[c-nsp] CEF Load Sharing question

[Rodney Dunn]

The *'s are probably because of the rate limiting of ICMP
packets to the box.

Do the trace and enable "debug ip packet <ACL>" and
see if you see the IP packet come back in.

Try turning off the unreachable rate-limit and see if that
does it:

On Wed, Jan 28, 2009 at 04:06:13PM -0500, Drew Weaver wrote:
> I have a 3560-48TS which is connected via gig-e to 3 switches(routers).
> 
On Wecape sequence to abort.
Tracing the route to 1.1.1.1

  1 1.1.1.1 20 msec 20 msec * 
102_#tr 1.1.1.1

100_(config)#no ip icmp rate-limit unreachable ?
  DF  code 4, fragmentation needed and DF set
  <cr>

100_(config)#no ip icmp rate-limit unreachable 
100_(config)#


Type escape sequence to abort.
Tracing the route to 1.1.1.1

  1 1.1.1.1 20 msec 20 msec 28 msec
102_#

Rodney


, Jan 28, 2009 at 04:06:13PM -0500, Drew Weaver wrote:
> On the 3560:
> 
> G0/1, g0/2, and g0/3 are part of VLAN 2
> 
> interface Vlan2
>  ip address x.x.x.4 255.255.255.248
> end
> 
> Ip route 0.0.0.0 0.0.0.0 Vlan2 x.x.x.1 track 1
> Ip route 0.0.0.0 0.0.0.0 Vlan2 x.x.x.2 track 2
> Ip route 0.0.0.0 0.0.0.0 Vlan2 x.x.x.3 track 3
> 
> Routing entry for 0.0.0.0/0, supernet
>   Known via "static", distance 1, metric 0, candidate default path
>   Routing Descriptor Blocks:
>     x.x.x.3, via Vlan2
>       Route metric is 0, traffic share count is 1
>   * x.x.x.2, via Vlan2
>       Route metric is 0, traffic share count is 1
>     x.x.x.1, via Vlan2
>       Route metric is 0, traffic share count is 1
> 
> On VLAN 3 (x.x.x.129) I have two servers, one is a Windows 2003 server, and one is a Redhat 5 server (the 3560 is the gateway for these two servers).
> 
> I know this is kind of nit-picky, but when I run a traceroute from the Linux server it looks like total * city:
> 
> [~]# traceroute x.x.x.1
> traceroute to x.x.x.1 (x.x.x.1), 30 hops max, 40 byte packets
>  1  x.x.x.129 (x.x.x.129)  2.476 ms  2.662 ms  2.882 ms
>  2  x.x.x.1 (x.x.x.1)  0.744 ms * *
> 
> If I trace route past the routers to the internet there are many more dozens of *** along the way.
> 
> The windows 2003 server never shows any * * * during trace routes so I am assuming this has something to do with the way linux is doing the trace route?
> 
> Is there a better way to do load sharing between l3 devices?
> 
> Thanks,
> -Drew
> 
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/