The last weekend was claimed by 4 different CCC events. I decided to go to the Goulash programming night of the Entropia e.V. Karlsruhe as a member of the Network Operating staff.
The former was nice, the latter was a bad choice in some way.
I met a lot of old and new friends at the GPN. Due to some circumstances, it was also a chance to get in contact with a great amount of new people I had never met before. The University in the middle of the city was also a nice starting point for discoveries, shopping tours and simply going to the restaurant.
While the restaurant made me pay € 10.- too much for the food which had bad effects on my digestive system after a short while, the City Rallye Ralf created was clearly one of the best highlights of the entire event. However, there were also some two or three interesting lectures (I am hard to impress).
The network didn't run smoothly at all most of the time. The initial setup took slightly longer than intented, but that was mainly our fault since we tried to debug a VLAN config while filtering the probes. But later on the first day, the network was running.
Until the afternoon of the second day, all network traffic was running over a Cisco VPN to the VPN server, which was shaped to 1.5Mbit. This is enough usually for a single person, but with a hundred people who want internet access, the per-user bandwidth was unbearably low. So on the second day we got a technician of the University of Karlsruhe to provide us with a network cable to the University datacenter.
However, that's when the differential problems started to kick in. A couple of hours after activation, our first Intel e1000 died. We replaced the router since it was onboard hardware, and used the replacement machine. This machine had an e100 and four Sun Happy Meal cards. On the morning of the third day, both e100 ports died and everything ran over the Happy Meal card which is way more resilient with regard to voltage potential.
Unfortunately, the death of the e100 happened at a time when only me was around and I didn't have root passwords to the router itself, so I had to bridge around it for the users VLAN and reconfigure the name and DNS servers to work around the defunct admin VLAN. This worked more or less, but took away my capability to make contact with the rest of the admin staff who were at home. Unfortunately I also wasn't in possession of a telephone number of the staff members.
Another problem was that the VPN connection kept coming back and buggering up the routing, which was also a source of problems twice.
As a conclusion, we need high voltage resilient hardware next year, and we need to ensure that internal organization between the netadmins takes place before the first day of the event.