I also put this question on the General forum but since no one has an answer for it I'm going to try here :

I've 2 ISP connected to an RB2011LS-IN both working correctly on failover, using diferent distances on the route table, however if I've a connection already established until it ends it wont switch to the secondary route in case the first one died.

I know is a delay of about 10 seconds but after configuring and seeing working a VRRP setup it looks too much.

A bit more of data :

WAN1 : 192.168.100.1/24 Distance 1
WAN2 : 192.168.200.1/24 Distance 2
LAN : 192.168.0.1/24

Now if I do a continuous ping from say 192.168.0.10 to 8.8.8.8 and I pull the WAN1 cable, the connection won't switch to the next route (in this case 192.168.200.1) until I stop the ping and wait around 10 seconds for the connection to be drop.

So the question remains, is it posible to improve this or is this the way it works ?

Thanks in advance.

Deleted because not related.

No, I've 1 Lan -> 2 MT -> 2 WAN, the problem is that if the WAN connection with distance 1 gets down, until the connection is severed it will try to get out through the same route.

Deleted because not related.

I've 2 ISP both connected to 2 RB2011 and this two also conected to the same Lan with VRRP on the three interfaces.

The problem is that if I do a ping from the Lan to the WAN and what I lose is the WAN connection with Distance 1 then, I've to drop the connection, wait around 10s and then start again. However is what a lose is the master RB2011 all work without interruption.

Deleted because not related.

Yes, I've tried changing the default route when I disconect the ISP1 link but with no joy.

I've setup an event on the Slave RB on the VRRP3 interface that onces it becomes master makes the other VRRP interfaces also master setting the priotity to 250.

Now the problem is that if using the Slave or the Master I lost link with the ISP1 during a connection from the Lan the default route changes but the connection is not reset.

Deleted because not related.

Can you do a test using scenario 2 ?

Open a cmd and type "ping -t 8.8.8.8" for a continous ping. Once it has started pull from the RB the cable of the ISP1 and see what happends.

The desired effect is that the RB automaticly detects that the ISP1 path is down and rerouted the connection throught ISP2 but it doesn't do it.

Deleted because not related.

I've just figured a way to do this using netwatch.
With that the connection is drop and I have 2-3 packets lost at most. This probably is not the best way to do it but it works.

As for the questions, this is for a medium business that have currently 5 WAN connections (2 Fiber , 3 ADSL), and the goal which I've almost reach is that all the equiment on the client side is redundant (I've now two RB2011UAS-RM installed) using VRRP in both sides (WAN/LAN).

Deleted because not related.

Yeah, I know that as long as it works that's what matter but I don't think that the solution for this is to disable and enable the interface that holds the route.

In any case, now I have to check the VPN options since in this routers I use OVPN and every time I restore a backup I lost the certificates which makes a copy of the master configuration on the backup useless (I've a script that on start determines if the RB is the master or slave and changes the identity, IP's and VRRP priorities. )