On Wed, Apr 13, 2016 at 05:17:13PM -0700, Ben Pfaff wrote: > On Wed, Apr 13, 2016 at 05:06:16PM -0700, Joe Stringer wrote: > > On 13 April 2016 at 16:12, Ben Pfaff <b...@ovn.org> wrote: > > > On Tue, Apr 12, 2016 at 10:56:50PM -0700, Joe Stringer wrote: > > >> On 12 April 2016 at 21:13, Ben Pfaff <b...@ovn.org> wrote: > > >> > On Tue, Apr 12, 2016 at 11:38:38AM -0700, William Tu wrote: > > >> >> Should we expose "truncate" to the ovs-ofctl action list? > > >> >> > > >> >> I was thinking about this ovs-ofctl syntax: > > >> >> actions='output(max_len=64, port=1), output:2' > > >> >> > > >> >> then at datapath it translates to actions > > >> >> truncate(64), output(1), output(2) > > >> >> > > >> >> So 64B to port1, and 100B to port2. > > >> > > > >> > I think that's OK. > > >> > > > >> > Pravin or Joe, do you have an opinion? > > >> > > >> Seems fine. > > >> > > >> As an aside, it might be worth creating some tests that output to a > > >> bond port to ensure that case works, in addition to the existing > > >> cases. > > > > > > I don't know what that means; there are no "bond ports" at the OpenFlow > > > level. > > > > I meant, to configure a bond and use that port as the output, to check > > for corner cases where the datapath flows break up the output across > > two flows (recirc + actual output) > > Oh, I guess that could be possible if one used "output(max_len=64, > port=NORMAL)". It's quite a corner case!
Also I wonder whether truncated output to FLOOD or ALL should be well-defined. We could just forbid them I suppose. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev