Funny and true story: 2 years ago we we needed to move in from the cloud to a internal network. Because of time pressure we got a PC as our first server, where

Funny and true story: 2 years ago we we needed to move in from the cloud to a internal network. Because of time pressure we got a PC as our first server, where we installed a set of tools for our embeded developers, from Gitlab, over Jenkins, LDAP backend, Nagios, Rocket.chat, Crowd, file sharing, nginx, Volumerize backups, nfs sharing, build artifact storage, a private Docker registry, we even run a build slave instance on it. This was supposed to be a intermediate solution until we move to the real infra. In the new infra they have all of the things like several stages of load ballancers, tons of firewalls between the servers, the slaves are physically in a different network, everything is insanely complex and takes at least 20 times more time to set up than you'd expect. This is why we still run on that old PC (we added 7 more as build slaves) and ca. 200 people use it now for 2 years on a daily basis, which seems pretty weird but it just works. Lately we saw that the network card started hanging and we needed to do a hard reboot which is not nice. The guys who are close to that PC (we use it remote) had a USB network card lying around and we asked them to connect it because the old one in the tower most probably had it's end of life because it has been used so much during the last 2 years. We're still on boarding more and more people and it's not clear when we will be able to move to the new very complex infra.