vpn - gre tunnel - continuous pings require to keep tcp connection alive

[james.rohan]

Have a situation wherby a server and client exchange data on tcp ports 3299 (client) and > 1023 (server). They are hosted by different service providers with a gre tunnel serviced by static routing running between the two provider's border routers.

The tunnel is shared with other clients but only one complains of timeouts after approx 5 mins.To counteract this the application support people at the server end run continuous pings to the client which allows the traffic flow between server and client as normal.

The tunnel config is as follows

interface Tunnel99

bandwidth 64

ip address 193.95.x.x 255.255.255.252

no ip directed-broadcast

no ip route-cache cef

tunnel source Loopback0

tunnel destination 195.182.x.x

I wonder if anyone else has had experience of a similar fault on a gre tunnel or

if the fault is most likely elsewhere..

Many thanks..

[kylseverson]

GRE tunnels are usually used in combination with a dynamic routing protocol that keeps the tunnel up and active. If you are using static routes there is no traffic to keep the tunnels up when the applications are not transmitting.

You have two options:

1. Increase the isakmp/ipsec lifetimes.

http://www.cisco.com/en/US/products/sw/iosswrel/ps1831/products_configuration_guide_chapter09186a00800d981f.html#1001095

2. Setup keepalives (Best Option)

crypto isakmp keepalive (Number of seconds between keep alives) (Number of seconds between retries if keepalive fails)

[james.rohan]

Many thanks for the response.

We are using statics and this is a simple GRE tunnel (no ipsec).

We haven't explicitly configured timeouts but sh int tunnel indicates that

keepalives are set to 10 secs.

Since logging the call we have changed one of the tunnel end points and the

problem now seems to be resolved but we are none the wiser as to its original cause.