-
Bug
-
Resolution: Done
-
High
-
None
There is a code change [0] in NeutronvpnManager.java which checks if rdList size < number of
nexthops configured for an extra route. (During neutron router change event)
This is checked when vpn interface(config) has to be updated with extra route adjacency.
In case when router is not associated with bgp, this check will fail.
This prevents addition of extra route adjacency in the vpn interface config DS.
And thus ECMP [1][2] code doesn’t get invoked.
if (rdList.size() < nbNextHops) {
neutronvpnAlarm.raiseNeutronvpnAlarm(typeAlarm, detailsAlarm.toString());
LOG.error("there are too many next hops for prefixe in vpn {}", vpnId);
routesError.add(route);
}
Was there any particular reason for adding this check (inside checkAlarmExtraRoutes method).
[0] [ https://git.opendaylight.org/gerrit/#/c/60881/|https://git.opendaylight.org/gerrit/#/c/60881/]
[1] [ https://git.opendaylight.org/gerrit/#/c/48915/|https://git.opendaylight.org/gerrit/#/c/48915/]
[2] [ https://git.opendaylight.org/gerrit/#/q/status:merged+project:netvirt+branch:master+topic:l3vpn_ecmp|https://git.opendaylight.org/gerrit/#/q/status:merged+project:netvirt+branch:master+topic:l3vpn_ecmp]