[6bone-pl] [kuznet@ms2.inr.ac.ru: Re: problems with v6 TCP on Sparc]

Rafal Maszkowski rzm w icm.edu.pl
Wto, 27 Paź 1998, 15:51:36 MET


Jezeli ktos nie czyta netdev. Bardzo ladnie to wyglada w praktyce:

root w 6bone-gw:/usr/src,0# ifconfig
lo        Link encap:Local Loopback
          inet addr:127.0.0.1  Mask:255.0.0.0
          inet6 addr: ::1/128 Scope:Host
          UP LOOPBACK RUNNING  MTU:3924  Metric:1
          RX packets:2677 errors:0 dropped:0 overruns:0 frame:0
          TX packets:2677 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0

eth0      Link encap:Ethernet  HWaddr 08:00:20:21:85:26
          inet addr:193.219.28.246  Bcast:193.219.28.247  Mask:255.255.255.252
          inet6 addr: fe80::a00:20ff:fe21:8526/10 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:2774720 errors:252 dropped:0 overruns:0 frame:504
          TX packets:928248 errors:13 dropped:0 overruns:0 carrier:13
          collisions:28184
          Interrupt:38 Base address:0x6600

zatoka    Link encap:IPv6-in-IPv4
          unspec addr:[NONE SET]  P-t-P:[NONE SET]  Mask:[NONE SET]
          inet6 addr: 3ffe:902:b::2/126 Scope:Global
          inet6 addr: fe80::c1db:1cf6/128 Scope:Link
          UP POINTOPOINT RUNNING NOARP  MTU:1480  Metric:1
          RX packets:20273 errors:0 dropped:0 overruns:0 frame:0
          TX packets:30557 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0

Tablice BGP:

>iB   20 04:27:33 3ffe:2600::0/24            3ffe:900:0:1b::1         zatoka 1225 5623 5609 33 2914 1849 6175 3274 i
>iB   20 09:19:22 3ffe:2800::0/24            3ffe:900:0:1b::1         zatoka 1225 6175 145 i
>iB   20 16:39:05 3ffe:2802::0/32            3ffe:900:0:1b::1         zatoka 1225 1312 i
                                                                      ^^^^^^
Interfejs nazywa sie zatoka...

'NEVER USE' dotyczy na pewno IPv6, w IPv4 mozna uzyc route bezkarnie. Nie
jestem tylko calkiem pewien czy z jadrem Linusa (nie CVS) bedzie tak samo, ale
nie widze istotnych roznic w zrodlach sieci.

R.

----- Forwarded message from kuznet w ms2.inr.ac.ru -----

Delivered-To: netdev-outgoing w peyote-asesino.nuclecu.unam.mx
X-Authentication-Warning: athena.nuclecu.unam.mx: majordom set sender to owner-netdev using -f
From: kuznet w ms2.inr.ac.ru
Subject: Re: problems with v6 TCP on Sparc
To: rzm w icm.edu.pl (Rafal Maszkowski)
Date: 	Sun, 18 Oct 1998 19:13:04 +0400 (MSK DST)
Cc: davem w dm.cobaltmicro.com, netdev w nuclecu.unam.mx
In-Reply-To: <19981017213936.U9206 w icm.edu.pl> from "Rafal Maszkowski" at Oct 17, 98 09:39:36 pm
X-Mailer: ELM [version 2.4 PL24]
Precedence: bulk

Hello!

> I do feel relief too - it is a real bug, not something I've done wrong and
> bother the gods. I'm not sure if mrtd and such will work correctly with /126
> routes but I can delay moving my some 10 tunnels to this machine - the old one
> is going to work for some time still. Mrtd is crashing but it is easier
> debuggable...

mrtd will be lucky with this setup.

Moreover, I bet it cannot work with your original setup.
mrtd (and all another routing daemons) DOES NOT understand
on link routes at all. They require corresponding prefix added
on interface i.e. you must make ifconfig 3ffe:xxxx::1/126 in any case.

IPv4 still support bsd-like PtP addressing semantics,
IPv6 does not and is not supposed to support it.


To resume:

How to make mrtd happy.
-----------------------

1. Create tunnels with "ip tunnel add INR mode sit local YOUR-ADDR remote 192.203.80.142 ttl 64"
2. ifconfig INR add 3ffe:xxxx:...:1/126
3. Start mrtd.
4. That's all, provided you:

   - NEVER USE "route add" or "ip route add" outside of mrtd.conf, no matter
   before mrtd started or when it is running.

   - Avoid to use static interface routes in mrtd.conf, if it is possible.

Never fiddle with interface routes by hands: kernel and mrtd
will make all this work. You can only break them.

Alexey

----- End forwarded message -----



Więcej informacji o liście dyskusyjnej 6BONE-PL