Added support for the aurweb role to the new openssh include mechanism,
that's baked into our sshd role.

Added the uwsgi_modifier1 option to nginx as described on [0] and also
change the chmod option on the socket to allow nginx to connect to it.

[0] https://gist.github.com/janoliver/85b682227bd9fcb8942885e60208bd76

Add the aur-dev changes to the aur playbook.

Add memcached to the playbook and also change the php extensions to use memcached.
Removed the apcu options from defatuls and added memcached settings. Added the php-memcached
packages to the list of needed packages and also remove the apcu tasks. Added the memcached
systemd unit file and enable and start it from the tasks.

Since git is installed on a separate step, remove it from the main list.

The original aur playbook file was missing a lot of the roles that were
added for the aur-dev playbook. Add them and renamed the file too.

Added a smartgit_socket option to the defaults. Reworked the tasks package
installation to look cleaner and also separated the cgit and git package installations
so we can trigger uwsgi reloads on updates. Changed the tubylaws repo update variable
to trigger the bylaws changes only when the tubylaws repository change, not the aurweb
one. Added tasks to install the apcu configuration, cgit uwsgi ini file, cgit rc file and
smartgit uwsgi ini file. Trigger an uwsgi reload in case the cgit-aurweb or git packages
change. Also added a few missing options to the aurweb configuration file. Rework the nginx
configuration file to use the cgit and smartgit uwsgi services.

Use the cgit_socket variable on the uwsgi template for cgit.

Added the template for the cgitrc configuration file.

The user for the socket was set incorrectly, fixed it.

roles/aurweb: Add the apc configuration variables

Added the apc to the php extensions on the playbook, instead of adding
it on the conf.d directory. Added the apc variables to the defaults file.

Added the templates for the apcu.ini configuration for php and for the uwsgi
config for cgit.

We need to use uwsgi for cgit hosting, so add the uwsgi role to the playbook.

Since the AUR has its own maintenance mode, there is no need for using the
maintenance role, so it was removed. Added a check on the aurweb config file
that checks for the maintenance variable and uses the variable for allowing
the ip address of the person running the playbook to access the AUR.

Add a variable that takes the SSH_CLIENT environment variable to save the
ip address of the machine actually running the playbooks. This can be used
on maintenance mode to allow in the person running the playbook and let
everybody else to see the maintenance page.

playboooks/aur-dev: Remove the intl php extension

Added maintenance mode to the aurweb role using the AUR internal mode. Also,
add the php-apcu-bc package to the list of required packages. Also running the
make required to create the translations.

We have reworked the entire tasks file to do an actual AUR installation.

Added the TU bylayws repository path and the wanted version for the TU bylaws repository.

Added a vault that contains the database password for the AUR.

Changed the playbook to add the missing php modules and add the AUR role. Also set the
domain and the git branch variables so the right version of the AUR is installed.

Moved all the systemd files to templates and also made sure every aur user
reference is using the variable. Changed the config to use the variables and
also changed the sshd config to use the ansible variables.

Made some changes to the defaults file to make the installation work.
Moved the db password to the vault, added the cache and limits variables
and also changed the username back to aur.

Renamed to aurweb

To make things consistent, rename the role to aurweb.

Jelle van der Waa authored 7 years ago and Giancarlo Razzolini committed 4 years ago

The ansible role for the Arch User Repository.

Thanks-to: Eli Schwartz <eschwartz@archlinux.org>

Now uses Github links for source/history of PKGBUILDs.

Also set zram_count to 1 because creating 48 devices on dragon is not
usable for swap. We don't need multiple devices anyway; SMP is handled
through multiple compression streams.