Ethernetsubinterface indirect link failure direct connected subnet

_networker_ · ‎01-10-2014

Hello. Given the following topology:

R1#SW1-----
           |
          SW3--Host
           #
          SW4--Host
           |
R2#SW2-----

R1 and R2 are connected via an Ethernet dot1q Trunk to SW1/SW2. On SW1/SW2 there're static access ports which connects to SW3/SW4. SW3 is connected to SW4 via Trunk.

SW1/SW2 and SW3/SW4 are under different administration.

The following problem: R1 and R2 are member of the same IPv4 /25 Network. That means the subinterface on R1 uses 1.0.0.1/25 and R2 uses 1.0.0.2/25. The Hosts are also Members of this network.

The following Problem: If the Link SW1 to SW3 fails, the Subinterface on R1 will remain up and R1 will send traffic to the network 1.0.0.0/25 to SW1 where the traffic will be dropped. The same situation if the link SW2-SW4 fails.

Are there any possibilites to automatically shutdown the subinterface on R1 or R2, so that the directly connected route will be removed?

R1/R2 use IOS XR

Jon Marshall · ‎01-10-2014

You may be able to use IP SLA and EEM to do this ie. you use IP SLA to keep track of the state of the link and if it fails you then call an EEM script to shut down the interface.

If so then you should move this post to the EEM forum -

https://supportforums.cisco.com/community/netpro/network-infrastructure/eem?view=discussions

one thing that is a bit confusing though is you have a trunk from SW1 to R1 and then a trunk from SW3 to SW4 but only an access port link from SW1 to SW3 so are you only running HSRP for one of the subinterfaces ?

Jon

_networker_ · ‎01-10-2014

Hello,

thank you for you reply. IP SLA and EEM could be indeed a solution. Is this a normal "best practice"?

Regardinf your question: As written before, SW3 and SW4 are not under the same administration, e.g. you could think of virtual separation betweeen R1/SW1/R2/SW2 an the other devices regarding logical and physical setup.

HSRP is another sub-question: It´s indeed configured on the subinterfaces of R1/R2 to provide a redundant gateway towards the hosts

Jon Marshall · ‎01-10-2014

thank you for you reply. IP SLA and EEM could be indeed a solution. Is this a normal "best practice"?

There is no best practice here really because your setup is not best practice ie. your problem is because you have so many switches between the routers.

HSRP is another sub-question: It´s indeed configured on the subinterfaces of R1/R2 to provide a redundant gateway towards the hosts

Yes, but my question was if there is no trunk between SW1/SW3 and SW2/SW4 ie it is just an access port in a particular vlan then HSRP would only work for that vlan because the HSRP messages for any other vlan would not be allowed across the links because they are not trunk links.

Jon

_networker_ · ‎01-12-2014

jon.marshall schrieb:

HSRP is another sub-question: It´s indeed configured on the subinterfaces of R1/R2 to provide a redundant gateway towards the hosts

Yes, but my question was if there is no trunk between SW1/SW3 and SW2/SW4 ie it is just an access port in a particular vlan then HSRP would only work for that vlan because the HSRP messages for any other vlan would not be allowed across the links because they are not trunk links.

Jon

No, there`s no trunk. It should be an access-port, because i don't want to mix different administrative topologies. HSRP should only work for this vlan. On SW3/SW4 there`re some other vlans, which don't exist on SW1/SW2...