This may be a little off-topic for DNSOP, but has anyone considered submitting 
Errata for RFC 4291 to add the word "physical" before the word "interface" to 
the sentence

"A packet received on an interface with a destination address of loopback must 
be dropped"

?

Because, as it stands, if taken literally, even packets to lo0 should be 
discarded, thus effectively transforming so-called "IPv6 loopback" into "IPv6 
sinkhole", and leaving no formal address assignment whatsoever for true 
*loopback* functions in IPv6.

Be that as it may, since the "reservation" of the entire ::/8 block in 
http://www.iana.org/assignments/ipv6-address-space/ipv6-address-space.xhtml 
isn't backed up by any corresponding /8-level definition in 
http://www.iana.org/assignments/iana-ipv6-special-registry/iana-ipv6-special-registry.xhtml,
 I think the portions of ::/8 not mentioned in the latter registry are 
legitimately available for loopback purposes. Just don't put any packets on the 
wire with source or dest addresses in that range.

Or, as Mark said, just use a ULA address configured on a local interface.

                                                                                
- Kevin

-----Original Message-----
From: DNSOP [mailto:dnsop-boun...@ietf.org] On Behalf Of Paul Vixie
Sent: Wednesday, September 30, 2015 9:47 PM
To: John Levine
Cc: e...@isc.org; dnsop@ietf.org
Subject: Re: [DNSOP] Brian Haberman's No Record on 
draft-ietf-dnsop-root-loopback-04: (with COMMENT)



John Levine wrote:
>> It should be easy enough to create a local alias address for the 
>> purpose though.  "ifconfig lo inet6 add ::2 alias", salt to taste.
>
> Uh, no.  The *only* loopback address is ::1.  The rest of 0000::/8 is 
> reserved.

right. just like 127.0.0.0/8 is reserved. yet i use 127.0.0.2, .3, and so on, 
all the time. i think it's probably safe to intrude on this "reservation" for 
this use case.

> If you have a loopback software interface, you could set up a link 
> local address like fe80::1, but now your DNS software has to 
> understand link scoped addresses like fe80::1%lo.
>
> Having set up a DNS cache on my LAN using link local IPv6 addresses, I 
> can report that it doesn't work very well.

agreed.

> All in all, I think the advice to stick with IPv4 loopback addresses 
> is reasonable.  We can revisit this in 2050 when IPv4 is starting to 
> be phased out.

disagreed. ipv4 should die a-s-a-p. don't bring up any new ipv4 services unless 
you are sure they have to talk to the legacy internet. which is demonstrably 
not the case for localhost dns service.

now you don't see it:

root@family:/home/vixie # ifconfig lo0
lo0: flags=8049<UP,LOOPBACK,RUNNING,MULTICAST> metric 0 mtu 16384
        options=600003<RXCSUM,TXCSUM,RXCSUM_IPV6,TXCSUM_IPV6>
        inet6 ::1 prefixlen 128
        inet6 fe80::1%lo0 prefixlen 64 scopeid 0x2
        inet 127.0.0.1 netmask 0xff000000
        nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL>

now you do:

root@family:/home/vixie # ifconfig lo0 inet6 ::2/128 alias 
root@family:/home/vixie # ifconfig lo0
lo0: flags=8049<UP,LOOPBACK,RUNNING,MULTICAST> metric 0 mtu 16384
        options=600003<RXCSUM,TXCSUM,RXCSUM_IPV6,TXCSUM_IPV6>
        inet6 ::1 prefixlen 128
        inet6 fe80::1%lo0 prefixlen 64 scopeid 0x2
        inet 127.0.0.1 netmask 0xff000000
        inet6 ::2 prefixlen 128
        nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL>

ntpd is a grabby little thing:

root@family:/home/vixie # netstat -an | grep ::
tcp6       0      0 ::1.465                *.*                    LISTEN
tcp6       0      0 ::1.587                *.*                    LISTEN
tcp6       0      0 ::1.25                 *.*                    LISTEN
tcp6       0      0 ::1.993                *.*                    LISTEN
tcp6       0      0 ::1.143                *.*                    LISTEN
tcp6       0      0 ::1.995                *.*                    LISTEN
tcp6       0      0 ::1.110                *.*                    LISTEN
udp6       0      0 ::2.123                *.*
udp6       0      0 fe80::1%lo0.123        *.*
udp6       0      0 ::1.123                *.*
udp6       0      0 fe80::2a0:98ff:f.123   *.*

i had to alter these lines of my ipfw configuration:

add     pass    all     from any to any via lo0
add     deny    all     from any to { ::1 or 127.0.0.0/8 }
add     deny    ip      from { ::1 or 127.0.0.0/8 } to any

they now read:

add     pass    all     from any to any via lo0
add     deny    all     from any to { ::1 or ::2 or 127.0.0.0/8 }
add     deny    ip      from { ::1 or ::2 or 127.0.0.0/8 } to any

i had to add a line to ntp.conf:

restrict -6 ::1
restrict -6 ::2

noting, the other lines in that vicinity tell us things about
127.0.0.0/8 that the IETF might not know:

restrict 127.127.1.0

but anyway, it works:

root@family:/home/vixie # ntpq -p ::2
     remote           refid      st t when poll reach   delay   offset 
jitter
==============================================================================
 mm1.redbarn.org 108.61.194.85    3 u    1   64    1    0.287    3.617  
0.075
 ks.redbarn.org  208.75.88.4      3 u    2   64    1    1.171   -0.744  
0.000

--
Paul Vixie

_______________________________________________
DNSOP mailing list
DNSOP@ietf.org
https://www.ietf.org/mailman/listinfo/dnsop

_______________________________________________
DNSOP mailing list
DNSOP@ietf.org
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to