Ordinarily, if restart openvswitch, we should restart neutron-openvswitch-agent as well. For instance, for some RPM release, it has systemd services which will restart neutron-openvswitch-agent after openvswitch is restarted implicitly. So, for container based deployment, I'm not sure if there is such similar mechanism. It's better to restart neutron-openvswitch-agent container after restart ovs manually to avoid some flows missing, besause restart ovs is mean to re-process all ports for neutron-openvswitch-agent.
Ordinarily, if restart openvswitch, we should restart neutron- openvswitch- agent as well. For instance, for some RPM release, it has systemd services which will restart neutron- openvswitch- agent after openvswitch is restarted implicitly. So, for container based deployment, I'm not sure if there is such similar mechanism. It's better to restart neutron- openvswitch- agent container after restart ovs manually to avoid some flows missing, besause restart ovs is mean to re-process all ports for neutron- openvswitch- agent.