Added a new server for the mail migration.

Both boxes died and are no longer responsive, retire them.

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.

Apparently the host without prio is the traffic shaped one which has been very slow for us.

It's an old box that's died and we have no contacts for the box. We haven't been able to reach it for a long time. Let's consider it dead and move on. It hosted no critical infrastructure.

Create a new VPS for running the master rebuilderd daemon which
coordinates the rebuild tasks to rebuilderd ndoes.

Signed-off-by: Jelle van der Waa <jelle@vdwaa.nl>

sgp and mex both have disk errors and died, so remove them to hosts.

Signed-off-by: Jelle van der Waa <jelle@vdwaa.nl>

Given flyspray is not maintained at all and that it is a security liability,
we are creating a VPS for it and isolating it from the rest of the resources.

This took quite a bit of manual work on the particular systems as they were partitioned in all sorts of ways that didn't allow for a big continuous chunk of storage such as required by the archive. I've manually repartitioned everything and it seems to work just fine. Now all systems have a big 4TB / and that's that.

add to mysql-servers for monitoring and fix the php-fpm extensions.

Signed-off-by: Jelle van der Waa <jelle@vdwaa.nl>

This is a new role for syncing the archive. Currently only ger.mirror.pkgbuild.com is capable of this due to disk space requirements. It can probably be extended fairly easily to other hosts.

Ansible does not allow group names to include a hyphen.