rpsltool & odd prefixes in the RIPE db
Hello, Running rpsltool[1] against the RIPE database, it broke on a rather odd object in the RIPE database:
inet6num: 2a02:4c8::1/64 netname: ITDNET descr: ITD Network, AD country: BG ........ route6: 2a02:4c8::1/64 descr: ITD Network AD origin: AS9070 .........
That doesn't seem right - the correct prefix would be 2a02:4c8::/64. I've asked the RIPE NCC about this, and they explained that internally, the database silently discards the host bits. So for the database, this is 2a02:4c8::/64. This is also visible in whois queries: searching for 2a02:4c8::1/64 yields no results, you have to search for 2a02:4c8::/64. Upon asking, the RIPE NCC responded that this is technically valid according to the relevant RFCs. But, they acknowledged that this behaviour can be confusing, and say they intend to disallow it at a later time. Down to my specific question: has anyone made the appropriate modifications to rpsltool to handle these kind of objects, and also silently discard the host bits in the address? I imagine we can't be the first people running into this. [1] http://www.linux.it/~md/software/ cheers, Erik Romijn
Erik, On Friday, 2012-05-04 14:18:01 +0200, Erik Romijn <eromijn@solidlinks.nl> wrote:
That doesn't seem right - the correct prefix would be 2a02:4c8::/64.
I've asked the RIPE NCC about this, and they explained that internally, the database silently discards the host bits. So for the database, this is 2a02:4c8::/64. This is also visible in whois queries: searching for 2a02:4c8::1/64 yields no results, you have to search for 2a02:4c8::/64.
I agree that the current situation is non-optimal. :) Note that the lookup does tell you what it is doing on query: %WARNING:905: fixed lookup key % % The key "2A02:4C8::1/64" has been changed to "2a02:4c8::/64" for lookup. Regarding the update path, I think we should reject entries with non-zero bits after the CIDR portion of the network address. I'm not sure what to do about the existing entries. I'd recommend a cleanup, actually, since they imply confusion on the maintainer's part. -- Shane
participants (2)
-
Erik Romijn
-
Shane Kerr