Neutron ovs config instructions are not clear
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
openstack-helm |
Fix Released
|
Undecided
|
Artur Korzeniewski |
Bug Description
values.yaml indicates, we can leave tunnel if not vxlan, instead it should be mandatory.
# External interface will be automatically added to external_bridge. Default is null.
# Tunnel interface will be used for VXLAN tunneling. Default is null, with
# fallback mechanism to search for interface with default routing.
external: null
tunnel: null
If left null the "local_ip" never gets assigned and neutron fails to deploy
{{{
#_neutron-
tunnel_
if [ -z "${tunnel_
# search for interface with default routing
tunnel_
fi
# determine local-ip dynamically based on interface provided but only if tunnel_types is not null
IP=$(ip a s $tunnel_interface | grep 'inet ' | awk '{print $2}' | awk -F "/" '{print $1}')
cat <<EOF>/
[ovs]
local_ip = $IP
EOF
}}}
The general idea is that if tunnel_interface is null, then tunnel interface should be discovered by line: interface= $(ip r | grep default | grep -oP '(?<=dev ).*')
tunnel_
What is the tunnel_interface set to when you run it by hand in your setup, where ovs-agent is failing to start?
Can you paste logs from ovs-agent pod?