Gitlab Backups Rework
Currently we are reliant on the usage of Replicant (sysadmin/replicant) to mirror the repositories from Gitlab (invent.kde.org) and keep a backup copy of them.
Unfortunately, Gitlab itself is much more efficient at organising it's storage when compared with Replicant, so while the actual repositories only take up 218GB on Leptone (including shared artifacts such as those from CI builds), the 500GB of space allocated to Seges where Replicant currently runs has been exhausted.
Given that Replicant also misses backing up Wiki and Design repositories (with the second not being accessible externally either) we need to rethink how we handle backups of Gitlab.
My current thinking on this matter is that given that Gitlab's database and the underlying repositories should generally be in alignment, that we should use Borg to take backups of Gitlab much like we do with other aspects of our infrastructure. In the event of a failure, this would leave us with up to 24 hours of lost records which is an acceptable trade-off given that people can likely repush any missing code from their local copies.
While this would fit within the current free space of our Hetzner backup space (currently at 383GB free) it wouldn't give us much in the way of spare space to accomodate for future growth, and certainly wouldn't allow for the proposed use of that space for non-Hetzner server backups proposed in #23 (closed).
I'm therefore proposing we ask the Board to approve us increasing the plan from a BX40 (2TB - 9.90 EUR + VAT / month) to a BX50 (5TB - 21.90 EUR + VAT / month), which should allow us to move the Gitlab backups as well as those of our other servers.
This would also have the benefit of keeping the load of other server backups off of Hersili, which was one of the issues that impacted on Micrea during late evenings.