[c-nsp] IOS XR - vrrp and protocol tracking
Pshem Kowalczyk
pshem.k at gmail.com
Mon Feb 21 19:00:08 EST 2011
Hi,
On 8 February 2011 15:22, Pshem Kowalczyk <pshem.k at gmail.com> wrote:
> Hi,
>
> We have a relatively simple setup with two ASR9k (running 3.9.1)
> providing a virtual gateway (using VRRP) for a set of servers
> (appliances) connected through a few switches. VRRP is augumented with
> BFD for faster detection of problems. VRRP runs on subinterfaces. When
> BFD detects that the other side is gone as expected assumes the IP
> address and forwarding from the servers continues. Towards the servers
> though the forwarding is broken, as the 'local' interface remains up
> and connected network is still in the routing table. We can run BFD
> with the appliances (only one appliance per vlan), but we have not
> found way to take the subinterface down based on bfd status. we can do
> this for static routes, but not for connected interfaces.
> Any ideas on how this could be fixed? I want the ASR that does not
> have visibility of the appliance (using BFD) to NOT advertise the
> connected network.
Just a follow-up on the subject, in case someone else needs it. We
have found a method of disabling the line protocol on the subinterface
using ethernet CFM to trigger EFD (Ethernet Fault Detection). The
convergence time is not as good as BFD it still well under 1 sec,
which is good enough for us. This method also requires the remote end
to support CFM.
Relevant config:
ethernet cfm
domain TEST level 2
service CUST1 down-meps
continuity-check interval 100ms
mep crosscheck
mep-id 102
!
efd
log efd
log crosscheck errors
interface TenGigE0/1/0/0.101
vrf Cust
ipv4 mtu 1500
ipv4 address 172.28.252.49 255.255.255.240
ethernet cfm
mep domain TEST service CUST1 mep-id 101
!
!
encapsulation dot1q 101
We also going to configure this only on the 'primary' path to make
sure that a CFM misconfiguration does not take down both of the links.
kind regards
Pshem
More information about the cisco-nsp
mailing list