This year, I took the time to place a “Server” at the YoloColo the C3Noc (Network Operations Center) runs at the Chaos Communication Congress. The helpful angel at the YoloColo asked me if I need a IPv4 legacy IP address. Saying no to this question made his day (and mine).

The server mostly came up without issues. As I didn’t run a IPv6 only setup, I ran into a few entertaining issues. SLAAC worked out of the box, which I was able to test earlier at an assembly desk. This brought down the whole setup at the YoloColo to around 5 minutes.

About time?

Most NTP Timeservers via *.pool.ntp.org are mostly only IPV4. Finding this took a bit of time (pun intended) and fixing it was not as straight forward I hoped. Luckily, I found a List of some IPv6 enabled NTP servers in the Linux Reviews wiki. This solved the clock drift (no RTC Battery on the board) and brought up the services that didn’t like having a time that was off by a few hours.

Apparently the NTP servers starting with 2 should be also available via IPv6, but this wasn’t true for me. I hoped that the NTP servers are all IPv6 ready, but apparently this is not the case.

Docker & IPv6

My next issue was that the outbound IPv6 connections from my containers were not working. A bit of digging, cursing and a few sips of club mate later I finagled this into the live config and there’s actually documentation for this already existing. Neat.

In the end, I took the shortcut and bound everything I needed to host via network_mode: host. This works as I only needed a few ports to be working for the services running.

After getting past those challenges, I got everything running and ended up pushing almost 6 TB of data during congress.