Create a Post
cancel
Showing results for 
Search instead for 
Did you mean: 
nabs_nabs
Contributor

clusterXL failover and dynamique routing

Jump to solution

 

Hi all,

I'm doing ebgp with a cisco router and gateway clusterXL; And that works fine. It uses the clusterXL VIP of the gateway to establish the ebgp session with the cisco router.
But when i do a clusterxl failover, the eBGP session on the standby member (newly master) takes about 40sec for passing from "idle state" to "established state" while clusterxl switch was done imediately..
The problem is that this switchover makes a blackhole trafic during 40sec.
Do you know if there is a way to tune the switchover ? for exemple to synchronize clusterXL switchover with ebgp session ?
or something else ton decrease this outage time ?

 

Nb: I use r81 version on security gateway

Thank you for your reply.

kind regards

0 Kudos
1 Solution

Accepted Solutions
Chris_Atkinson
Employee
Employee

Yes it's recommended in such scenarios.

Sets the Check Point system to maintain the forwarding state advertised by peer routers even when they restart.

This minimizes the negative effects caused by the restart of peer routers.

View solution in original post

0 Kudos
5 Replies
Chris_Atkinson
Employee
Employee

Is graceful restart configured in this scenario? Please see sk100499.

0 Kudos
nabs_nabs
Contributor

Hi @Chris_Atkinson ,

GR is not enabled. 

Does it means that if i enable it,  the newly master will forward packet even if the bgp session is not yet bring up ?

 

0 Kudos
Chris_Atkinson
Employee
Employee

Yes it's recommended in such scenarios.

Sets the Check Point system to maintain the forwarding state advertised by peer routers even when they restart.

This minimizes the negative effects caused by the restart of peer routers.

0 Kudos
nabs_nabs
Contributor

Ok. Thank you very much for your lights @Chris_Atkinson .

 

0 Kudos
Chris_Atkinson
Employee
Employee

Not a problem.

Also Note - In a cluster, you must select a Router ID and make sure that it is the same on all cluster members.