benjojo.co.uk

one honk maybe more

benjojo posted 21 Aug 2026 09:06 +0000

One of the bigger mistakes of the RIR/Internet Addressing world was allowing non CIDR aligned stuff to exist on v4, all be it that sin was committed a long time ago (and you could argue they knew no better...), but my god it's such a bug magnet.

Huge hate for the fact that you have to support two kinds of IP blocks, a range 1.2.3.4 -> 1.2.3.7 and a CIDR 1.2.3.0/24

wolf480pl@mstdn.io replied 21 Aug 2026 09:08 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/qT3WzH43vdxqGz4nNT

@benjojo
is the end of the range inclusive?

benjojo replied 21 Aug 2026 09:15 +0000
in reply to: https://mstdn.io/users/wolf480pl/statuses/117132708404189303

@wolf480pl sorta? you get a row that looks like this in most situations:

Name        |Value      |
------------+-----------+
type        |ipv4       |
record      |192.58.41.0|
size        |12544      |

Most RPSL databases (See almost every RIR except ARIN) do have a start and end, but I dont know if the database actually cares or if that is a display thing

benjojo replied 21 Aug 2026 09:33 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/qT3WzH43vdxqGz4nNT

If you dig deep enough into the RIR allocation files you will see some wild stuff, stuff that makes you think "what was the story there"

My "favourite" is this 156.67.6.0/24 that was blasted into 6 bits, but most of the bits going to de.terratransit and the rest going to other uniq people in tiny 8 address chunks

record     |size|accountowner    
-----------+----+----------------
156.67.6.0 |   8|f02ca83b-3aaa-41
156.67.6.16|   8|9a15b124-5677-46
156.67.6.24|   8|8c5cf919-7bab-4b
156.67.6.32|   8|b87411e5-c3a5-40
156.67.6.40| 216|de.terratransit 
156.67.6.8 |   8|92ce2ff5-c275-4d

Needless to say, this block is announce-able (at least "legally"), so this block got extremely burned for internet use, and most reasonable CIDR use too.

job@bsd.network replied 21 Aug 2026 09:59 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/552vWs33TxHDNZd8vL

@benjojo the global routing system is not the only use case for globally unique IPs. So it makes sense to me that you can request smaller-than-/24 blocks if you need less than /24

benjojo replied 21 Aug 2026 10:02 +0000
in reply to: https://bsd.network/users/job/statuses/117132912164887527

@job Sure, but most(*) addresses assigned are used for the global routing system, I assume based on the dates of those buckshot allocs (1992) they were not intended for the GRT, but /shrug, my side has to parse it all the same

phessler@bsd.network replied 21 Aug 2026 09:45 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/qT3WzH43vdxqGz4nNT

@benjojo we have a single block that is basically a /20+/21 combined and I've given up trying to represent it as the official assignment. In most of our internal databases and docs I did the above, referenced the official inetnum in both and moved on.