'dict object' has no attribute 'wordpress_sites'

[grahamharper]

First, thanks for creating this project, I've been looking for something like this.

I followed the installation instructions and ran ./bin/database.sh production <my_site_name> pull but I'm getting this error:

TASK [Search for {{ url_from }} and replace with {{ url_to }} on development] *****************************************************************************************************
System info:
  Ansible 2.3.0.0; Darwin
  Trellis at "Add Vagrant config"
---------------------------------------------------
the field 'args' has an invalid value, which appears to include a variable
that is undefined. The error was: {{
from_host.wordpress_sites[site].site_hosts.0.canonical }}: 'dict object' has
no attribute 'wordpress_sites'

I'm on macOS 10.12.3 and using Trellis 1.0.0-rc1.

[grahamharper]

My apologies, never mind, I misunderstood the instructions about the hosts configuration.

[BrunoFenzl]

Hi @grahamharper could you tell what did you misunderstood? I'm having the same problem and can't get it working…

[grahamharper]

@BrunoFenzl if I remember correctly, I think I appended the example hosts config to my existing hosts config, but what I needed to do was replace my existing hosts config with the example config (and of course modify your_server_hostname).

[BrunoFenzl]

@grahamharper Thanks for the quick reply! Well that's what I have… my staging host file looks like:

staging_host ansible_host=SERVER_IP

[staging]
staging_host

[web]
staging_host

and development host file:

# This file is only used for Windows hosts.
#
# Windows
# -------------------------------------------------------------
# If you want to run `dev.yml` manually you can SSH into the VM
# to the directory with the `dev.yml` playbook and run:

# `ansible-playbook dev.yml`
#
# Non-Windows
# -------------------------------------------------------------
# If you want to run `dev.yml` manually via the `ansible-playbook`
# command (vs. `vagrant up` or `vagrant provision`), you might be
# inclined to define your development host information in this file.
# We recommend instead that you use the `-i` (inventory) option with
# your `ansible-playbook` command to specify the custom inventory file
# Vagrant has created for the VM. Vagrant's custom inventory
# includes necessary non-standard SSH connection information.
#
# Here is an example command:
#
# `ansible-playbook dev.yml -i .vagrant/provisioners/ansible/inventory/vagrant_ansible_inventory`
#
# The `.vagrant` directory above is usually in the same directory as
# your `Vagrantfile`. If not, you will need to adjust this path in the
# command.
#
# Why run `dev.yml` manually?
# -------------------------------------------------------------
# One reason you may want to run `dev.yml` via the `ansible-playbook`
# command is for the convenience of adding Ansible options via the
# command line (e.g., `--tags`, `--skip-tags`, or `-vvvv`). In contrast,
# the commands `vagrant up` and `vagrant provision` would only run the
# `dev.yml` playbook with such options if you were edit the options
# into the Vagrantfile's `config.vm.provision` section.

development_host ansible_host=192.168.50.5 ansible_connection=ssh ansible_user=vagrant ansible_ssh_private_key_file=.vagrant/machines/default/virtualbox/private_key ansible_ssh_extra_args="-o StrictHostKeyChecking=no -o GlobalKnownHostsFile=/dev/null -o UserKnownHostsFile=/dev/null"

[development]
192.168.50.5 ansible_connection=local

[web]
192.168.50.5 ansible_connection=local

Just in case, I'm running trellis 1.0.0-rc.2, Ansible 2.5.3 and macOS 10.13.4. Any hint?

[BrunoFenzl]

@grahamharper is the opposite of what you wrote. I had to append the configurations to my host files. Now everything works fine. Thanks again for the input!

Now my staging host file looks like:

# Add each host to the [staging] group and to a "type" group such as [web] or [db].
# List each machine only once per [group], even if it will host multiple sites.

[staging]
IP_ADDRESS

[web]
IP_ADDRESS

staging_host ansible_host=IP_ADDRESS

[staging]
staging_host

[web]
staging_host

@valentinocossar Is it supposed to be this way? Thanks for the tool btw!

[grahamharper]

Ok, maybe that was it, sorry it's been so long.

[valentinocossar]

Hi, the right way is to replace the content of each host file with those available in the README. But I've never tested the tool with Ansible 2.5.3. Maybe with Ansible >= 2.5.0 I'll have to update the tool to improve compatibility.

Could be a good idea adding somewhere in the README a notice like "Tested up to Ansible 2.X.X".

[eriteric]

ansible 2.4.3.0 While trying to database push I get this error with both the append and replace method of editing the hosts file.

TASK [Search for {{ url_from }} and replace with {{ url_to }} on development] ***********************
System info:
  Ansible 2.4.3.0; Darwin
  Trellis version (per changelog): "Enable `fastcgi_cache_background_update` by default"
---------------------------------------------------
The task includes an option with an undefined variable. The error was: {{
hostvars.development_host.wordpress_sites[site].site_hosts.0.canonical }}:
'dict object' has no attribute 'wordpress_sites'

The error appears to have been in
'/myfolder/database-push.yml': line 89,
column 7, but may
be elsewhere in the file depending on the exact syntax problem.

The offending line appears to be:

    - name: Search for {{ url_from }} and replace with {{ url_to }} on
development
      ^ here
We could be wrong, but this one looks like it might be an issue with
missing quotes.  Always quote template expression brackets when they
start a value. For instance:

    with_items:
      - {{ foo }}

Should be written as:

    with_items:
      - "{{ foo }}"

exception type: <class 'ansible.errors.AnsibleUndefinedVariable'>
exception: {{
hostvars.development_host.wordpress_sites[site].site_hosts.0.canonical }}:
'dict object' has no attribute 'wordpress_sites'
fatal: [production_host]: FAILED! => {}
    to retry, use: --limit @/myfolder/database-push.retry

PLAY RECAP ******************************************************************************************
production_host            : ok=11   changed=8    unreachable=0    failed=1

[valentinocossar]

Hi, Trellis version?

