[openstack-dev] [nova] Continued reduction in NFV tech debt - input?

Matt Riedemann mriedem at linux.vnet.ibm.com
Thu Dec 1 13:57:45 UTC 2016

On 12/1/2016 4:51 AM, Stephen Finucane wrote:
> nova has worked hard over the last few cycles to identify and reduce
> the amount of tech debt we're carrying. One such item identified early
> on was the lack of automated testing and upstream documentation for NFV
> features, such as CPU pinning, realtime, hugepages, SR-IOV, and PCI
> passthrough. We discussed this at the Austin (M) summit and have been
> slowly chugging away at it since through a combination of weekly SR-IOV
> meetings [1] and changes to the docs [2][3][4][5], code and in- [6] and
> out-of-tree tests [7].
> This work will continue for this cycle, however, we'd like to ask is
> there any of these features where developers or operators can still
> identify significant holes in either validation or (upstream)
> documentation? If so, please give us a shout. This is something we'll
> probably double back on in a PTG session, but we'd like to make full
> use of the few months between now and then to really build on these
> features.
> I'd also like to take the opportunity to promote the weekly SR-IOV
> meeting. We regularly address all things NFV at this meeting, so if
> you've ideas on how to improve or build upon these features, this is as
> good a place to bring said ideas up.
> Cheers,
> Stephen (sfinucan)
> [1] http://eavesdrop.openstack.org/#SR-IOV/PCI_Passthrough_Meeting
> [2] http://docs.openstack.org/networking-guide/config-sriov.html
> [3] http://docs.openstack.org/admin-guide/compute-pci-passthrough.html
> [4] http://docs.openstack.org/admin-guide/compute-cpu-topologies.html
> [5] http://docs.openstack.org/admin-guide/compute-huge-pages.html
> [6] https://github.com/openstack/nova/tree/master/nova/tests/functional
> /libvirt
> [7] https://github.com/openstack/intel-nfv-ci-tests
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Stephen, thanks for asking for input. You might want to cross-post to 
the openstack-operators mailing list if you're looking for operational 
input on these features in nova. It's fun to fish either way. :)

One immediate thing is we approved this spec for Ocata:


But there is no code up for review yet. I'm getting a bit antsy about 
blueprints that don't have code yet for Ocata, and this change seems 
pretty straight-forward, so it would be good to get some people driving 



Matt Riedemann

More information about the OpenStack-dev mailing list