I have a scenario I have been testing with what appears to be inconsistent results, and I was wondering if anyone could help me understand why.
Please see the attached diagram.
* HQ is the company head quarters (hub site), and has dual connections to SITE1 and SITE2 via GRE tunnels (/30) (in IPSEC) over separate internet connections.
* SITE1 and SITE2 also have a single connection, and also a GRE tunnel between them.
* OSPF costs are configured as such that the BLUE links are primary, RED links are secondary
What I observed when running a continual ping from HQ to SITE1:-
* Normally traffic goes over A1
* When A1 between HQ and SITE1 goes down, A2 takes over and traffic resumes after approx 30-40 seconds (dead-interval?)
However, if A2 is already down it would instantly go via A3 (via primary link to SITE2), and then C - hardly any disruption to ping.
So I do not understand why it takes longer to fail onto A2 (a direct link), than traversing another site (link A3, then link C).
All routers are RouterBOARD 3011UiAS running 6.49.10
HQ config for OSPF:-
Code: Select all
/routing ospf instance
set [ find default=yes ] redistribute-connected=as-type-1 redistribute-static=as-type-1 router-id=192.168.91.50
/routing filter
add action=accept chain=ospf-out comment=Internal prefix=192.168.0.0/16 prefix-length=16-32
add action=discard chain=ospf-out comment=Discard
/routing ospf interface
add comment=A1 cost=200 interface=gre-tunnel-A1 network-type=point-to-point
add comment=A2 cost=300 interface=gre-tunnel-A2 network-type=point-to-point
add comment=A3 cost=520 interface=gre-tunnel-A3 network-type=point-to-point
add comment=A4 cost=620 interface=gre-tunnel-A4 network-type=point-to-point
add passive=yes
/routing ospf network
add area=backbone comment=A1 network=192.168.90.4/30
add area=backbone comment=A2 network=192.168.90.104/30
add area=backbone comment=A3 network=192.168.90.12/30
add area=backbone comment=A4 network=192.168.90.112/30
add area=backbone comment=Loopback network=192.168.91.50/32