[c-nsp] IPv6 not Fast/CEF-switched on 3640 12.3T

[Daniel Roesen]

On Mon, Feb 28, 2005 at 11:42:04AM +0100, Oliver Boehmer (oboehmer) wrote:
> >> Can you use a different tunnel encaps? ipv6-over-ipv4 (tunnel mode
> >> ipv6ip) should provide better results..
> > 
> > Unfortunately not, as the tunnel must support MPLS and IS-IS...
> 
> Hmm, I fear there's not much I can do at this time if you need ISIS for
> IPv6.. Otherwise you could create two tunnels, one GRE for MPLS and
> IPv4, and one ipv6ip for IPv6 with possibly another routing protocol..

I've now bypassed the 3640 completely by using an ipv6ip tunnel ontop
the GRE tunnel. The ipv6ip tunnel is terminated by different routers
than the GRE tunnel (where one side is done by the 3640), so the 3640
should see only ipv6ip (proto 41) packets coming out of the GRE tunnel
and being forwarded to another router terminating the ipv6ip tunnel.

Unfortunately, the same overload condition is still there, but all
"sh int ... switching" counters of all involved interfaces show that
the traffic is fast/CEF-switched. "sh cef not-cef-switched" indicates
that the ipv6ip packets are being counted in the "Receive" column, not
the "Unsupp'ed".

I'm even trying a much lower pps group (25 pps), and still things become
unusable.

I'm pretty much out of ideas now (no, I can't open TAC cases).


Best regards,
Daniel

-- 
CLUE-RIPE -- Jabber: dr at cluenet.de -- dr at IRCnet -- PGP: 0xA85C8AA0