> I'm willing to believe that.  Anyway, it can't cause much extra work,
> since it only has an effect for facets that have stats that haven't
> previously been pushed.  (I guess you verified that this can happen in
> cases where we don't already push the stats?  I can believe that, but I
> don't know off-hand, and I did not verify it myself.)

Yes, as a result of commit 5c0243a, when we handle a miss we credit
the statistics to the facet without pushing it to avoid the additional
action translation.  At any rate, as part of my wildcarded facet
series, we won't even be doing pushes in update_stats().  Hopefully
this measure will avoid a bit of confusion.

Ethan

>
> Looks good.
>
> Thanks,
>
> Ben.
_______________________________________________
dev mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/dev

Reply via email to