Hi!
I believe that the address assignment chain (the data you can find in RIPE/ARIN/… databases) does not really tell much about where are the addresses really BGP-announced and used. And GeoIP databases are much less reliable for IPv6 compared to IPv4 in my experience.
I think the best way to pinpoint the physical location is by measuring RTT from the public probes and “triangulate” from the nearest ones. Bearing in mind that light propagation delay in common fibers is around 1 ms for each 200 km and that active network elements (switches, routers) usually add 1-5 ms delay to each datagram transiting them, we can make approximate assumption about maximum real-world distance from seeing low (<10 ms) delays. I tend to believe that anything below 10 ms is probably in the same city.
On the other hand the higher RTTs are not significant for this reasoning - there can be many reasons for suboptimal routing over distant Internet Exchange etc.
Luckily, there are multiple public probe networks that fielded thousands of small low power computers that can be asked to run ping+traceroute for us and report back the results. RIPE Atlas for instance: https://atlas.ripe.net/ or GlobalPing: https://globalping.io/ .
Long story short, I tried this method for both nodes in question and I think this places 2a0c:b641:b21:2b:6801:5eff:fe75:a9e7 to Seoul, KR:
For Panama the evidence is scarce but 14 ms to San Jose, Costa Rica seems to me close enough:
