[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: FC5: Error at ip -6 route add

On Mon, 2006-09-11 at 23:20 +0200, Eder, Norbert wrote:
> Hello,

> I have a problem with FC5 and the Command "ip -6 route add fd25:8c9a:a296:a000::/64 via fd25:8c9a:a296:3300:20d:8800:68:4afb".
> This is not working. On FC3 it was working very well.

	fd25:  ...  What in blue blazes is THAT???

	What scope is that address suppose to be?  Those addresses don't look
legitimate.  It's not multicast, or link-local, or site local, or global
unicast.  I know they were formulating something for "globally unique
local addresses" but that doesn't look familiar either (I'll admit I'm
not up on the latest debates and resolutions in that arena).  If you are
using globally unique local addresses, that would be outside of my

> The error-message is:   "RTNETLINK answers: Invalid argument"

	Which is telling me the kernel doesn't know how to deal with an
argument passed to it (like one or both of those addresses).

> On FC5 shows me ip -V : ip utility, iproute2-ss060110
> On FC3: ip utility, iproute2-ss040831

	Looking at the latest ip command pages, it appears that they have now
removed the old scope value of "site", leaving "host", "link", "global"
and numerical.  The IETF deprecated site locals a couple of years ago.
I have no doubt the FC3 utilities would have still supported them (I
used them back then) but they seem to be gone now.  They may no longer
be valid and that may be the source of your problem.  FC3 may have
considered that weird fd25 TLA prefix to be site local and now it's link
local and now you are broken.  Are you really using those addresses?  If
so...  Why?

> What is wrong at this command?
> Can anybody help me??

	Just my personal opinion but I think it's the addresses you are using.
It probably doesn't like that "via" but it really could be either of
them.  If it now thinks that's a link local, you may also have to
specify a "dev" to define what device to associated it with (one of the
reasons why you should use a prefix from an appropriate scope).  BUT,
you can't ROUTE link locals, so the address you are trying to route
would be invalid.  Damned if you do and damned if you don't.

	IPv6 addresses are not like IPv4 addresses.  You have to take note of
the scope of the addresses.  You can't just chose arbitrary addresses
and something non-standard that worked at one time may well fall on its
face in another revision.

	Try using something more standard.  You can get free /48 and/or /64
blocks of global unicast addresses from numerous sources and even get
them tunneled.  Set it up right.

	Better yet...  Why are you using static routes like this?  I've been
working with IPv6 for over 5 years now and never needed to do this.
Just use autoconf from your routers (radvd or quagga) with the
appropriate prefix and let your leaf nodes autoconfigure their own
addresses and default routes.  Unless you got some funky static routes
in there but then you recurse back to the bad address problem.  You
could try specifying the scope explicitly when configuring the addresses
but I really think that's just shooting your self in the foot when you
are getting that far out of standard.

> Thanks.
> With kind regards 
> Norbert 

Michael H. Warfield (AI4NB) | (770) 985-6132 |  mhw WittsEnd com
   /\/\|=mhw=|\/\/          | (678) 463-0932 |  http://www.wittsend.com/mhw/
   NIC whois: MHW9          | An optimist believes we live in the best of all
 PGP Key: 0xDF1DD471        | possible worlds.  A pessimist is sure of it!

Attachment: signature.asc
Description: This is a digitally signed message part

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]