Router failover issue with external network failure

[Aymen Frikha]

After testing some failover scenarios for the HA router when we use a disaggregated architecture (separated network nodes), we detected a router failover issue. The following are the scenarios tested:

1- When we completely disable the interfaces of the network node (that contains the active router) that are related to the Overlay Geneve network: The router failover succeed

2- When we completely disable the interfaces of the network node (that contains the active router) that are related to the external network: The router failover Failed

Is there any solution to cover the second failed scenarios ?

[Aymen Frikha (aym-frikha)]

subscribed ~field-high

[Frode Nordahl (fnordahl)]

> 2- When we completely disable the interfaces of the network node (that contains the active router) that are related to the external network: The router failover Failed

> Is there any solution to cover the second failed scenarios ?

For this to work there needs to be some sort of liveliness detection between the OVN virtual router and the default gateway. OVN itself has support for this through the use of BFD to monitor next hop of static routes.

Support for this has been available in OVN for some time [0][1], however support for making use of it it was only recently added to OpenStack [2].

What versions are in use here?

0: https://patchwork.ozlabs.org/project/ovn/list/?series=223388&state=*
1: https://patchwork.ozlabs.org/project/ovn/list/?series=393345&state=*
2: https://bugs.launchpad.net/neutron/+bug/2002687

[Launchpad Janitor (janitor)]

[Expired for charm-ovn-dedicated-chassis because there has been no activity for 60 days.]