notice: Undefined property: stdClass::$public in /srv/www/example.com/releases/20160205091556/web/wp/wp-includes/ms-settings.php on line 199
WordPress database error Table 'div_prod.wp_sitemeta' doesn't exist for query SELECT meta_value FROM wp_sitemeta WHERE meta_key = 'site_name' AND site_id = 1 made by include('phar:///usr/bin/wp/php/boot-phar.php'), include('phar:///usr/bin/wp/php/wp-cli.php'), WP_CLI\Runner->start, WP_CLI\Runner->load_wordpress, require('phar:///usr/bin/wp/php/wp-settings-cli.php'), require('wp-includes/ms-settings.php'), WP_Network->__construct, WP_Network->_set_site_name, get_network_option
WordPress database error Table 'div_prod.wp_sitemeta' doesn't exist for query SELECT meta_value FROM wp_sitemeta WHERE meta_key = 'siteurl' AND site_id = 1 made by include('phar:///usr/bin/wp/php/boot-phar.php'), include('phar:///usr/bin/wp/php/wp-cli.php'), WP_CLI\Runner->start, WP_CLI\Runner->load_wordpress, require('phar:///usr/bin/wp/php/wp-settings-cli.php'), wp_cookie_constants, get_site_option, get_network_option
Debug: Loaded WordPress (0.218s)
Debug: Running command: core is-installed (0.219s)
I checked for any known bugs on wp-cli, but haven’t found any reports. I also updated wp-cli to the latest version without any luck. @swalkinshaw do you have any suggestions where to start debugging?
Edit:
I added a bug report with a temporary workaround https://github.com/roots/trellis/issues/482.
For more context from my end, the exact same local installation deploys just fine if I do not enable multisite.
I was able to resolve this by looking for instances of my old url in the database sql dump and changing it with my new desired production url.