13:00:35 <moshele> #startmeeting sriov 13:00:36 <openstack> Meeting started Tue May 17 13:00:35 2016 UTC and is due to finish in 60 minutes. The chair is moshele. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:37 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:00:39 <openstack> The meeting name has been set to 'sriov' 13:00:39 <moshele> hi 13:00:46 <lbeliveau_> hi 13:00:47 <gjayavelu> hi 13:00:48 <lennyb> moshele hi 13:01:21 <moshele> let get started 13:02:05 <moshele> #topic Improving SR-IOV/PCI Passthrough CI 13:02:39 <moshele> regarding CI I know lennyb is still working on the Mellanox CI to move it to containers 13:03:19 <moshele> anyone have anything to add or we can move on 13:03:47 <lbeliveau_> when do you think this is going to be done and working ? 13:04:02 <moshele> end of this week 13:04:07 <lbeliveau_> awesome ! 13:04:44 <moshele> #topic Documentation 13:05:13 <moshele> so it seem that all the SR-IOV/NUMA/PCI doc update are merged 13:05:20 <lbeliveau_> yes 13:05:44 <moshele> thanks to lbeliveau_ and sfinucan 13:06:02 <lbeliveau_> the only think that I still need to do is to document the feature taht ndipanov did for pf and vf 13:06:37 <moshele> beliveau: yes I will add this in our TODO 13:07:27 <lbeliveau> also, I think the integration of pci passthrough with neutron is done, but I have never tested that 13:07:34 <lbeliveau> this would need to be documented as well 13:07:52 <moshele> lbeliveau: it done I review the code on the neutron side 13:08:11 <moshele> if I have time I will try to test it 13:08:52 <lbeliveau> ok, I should be available also to work on this later this week 13:09:03 <moshele> ok I added pf passthough doc task in the agenda https://etherpad.openstack.org/p/sriov_meeting_agenda 13:09:36 <moshele> anything else or we can move on? 13:09:38 <gjayavelu> i had one question on documentation. Currently, the step to create pci_alias is missing on create virtual functions page: http://docs.openstack.org/mitaka/networking-guide/adv-config-sriov.html#create-virtual-functions-compute 13:09:55 <gjayavelu> should we add that ? 13:10:51 <lbeliveau> yeah, it could be done with pci_alias as well, but you wouldn't get your device associated with a neutron port 13:11:05 <lbeliveau> but you are right, we should clarify 13:11:06 <moshele> you can put a note 13:11:18 <moshele> I don't think anyone will use it like this 13:11:18 <gjayavelu> ok 13:11:58 <gjayavelu> because I found that info on the new page for pf https://review.openstack.org/#/c/312234/14/doc/admin-guide/source/compute-pci-passthrough.rst 13:12:22 <gjayavelu> so thought would be good to keep it consistent 13:13:12 <lbeliveau> yeap good point, we can add that and clarify 13:13:19 <moshele> I agree 13:14:04 <moshele> #topic bug fixes 13:14:04 <gjayavelu> i'll add it. 13:14:22 <moshele> gjayavelu: and add you patch to the agenda so we can review ti 13:14:23 * mriedem joins late 13:14:36 <lbeliveau> hi mriedem 13:14:46 <mriedem> o/ 13:14:57 <moshele> hi mriedem 13:15:18 <moshele> so my resize patch https://review.openstack.org/#/c/307124/ 13:16:19 <moshele> I fix the issue of clearing all the allocated pci device after confirm migration 13:16:44 <moshele> so now I am freeing only the old pci devices 13:17:10 <lbeliveau> cool, I'll test it in my testbed later today 13:17:10 <moshele> lbeliveau: I test it manually can you run also some tests on it 13:17:46 <moshele> Intel PCI is failing so I can see if it passes the resize test 13:17:49 <lbeliveau> I'll also test cold migration with it 13:18:02 <moshele> yonglihe: around? 13:18:13 <lbeliveau> have they fixed their test code as I suggested ? 13:18:29 <moshele> lbeliveau: I don't know 13:18:50 <lbeliveau> let me have a quick look at their git 13:18:51 <moshele> they are not sharing the tests 13:20:30 <moshele> mriedem: I update the https://etherpad.openstack.org/p/sriov_meeting_agenda with patches that are ready for core review 13:20:38 <lbeliveau> https://github.com/intel-hw-ci/Intel-Openstack-Hardware-CI/commit/21cd67b3c72592044d1c5eda8333832d9631e08f 13:21:07 <moshele> cool they fix there resize test 13:21:10 <mriedem> do we have a section like that in the overal nova newton review priorities etherpad? 13:21:24 <mriedem> https://etherpad.openstack.org/p/newton-nova-priorities-tracking 13:22:27 <mriedem> ah we do, L168 13:22:52 <moshele> mriedem: I will update the https://etherpad.openstack.org/p/newton-nova-priorities-tracking with the paches 13:23:58 <moshele> anything else on bugs? 13:24:35 <lbeliveau> no, I think the next one will be to fix cold migration 13:24:56 <lbeliveau> I don't know if sfinucante has more numa bugs to review 13:25:42 <moshele> there is still one which I have him -1 13:25:57 <moshele> #topic SR-IOV Specs 13:26:23 <moshele> anyone want to talk about their specs? 13:27:00 <lbeliveau> I still need to fix mine, haven't had much cycles to put on that lately, will do for next week 13:27:22 <gjayavelu> moshele: i added one to agenda only yesterday, may be i could move it for next week to give others more time 13:27:57 <moshele> gjayavelu: yes I review yours had some questions 13:27:57 <lbeliveau> gjayavelu: I will review for next week 13:28:12 <gjayavelu> thanks 13:28:54 <moshele> anything else? 13:29:01 <mriedem> yeah 13:29:04 <mriedem> on this is_new patch https://review.openstack.org/#/c/310990/ 13:29:14 <mriedem> what was is_new used for? what was the hack? 13:29:34 <mriedem> i'm trying to figure out what was setting new=True in the db 13:30:02 <moshele> as far as I know nothing 13:30:57 <moshele> just a sec let me look for Nikola commet 13:32:18 <mriedem> https://review.openstack.org/#/c/310990/4/nova/objects/instance_pci_requests.py@122 is where it's set 13:33:13 <moshele> see Nikola's comment on this https://review.openstack.org/#/c/301307/ 13:34:28 <mriedem> this? https://review.openstack.org/#/c/301307/3/nova/compute/resource_tracker.py 13:34:46 <moshele> yes 13:35:39 <mriedem> and the new_ and old_ he's talking about is what you're doing in your migration patch 13:35:47 <mriedem> to store the pci_requests old/new in the migration context 13:35:55 <moshele> yes 13:36:12 <mriedem> why doesn't https://review.openstack.org/#/c/310990/ come after your migration fix then? 13:37:48 <moshele> I can make a dependency, but because it is broken and the is_new is never set anyway I don't think it is matter 13:39:15 <mriedem> moshele: ok, i left some comments in the is_new patch 13:39:17 <mriedem> see what you think 13:39:25 <mriedem> after the meeting is fine 13:39:32 <moshele> ok 13:39:43 <moshele> anything else? 13:40:05 <moshele> ok thanks everyone 13:40:15 <moshele> #endmeeting