RFC1918 Addresses in routing paths
Hi Folks, I am wondering about the etiquette of reporting RFC1918 addresses back to hosts performing a traceroute. Example below: Hostname 1. ourhost.ourprovider.net.nz 2. f3-0-2.core1.wlg.ourprovider.net.nz 3. f0-0-4.core3.wlg.ourprovider.net.nz 4. f4-0-5.core2.akl.ourprovider.net.nz 5. g1-0-1396.u12.brfd.otherprovider.net.nz 6. g1-0-1043.u12.brh.otherprovider.net.nz 7. 10.65.32.1 8. 10.65.32.250 9. 10.69.0.2 10. a12-3-23.u21.tar.otherprovider.net.nz 11. fa7-4-1042.bigrouter.otherprovider.net.nz 12. otherhost.otherprovider.net.nz Any comments on this practice? Cheers, Jon ------------------------------------ Araneo Limited info (at) araneo.net.nz 5th Floor Central House 26 Brandon Street Wellington, New Zealand tel: +64-4-473-3932 fax: +64-4-472-0960 http://www.araneo.net.nz ------------------------------------
It's pretty common and accepted. While not maybe the absolute best practise as it does break things like pmtu discovery, it still works. It's a recurring (every 2-3 months) question on NANOG, so you could probably do a search on their list archives (www.nanog.org -> mailing list) and you'll see the flame wars about it for say, the last 8 years or so... As far as Telstraclear using it in your traceroute below, they're certainly not the only carrier or isp in NZ doing so. aj
On Tue, 5 Apr 2005, Alastair Johnson wrote: : It's a recurring (every 2-3 months) question on NANOG, so you could : probably do a search on their list archives (www.nanog.org -> mailing : list) and you'll see the flame wars about it for say, the last 8 years or : so... Nah, NANOG has been around for 10 years, so the argument has been going on for 10 years and probably longer on the Regional Operators list (or something close to that) that NANOG evolved from. Then every 2-3 months after that... :-) scott
Discussions about RFC1918 addresses surely can't have started before February 1996. Before that we talked about RFC1597 addresses and then only after March 1994.
On Tue, 5 Apr 2005, Andy Linton wrote: : Discussions about RFC1918 addresses surely can't have started before February : 1996. Before that we talked about RFC1597 addresses and then only after March : 1994. Ahh, I stand corrected, but that's at least 11 years worth of arguments several times a year for the 10, 172.16 and 192.168 ranges. Just a silly joke anyway in response to Perry's email. Did I forget the smiley face? ;-) scott
Jonathan Brewer wrote:
Oooh goody, one of my favourite pet rants. :) RFC1918 addresses on routers is bad because if the router has to generate an ICMP error message it is sent with an RFC1918 source address (as your traceroute shows). These packets are often filtered by various devices using loose or strict reverse path filtering causing; * Traceroutes to have "*"'s in the middle of them. * pMTU discovery to break (ICMP Fragmentation required packets never make it back to the source) Other issues arise such as people trying to ssh to one machine inside their local network due to routing glitches ending up sshing into someone elses router. This also shows that ISP's in NZ aren't dropping packets sourced from bogus addresses at their boundary :) obrantlink: http://coders.meta.net.nz/wordpress/archives/2005/01/27/the-evils-of-rfc1918...
participants (7)
-
Alastair Johnson -
Andy Linton -
Jonathan Brewer -
Juha Saarinen -
Perry Lorier -
Scott Weeks -
Tony Wicks