On Thu, May 12, 2016 at 11:36 AM, Christos Zoulas <chris...@astron.com> wrote: > In article > <cakryomjiy13pnopc7hrv4e0gjnju5yejcedvacnbxphe9ez...@mail.gmail.com>, > Ryota Ozaki <ozak...@netbsd.org> wrote: >>On Wed, May 11, 2016 at 4:18 AM, matthew green <m...@eterna.com.au> wrote: >>> Thor Lancelot Simon writes: >>>> I do not think you should do any extra work to support kvm grovelling. >> >>Actually I do nothing other than leaving the original list >>as it is. I thinks it's reasonable as what we do now. >> >>> >>> existing tool functionality should not be broken, however. the >>> netstat -i groveller code should be either updated to the new >>> method or converted to a real sysctl, though the current work >>> around would be ok for now. simply breaking netstat -i is not >>> an acceptable solution. >> >>Yeah, someone should sweep kvm(3) users in the future... >> >>BTW netstat already has a feature to retrieve information via >>sysctl but its man page warns that it lacks some information >>compared to kvm(3). Do anyone know what's lost? > > grep the netstat code for XXX
Thanks. Hmm, if_snd.ifq_drops. We now have percpu RX queues and will have multiple TX queues (in each device driver, not ifnet though). We would need to revise how to show queue information of interfaces. ozaki-r
