[ipv6-wg] What to do with RIPE Atlas probes that have only a ULA as IPv6 address?

As to the ATLAS-ULA discussion, here is a datapoint that may be
interesting to some.

After some network changes / upgrades, my probe was not happy with IPv6.
For the built-in measurements, IPv4 was fine but all the IPv6 measurements 
were showing on-off behaviour with a cycle time of 2 hours.

Initially I had no clue what was happening. Set up various IPv6 ping tests
to the probe, and they were clean, and the measurements still
showed the same outages. Also, measurements to other SLAAC clients
were also spotless.

While I was investigating something else however, I noticed these packets
on the VDSL PPPoE-link to the ISP:
15:25:39.366673 IP6 fd0c:8f98:a1d2:80:6666:b3ff:fec5:dee.54536 > 2001:500:2d::d
.53: 61413 TXT CHAOS? hostname.bind. (31)
15:25:41.911297 IP6 fd0c:8f98:a1d2:80:6666:b3ff:fec5:dee.46086 > 2001:500:2::c.
53: 64278 TXT CHAOS? hostname.bind. (31)

Note the source address of these packets. Indeed, looking at the
ATLAS portal, both the public IPv6 IP and ULA IP was listed on the
info page of my probe, and the ULA IP didn't have connectivity.

I diked out the ULA config on the OpenWRT router (not sure how 
it got there, probably PEBCAK but I didn't explicitely configure it),
and things are a lot better now.

Thanks to Philip for spotting the problem,

Geert Jan