This is something I have been doing as a temporary workaround (the code
wrapped inside ifdef ANUP_EXTENSION). But, I am not sure if the aim was to
handle this somewhere else, and I am missing something.
features.capabilities = (OFPUTIL_C_FLOW_STATS | OFPUTIL_C_TABLE_STATS |
OFPUTIL_C_PORT_STATS | OFPUTIL_C_QUEUE_STATS);
#ifdef ANUP_EXTENSION
/* advertise support for group stats in features reply */
protocol = ofconn_get_protocol(ofconn);
version = ofputil_protocol_to_ofp_version(protocol);
switch(version) {
case OFP11_VERSION:
case OFP12_VERSION:
case OFP13_VERSION:
features.capabilities |= OFPUTIL_C_GROUP_STATS;
default:
break;
}
#endif
On Mon, Oct 20, 2014 at 3:25 PM, Anup Khadka <khadka...@gmail.com> wrote:
> Seeing this in OVS 2.1 encounter (but looks like the code is same now too)
>
> The group stats bit is set to false in capabilities inside
> OFPT_FEATURES_REPLY.
>
> Is there a reason why that's done, despite the fact that there is already
> support for group stats?
>
> Looking at the git source right now (function: handle_features_request,
> file: ofproto.c), and I still see:
> features.capabilities = (OFPUTIL_C_FLOW_STATS | OFPUTIL_C_TABLE_STATS |
> OFPUTIL_C_PORT_STATS | OFPUTIL_C_QUEUE_STATS);
>
> Why isn't OFPUTIL_C_GROUP_STATS included there?
>
> Thanks,
> Anup
>
>
>
_______________________________________________
discuss mailing list
discuss@openvswitch.org
http://openvswitch.org/mailman/listinfo/discuss
