Skip to main content
jjdope
Staff
Staff
March 23, 2026

Technical Tip: Reducing BGP convergence delay and packet loss during HA failover using route-ttl and stale routes

  • March 23, 2026
  • 0 replies
  • 677 views
Description This article describes an issue where significant packet loss and delayed BGP convergence occur during FortiGate HA failover, and how to mitigate it using HA route-ttl and BGP stale route retention (graceful restart).
Scope FortiGate.
Solution

In HA failover scenarios involving BGP, the following behavior can commonly be observed, especially in environments using overlay VPNs:

  • Temporary packet loss during failover events.
  • Delay in BGP route convergence.
  • Traffic disruption while routing protocols re-establish adjacencies.
  • This can occur even when session pickup is enabled, as traffic forwarding depends on routing convergence, which may take additional time after failover.

 

Failover process:

During failover, The primary unit goes down and Secondary unit takes over, BGP sessions are re-established and routes are relearned. If the Traffic is dropped until BGP reconvergence is completed, then the below config changes will resolve the issue. See Technical Tip: Controlling how HA synchronizes routing table updates.

 

Step 1: Configure HA route-ttl.

The route-ttl parameter allows FortiGate to retain routes temporarily after failover.

 

Configuration:

 

config system ha

    set route-ttl 210

end

 

Step 2: Enable BGP graceful restart and stale routes.

 

To further improve convergence, enable stale route retention on BGP neighbors.

 

Global configuration:

 

config router bgp

    set graceful-restart enable

    set graceful-restart enable

    set graceful-restart-time 120

    set stale-route enable

   end

 

Per-neighbor configuration:

 

config router bgp

    config neighbor

        edit <neighbor-ip>

            set capability-graceful-restart enable

        next

    end

   end

 

  • Routes are retained for the configured route-ttl seconds and prevents immediate route removal.
  • This allows BGP sessions to re-establish without traffic loss.
  • Neighbor devices retain routes during session interruption and traffic continues using stale routes.
  • HA failover can introduce routing gaps if routes are flushed prematurely, route-ttl ensures temporary route retention.
  • BGP graceful restart allows control-plane recovery without data-plane impact, combining both provides near hitless failover.

Notes:

  • Set route-ttl higher than expected BGP convergence time. For example, if the reconvergence takes 200 seconds, then configure route-ttl to 210 seconds.
  • Some ISPs might not support graceful-restart capability.

 

Related articles: