[openstack-dev] Multiple API/RPC process behavior

Miguel Angel Ajo majopela at redhat.com
Wed Jul 2 10:52:29 UTC 2014


I'd recommend you to go for ml2, as openvswitch alone
plugin will be removed during Juno, making an upgrade
path harder on your side.

I suppose there could be a bug/noimplementation for
multiplerpc workers within openvswitch plugin, but
as it's deprecated there are high chances that
it won't be fixed.

On 07/02/2014 09:18 AM, tfreger at redhat.com wrote:
> Hello,
>
> I've installed the latest version of RHOS5, set multiple API\RPC workers
> within neutron.conf file to 8 (the amount of cores of the host)
>
> After I ran "openstack-service restart" I've got different result for
> ML2 and Openvswitch deployments.
>
> In ML2 deployment all 17 (1-parent and 16 child processes)
> neutron-server processes exist as was expected.
> In Openvswitch deployment only 9 processes of API are spawned, RPC
> process didn't spawn.
>
> Should be any difference between those two installations?
>



More information about the OpenStack-dev mailing list