[eriteric]

@valentinocossar using the latest Trellis version from the master branch. When i commented that was including the June 3rd commit in Trellis (commit # e140c05e18db87bce01d92bd1433c17e415adac3). There's been one more commit since then

Followed the instructions and recommendations in this repo's readme. Really not sure why it's not working. Uploads works fine though.

Using OSX

[valentinocossar]

I've just tested it out with Ansible 2.4.3.0 and the latest Trellis version (from branch master) and works as expected. Could you show me the hosts files (with obviously a fake IP) and the wordpress_sites files (without sensitive information)? Thanks!

[eriteric]

Thanks for testing, Valentinocaossar. Here is a gist of the files (with redacted sensitive info) from my server which produces the error: https://gist.github.com/eriteric/85f2820892edb9024984916ae4ae2063

---

I've tried this with a fresh copy of both trellis and this repo, and it worked too.

I then added in my customizations one by one, provisioned both dev and prod and then attempted to run ./bin/database.sh at each step.

1. add wp-cli package - aaemnnosttv/wp-cli-login-command - ./bin/database.sh push and pull worked
2. add another website under wordpress_sites - ./bin/database.sh push and pull worked for all websites
3. Add SSL to dev - ./bin/database.sh push and pull worked for all websites
4. Ansible vault encrypt all the vault.yml files - ./bin/database.sh push and pull worked for all websites
5. Enable server cache + set skip_cache_uri & skip_cache_cookie: ./bin/database.sh push and pull worked for all websites

The trellis that produces the error has 4 websites as you'll see in the gist. Aside from that I only know of one more variable to try, and that's adding the Typisttech/trellis-cloudflare-origin-ca trellis add-on & turn on SSL on production. However I can't try that right now on the domain I'm testing with as it's in use, but i will be able to in the next few days.

I'm hoping there is something I'm just not seeing in my wordpress_sites files.

edit: Thought of another variable... I did provision the working server once before making the changes for trellis-database-uploads-migration. On the server that gives the error I provisioned with trellis-database-uploads-migration added from the start.

[eriteric]

Ok I started another server. And DB sync was working up until the point I added the cloudflare origin certificate trellis add-on. Any ideas as to why that would interfere?

[henscu]

Any progress on this? I am using Ansible 2.5.3, the Cloudflare Origin Certificate Trellis addon, and get an error...

My wordpress_sites file:

  example.com:
    site_hosts:
      - canonical: example.com
        redirects:
          - www.example.com
    local_path: ../bedrock # path targeting local Bedrock site directory (relative to Ansible root)
    repo: git@bitbucket.org:xxxxxx/example.com.git # replace with your Git repo URL
#    repo_subtree_path: '.' # relative path to your Bedrock/WP directory in your repo
    branch: master
    multisite:
      enabled: false
    ssl:
      enabled: true
      stapling_enabled: false
      provider: cloudflare-origin-ca
    cache:
      enabled: true

I issue the command:

    `./bin/database.sh   production   example.com  backup`

I get the following error:

The task includes an option with an undefined variable. The error was:
'unicode object' has no attribute 'wordpress_sites'

The error appears to have been in '/../../../trellis/database-backup.yml': line 48, column 7, but
may
be elsewhere in the file depending on the exact syntax problem.

The offending line appears to be:

    - name: Pull exported database from {{ env }} to development
      ^ here
We could be wrong, but this one looks like it might be an issue with
missing quotes.  Always quote template expression brackets when they
start a value. For instance:

    with_items:
      - {{ foo }}

Should be written as:

    with_items:
      - "{{ foo }}"

fatal: [xx.xx.xx.xx]: FAILED! => {}

[valentinocossar]

Hi, maybe could be an issue related to the hosts configuration? See #6.

[eriteric]

Any progress on this? I am using Ansible 2.5.3, the Cloudflare Origin Certificate Trellis addon, and get an error...

My wordpress_sites file:

  example.com:
    site_hosts:
      - canonical: example.com
        redirects:
          - www.example.com
    local_path: ../bedrock # path targeting local Bedrock site directory (relative to Ansible root)
    repo: git@bitbucket.org:xxxxxx/example.com.git # replace with your Git repo URL
#    repo_subtree_path: '.' # relative path to your Bedrock/WP directory in your repo
    branch: master
    multisite:
      enabled: false
    ssl:
      enabled: true
      stapling_enabled: false
      provider: cloudflare-origin-ca
    cache:
      enabled: true

I issue the command:

    `./bin/database.sh   production   example.com  backup`

I get the following error:

The task includes an option with an undefined variable. The error was:
'unicode object' has no attribute 'wordpress_sites'

The error appears to have been in '/../../../trellis/database-backup.yml': line 48, column 7, but
may
be elsewhere in the file depending on the exact syntax problem.

The offending line appears to be:

    - name: Pull exported database from {{ env }} to development
      ^ here
We could be wrong, but this one looks like it might be an issue with
missing quotes.  Always quote template expression brackets when they
start a value. For instance:

    with_items:
      - {{ foo }}

Should be written as:

    with_items:
      - "{{ foo }}"

fatal: [xx.xx.xx.xx]: FAILED! => {}

I never solved it. I ended up just using WP-CLI aliases to push & pull the database, but still use the uploads script. You can pull to your local in one command by using the pipe operator. like:

wp @prod db export - | wp @dev db import - && wp @dev search-replace 'prodsite.com' 'devsite.test' --skip-columns=GUID

[henscu]

Thanks, @eriteric that looks much simpler..!