Variable reset pr devel (#603)

* Need to check if is_local is defined in webserver meta dependencies. (#522)

* Ce dev refactor pr 1.x (#518)

* Making it easier to test with provision-target and ce-dev.

* Moving the provision forcing var back to plays so _init has it.

* Adding defaults vars and test script extra options.

* Adding a web server test to CI.

* examples string needs to be in quotes.

* Making sure is_local and _ce_provision_force_play are available to the _init role.

* Adding SSH keys to the provision user.

* Adding a --force to the test script.

* Explicitly adding vars to role.

* Fixing _init behaviour and adding SSH key for web role.

* Setting default PHP version to 7.4.

* Looking up the generated ce-dev SSH key instead of hard-coding one.

* We cannot run the ssh_server role locally, so excluding for tests of webserver role.

* Trying to remove user_root.yml in case it's breaking CI.

* Adding a verbose mode to the test script.

* Exposing the command in the test script.

* Trying hard-coded keys again.

* Changing location of data dir for test containers.

* Putting vars back and restricting CI to the 'web' example.

* Adding backup handling to ldap_server. (#525)

* Adding backup handling to ldap_server.

* Improving SSL docs and handling perms for openldap and letsencrypt.

* Cron user must be specified with file.

* Running as root, do not need a 'sudo' in this cron.

* Allowing 'gitLab' to disable Prometheus. (#530)

* Allowing 'gitLab' to disable Prometheus.

* Booleans to use in jinja2 as strings must be cast as strings.

* GitHub Actions - Rebuilt documentation. (#526)

Co-authored-by: Code Enigma CI <sysadm@codeenigma.com>

* Prometheus pr 1.x (#533)

* Allowing 'gitLab' to disable Prometheus.

* Booleans to use in jinja2 as strings must be cast as strings.

* Tidying up CI and adding a GitLab test.

* Fixing CI job description.

* Add private files support for Drupal in Nginx. (#535)

* Prometheus pr 1.x (#539)

* Allowing 'gitLab' to disable Prometheus.

* Booleans to use in jinja2 as strings must be cast as strings.

* Tidying up CI and adding a GitLab test.

* Fixing CI job description.

* Adding a firewall config preset to open port 80 for LetsEncrypt.

* Removing our unused ClamAV roles and adding a Galaxy role to common base. (#541)

* Revert "Moving OSSEC pkill to use process_manager role instead. (#258)" (#544)

This reverts commit 73c7bd0.

* Backing out of Packer logging.

* Moving key servers to a variable so we can set them. (#555)

* Moving key servers to a variable so we can set them.

* Allowing us to disable sending keys completely.

* Oops, doubled up on existing functionality.

* Fixing var name.

* Adding a reboot option to the patching role. (#557)

* Add minimal support for Aurora RDS instances (#567)

* Attempt to create an RDS read replica.

* Use new task to create Aurora RDS instances.

* Try and fix linting issues.

* Don't pass max_storage variable for Aurora instances.

* Remove more storage related vars from Aurora RDS instance creation task.

* Add profile and region to read replica creation.

* Try creating the Aurora read replica another way.

* Add some debug info.

* Work around the silly registering of variables in Ansible.

* Rename an RDS CloudWatch task for Aurora DBs and remove RDS debug info.

* Add some Aurora info to aws_rds README file.

* Use reader instead of replica for Aurora readers.

* Remove db_cluster_identifier variable from non-Aurora RDS task.

* Gpg servers fix pr 1.x (#571)

* Moving key servers to a variable so we can set them.

* Allowing us to disable sending keys completely.

* Oops, doubled up on existing functionality.

* Fixing var name.

* Using a pipe to grep with 'command' cannot work, refactoring.

* Making CI use the meta deploy role to test gitlab.

* We mustn't assume AWS servers for deploy and controller.

* Support termination protection in EC2. (#573)

* Support termination protection in EC2.

* Fixing CI vars.

* Fixing CI vars.

* Fix managed SSL key perms and the variable used for the private key. (#575)

* Ec2 subnet lookup pr 1.x (#583)

* First pass at EC2 subnet detection.

* Touching subnet file to ensure it exists.

* Trying a different approach, file module didn't work.

* Switching back to file module.

* We need to create the directory for new servers too.

* Bad variable name.

* Ec2 subnet lookup pr 1.x (#589)

* First pass at EC2 subnet detection.

* Touching subnet file to ensure it exists.

* Trying a different approach, file module didn't work.

* Switching back to file module.

* We need to create the directory for new servers too.

* Bad variable name.

* Changing subnet lookup order to check for defined subnet first.

* Fixing gitlab-runner overriders so upgrades do not break the runner. (#586)

* Fixing gitlab-runner overriders so upgrades do not break the runner.

* Fixing override file template.

* Hopefully fixing CI.

* Making sure the service directory exists.

* We cannot use the deploy meta role in CI because of LDAP.

* Changing dir perms and adding a force.

* Gitlab runner service override pr 1.x (#591)

* Fixing gitlab-runner overriders so upgrades do not break the runner.

* Fixing override file template.

* Hopefully fixing CI.

* Making sure the service directory exists.

* We cannot use the deploy meta role in CI because of LDAP.

* Changing dir perms and adding a force.

* Debugging gitlab-runner directory creation issues in CI.

* Fixing linting error.

* Removing verbosity again but leaving 'stat' command in.

* Making the _exit role reset variables to defaults prior to next run.

* Trying builtin Ansible meta clear_facts behaviour.

* Adding an inventory refresh to _init.

* Backing out of ce-provision changes, this can all be done in config.

Co-authored-by: EmlynK <emlyn.kinzett@codeenigma.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Code Enigma CI <sysadm@codeenigma.com>

Author: 4 people

roles/_exit/defaults/main.yml

@@ -11,7 +11,3 @@
     content: "{{ current_play_dir_md5 }}"
     dest: "{{ _ce_provision_data_dir }}/{{ current_play_md5_file }}"
   delegate_to: localhost
-
-# Clear all existing facts of targeted hosts so Ansible reloads them
-- name: Clear gathered facts from all currently targeted hosts.
-  ansible.builtin.meta: clear_facts

roles/_exit/tasks/main.yml

@@ -40,10 +40,6 @@
   ansible.builtin.set_fact:
     previous_play_dir_md5: "{{ lookup('file', '{{ _ce_provision_data_dir }}/{{ current_play_md5_file }}' ) }}"
 
-# If we are here then our play is already running, so it doesn't matter if we overwrite temp host data.
-- name: Refresh inventory to ensure we have correct host information.
-  ansible.builtin.meta: refresh_inventory
-
 - name: Stop play if playbook md5 is unchanged.
   ansible.builtin.meta: end_play
   when: