[openstack-dev] [OpenStack-dev][neutron] can't notify the broadcast fdb entries

Zang MingJie zealot0630 at gmail.com
Wed Jul 9 13:15:35 UTC 2014


Hi:

We are encountered the same problem here. some of our ovs-agent
haven't received any fdb entry after a restart

To solve the problem I'm going to add a rpc call to l2pop mechanism
driver, when triggered, the l2pop send all fdb entries to the agent.
The agent call the driver while starting.

On Wed, Jul 9, 2014 at 8:56 AM, Yangxurong <yangxurong at huawei.com> wrote:
> Hi, folks,
>
>
>
> Now , in order to improve broadcast fdb entries traffic, we do some
> restriction:
>
>
>
> if agent_active_ports == 1 or (
>
>                 self.get_agent_uptime(agent) <
> cfg.CONF.l2pop.agent_boot_time):
>
>             # First port activated on current agent in this network,
>
>             # we have to provide it with the whole list of fdb entries
>
>
>
>          but this restriction will bring about new issue (fdb entries could
> not be notified)base on the follow cases:
>
> 1.       Distribute deploy multiple neutron server and bulk create ports in
> short order, namely, agent_active_ports will be more than 1, so fdb entries
> could not be notified and thus lead to fail to establish the tunnel port.
>
> 2.       The time of booting Ovs-agent is more than
> cfg.CONF.l2pop.agent_boot_time, so, fdb entries could not be notified and
> thus lead to fail to establish the tunnel port.
>
>
>
> IMO, the restrictions are not perfect.
>
> Any good thought, looking forward to your response.
>
>
>
> Thanks,
>
> XuRong Yang
>
>
>
>
> _______________________________________________
> OpenStack-dev mailing list
> OpenStack-dev at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>



More information about the OpenStack-dev mailing list