Metro Ethernet Multiple Links

[JACQUES DU PLESSIS]

Hi Guys,

We have 3 offices connected to each other via a Metro Ethernet network. One office has 2 lines into the Metro (call it HQ), and the other two one line each. The links are configured as P2P from HQ to the other offices. Intermittantly the sites are loosing connectivity, and the fix (or rather workaround) is to shut and unshut the interfaces at HQ, but couple of days later it happes again.

What I saw was that the HQ switch seemed to have have mac addresses flapping between its interfaces. This makes sense because the Metro is effectively a big switch and because it is layer 3 interfaces spanning tree will not block one.

To support this as well, is that I put a layer 3 switch in the path of one of the two HQ links and we have been running ever since. However, we are adding more and more sites with 2 lines and each is now requiring a 2nd layer 3 switch which is becoming quite expensive.

My question is, is there a better way of doing this? I am out of ideas on this one.

Thanks!

Jacques

[Edison Ortiz]

Are you connecting your edge switches to the Metro-E cloud as Layer2 or Layer3 switchports?

What are the switch models at each of the locations?

Can you provide portion of the config?

Any error messages in the log?

[JACQUES DU PLESSIS]

Hi,

All the switch ports connecting to the cloud is layer 3 ports, and the models are Cat 4506 12.2(53)SG1. The config is pretty straight forward, the HQ switch has two links, both configured with /30 subnets, each pointing to a branch. At the branches it is also layer 3 ports, with matching /30 subnets. Static routing is used:

interface GigabitEthernet2/7

description **** Metro ****

no switchport

ip address 10.254.0.193 255.255.255.252

ip policy route-map pbr

qos trust dscp

tx-queue 1

   bandwidth 5 mbps

tx-queue 2

   bandwidth 25 mbps

tx-queue 3

   bandwidth 30 mbps

   priority high

   shape 30 mbps

tx-queue 4

   bandwidth 40 mbps

service-policy output xxxx

Jacques

[Edison Ortiz]

You mentioned mac-address flapping - can you post the error message from the log?

Is this a new implementation?

Are you having issues since day 1?

You have custom queue configuration in the physical port and also have a service-policy attached to it, why?

What exactly is within the service-policy?

Can you show the PBR portion of the config?

Thanks