On Wed, Aug 05, 2015 at 10:01:26AM -0700, Justin Pettit wrote: > > > On Aug 5, 2015, at 3:11 AM, Wenyu Zhang <wen...@vmware.com> wrote: > > > > > >> On Aug 4, 2015, at 12:17 PM, Ben Pfaff <b...@nicira.com> wrote: > >> > >> On Tue, Aug 04, 2015 at 04:05:58AM +0000, Wenyu Zhang wrote: > >>> In ipfix and sflow, the sampling rate can be set as 1, which means > >>> 100%. And in order to make 100% packets sampled, I work out a patch > >>> in datapath, which is for review on net-next. And the same change > >>> seems need to be added into lib/odp-execute.c too. Do you mean that it > >>> is not necessary here? > >> > >> I mean that we should modify the IPFIX and sFlow sampling code so that > >> in the 0% and 100% sample rates cases it doesn't generate a sample > >> action; there isn't any value in it. > >> > >> This situation is a little different from the kernel situation, since we > >> know exactly what actions can be fed to the userspace datapath in > >> odp-execute.c, since the actions that it executes are generated by > >> another part of the same source tree. The kernel datapath, however, > >> should be equally capable of handling actions from sources other than > >> the Open vSwitch userspace. > > > > It is OK that sample action shouldn’t be generated when the sampling > > rate is 0%. But I wonder why we shouldn’t generate sample action > > when the sampling rate is 100%. It should happen in kernel datapth. > > Or you means that the 100% sampling rate case won’t happen in the > > userspace datapath? > > Ben's on vacation, so I'll just have to guess what he meant. I > suspect his thinking was that you'd just call the actions that would > be called through "sample" directly. Since the run 100% of the time, > there's no need to call them through the sample action.
Yes, that's what I meant. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
