[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
(usagi-users 04099) Re: Default Router Preference - RFC4191, default route switching
- To: Linus LÃssing <linus.luessing@xxxxxx>
- Subject: (usagi-users 04099) Re: Default Router Preference - RFC4191, default route switching
- From: arno@xxxxxxxxxxxx (Arnaud Ebalard)
- Date: Wed, 07 Jul 2010 17:49:07 +0200
- Cc: usagi-users@xxxxxxxxxxxxxxxxx
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=natisbad.org; s=mail; h=From:To:Cc:Subject:References:Date: In-Reply-To:Message-ID:MIME-Version:Content-Type: Content-Transfer-Encoding; bh=brQO5B2QOxHMjezlsD+0YjJmuxGaNx5FGj XzBJh/n30=; b=u/rma3IkhCevuK9e7NCNJ2KYFRBK9xHg8ra4x8al0pRMEsNnpK eiElyb88TPIkTjt24ROMu8y5IaOwbLgzEhMgfXzlESQtAq9MSWw5TH5vLQ2NvVbZ /96b0uAmu+zR7oRwh31B7bLTjNpOGwrXwZW6/yvnhZzFeVtefOtE4MO+8=
- In-reply-to: <20100701230846.GA3783@Sellars> ("Linus LÃssing"'s message of "Fri, 2 Jul 2010 01:08:46 +0200")
- References: <20100610155436.GA5116@Sellars> <87hbkm54q8.fsf@xxxxxxxxxxxxxx> <20100701230846.GA3783@Sellars>
- Reply-to: usagi-users@xxxxxxxxxxxxxxxxx
- User-agent: Gnus/5.110009 (No Gnus v0.9) Emacs/23.1.50 (gnu/linux)
hi,
Linus LÃssing <linus.luessing@xxxxxx> writes:
> [snip]
>
> Cool! I now did some new tests with a custom-build 2.6.29.6
> kernel. It seems to work a little better, in that it now also
> switched the default route. The new netcat session goes over the
> right prefix and to the right router with the high preference.
> However, there still seems something wrong: The old netcat
> connection still uses the old prefix (which is good, otherwise I'd
> lose the connection, of course), but new packets take the
> high-preference default route immediately - which is going to fail
> if the gateway routers / the ISPs behind them do strict subnet
> checking.
Yes. I would have to reread the RFC (4861, 4191, 2461, 4863, ...?) to
check what the expected behavior is: IIRC, the prefix is associated with
the subnet (the link) and not the router. If I am right, the behaviour
you get makes sense.
> And just for clarification, a route ::/0 with the same preference
> is probably superfluous, right (though I tested both with the same
> result)?
I implemented that a long time ago in UMIP and do not remember what the
kernel maintains, i.e. one or two routes.
> The gateway is running a 2.6.32 kernel by the way.
That's a good version (stable and maintained).
Cheers,
a+