On Tue, May 8, 2018 at 11:21 AM, Mark Michelson <[email protected]> wrote: > > On 05/08/2018 01:26 PM, Han Zhou wrote: >> >> >> >> On Tue, May 8, 2018 at 8:53 AM, Ben Pfaff <[email protected] <mailto: [email protected]>> wrote: >> > >> > On Thu, May 03, 2018 at 07:30:56PM -0700, Han Zhou wrote: >> > > Most mc related flows are added only if the mc is related to local >> > > datapath, but there is a flow added before the local datapath >> > > check, out of the function consider_mc_group(). This patch fixes it. >> > > >> > > Signed-off-by: Han Zhou <[email protected] <mailto:[email protected]>> >> > >> > Thanks for the patch. >> > >> > Is there anything actually group-specific about this flow? That is, can >> > we just have a single flow of this type instead of one per multicast >> > group? >> > >> Good point! After taking a look at the original code, it seems the flag MLF_LOCAL_ONLY doesn't have to be used for multicast, although it was introduced for multicast use case (IPv6 RA). I think we can extract it out as a single flow per HV, removing the match condition for mc tunnel key. >> @Mark, could you confirm this, in case we missed anything? >> >> Thanks, >> Han > > > Yes, that will work just fine. Seems like a good improvement to me. > Thanks for the work, guys. > > Mark
Thanks for confirming! Here is the new patch to address it. https://mail.openvswitch.org/pipermail/ovs-dev/2018-May/346736.html (The current patch is now abandoned) Thanks, Han _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
