cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Announcements

769
Views
0
Helpful
3
Replies

LDP Neighbor Down detection for backup xconnect

Hi all

 

I have the following topology:

topo.png

 

On R3 I have a xconnect to R1 with a backup xconnect to R2. Something like this:

interface gigabitethernet 0/1
xconnect 10.1.1.1 1000 encapsulation mpls
backup peer 10.2.2.2 1000

R3 sees both neighbours, R1 & R2, over the same uplink interface.

 

Now R1 crashes and goes offline. What is the right/best mechanism, that R3 can detect this topology change in a fast way? There is no "port down" on R3, it must be some kind of timeout, that R3 switch over to the backup peer over R2.

Should I use BFD with OSPF? Is this enough, that R3 will switch to the backup peer? BFD with LDP neighbours don't exist, if I'm right. Or can I tune some LDP timers?

 

I also found BFDoVCCV, but this is not supported, when there is a backup peer configured.

 

I like to have something, that detects the issue in 3 seconds or less.

Everyone's tags (3)
3 REPLIES 3
Cisco Employee

Re: LDP Neighbor Down detection for backup xconnect

Re: LDP Neighbor Down detection for backup xconnect

I don't think that is possible, from the same page:

"Bidirectional Forwarding Detection over Virtual Circuit Connection Verification (BFDoVCCV) with status signaling is supported only on static pseudowires that do not have a backup peer. Explicit configuration of backup peers that violates this restriction is rejected."

 

I have adjusted the LDP timers to 2/6 and used BFD with OSPF in my lab setup. Currently I have a failover time from around 6 seconds.

 

I will make some more tests, also with the new ISR1100 series, in the next weeks. Maybe It's different in the current IOS XE Releases or there is something else that I can do.

Highlighted
Cisco Employee

Re: LDP Neighbor Down detection for backup xconnect

Are those devices directly connected? Having BFD for OSPF will not necessarily make it switchover to the backup peer (As the targeted LDP is a TCP session between the peers, so it would try to take any new route to that destination and maintain the session up). I am thinking you can make LDP session shorter maybe? mpls ldp holdtime XXX
CreatePlease to create content
Content for Community-Ad
August's Community Spotlight Awards