13:03:37 <moshele> #startmeeting sriov
13:03:38 <openstack> Meeting started Tue Nov 29 13:03:37 2016 UTC and is due to finish in 60 minutes.  The chair is moshele. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:03:39 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:03:41 <openstack> The meeting name has been set to 'sriov'
13:03:43 <lbeliveau> o/
13:03:45 <wznoinsk> o/
13:03:45 <moshele> hi
13:03:58 <moshele> sorry for being late
13:04:05 <edand> hi
13:04:22 <moshele> #topic Improving SR-IOV/PCI Passthrough CI
13:04:52 <moshele> lbeliveau: you revert patch is not merged yet to tempest
13:04:59 <moshele> it has only one +2
13:05:14 <lbeliveau> yes, will try to ping cores today
13:05:29 <lbeliveau> does the cold migration one has been integrated in mellanox CI ?
13:05:40 <moshele> I think so
13:05:40 <lbeliveau> I don't get mellanox CI to review my patches anymore
13:05:57 <moshele> which patches?
13:06:18 <lbeliveau> this one for e.g. https://review.openstack.org/#/c/370374/
13:06:58 <moshele> lennyb can you check this ^
13:07:12 <lennyb> moshele: yeap
13:07:34 <moshele> anything else on CI?
13:07:52 <lbeliveau> wznoinsk: any update on the intel CIs ?
13:08:13 <moshele> wznoinsk: I saw you are commenting on neutron as well, right?
13:08:18 <pkoniszewski> o/
13:08:45 <moshele> hi pkoniszewski
13:09:11 <moshele> let move on
13:09:37 <moshele> #topic Patches for Subteam Review
13:10:01 <moshele> lbeliveau: can you update all your new patches in https://etherpad.openstack.org/p/sriov_meeting_agenda
13:10:34 <lbeliveau> there is two new patches for fixing race conditions on migrate and migrate revert
13:10:36 <lbeliveau> https://review.openstack.org/#/c/370374/
13:10:43 <lbeliveau> https://review.openstack.org/#/c/400995/
13:10:49 <lbeliveau> that needs review
13:11:15 <moshele> ok I will take a look this week
13:11:33 <moshele> any other patches?
13:11:40 <lbeliveau> not from me
13:11:47 <wznoinsk> sry, lennyb moshele , no sriov ci in our lab yet and yes we vote on neutron but had a rough week and only now getting back to normal day
13:12:26 <moshele> wznoinsk: thanks for the update :)
13:13:04 <moshele> pkoniszewski: any progress with the live migration patches?
13:14:29 <moshele> pkoniszewski: we have multi node setup with the live migration patches. I want to test it to see what are the probelms
13:14:30 <pkoniszewski> sorry
13:14:40 <pkoniszewski> i haven't fixed the race condition
13:14:52 <moshele> but is it even working?
13:14:59 <pkoniszewski> I need to ask cores about their opinion on the fix that Chris provided
13:15:02 <pkoniszewski> it is working
13:15:15 <pkoniszewski> apart from this race condition that is a more like a corner case
13:15:29 <moshele> with macvtap port?
13:15:41 <moshele> or just NUMA?
13:15:52 <pkoniszewski> oh, i'm right now still focusing on resource tracking and NUMA
13:15:55 <pkoniszewski> no macvtap ports yet
13:16:14 <pkoniszewski> i need to do some more testing with NUMA anyway
13:16:25 <moshele> pkoniszewski: I see
13:17:18 <moshele> I would say let merge the basic functionality if it working and open bugs on the race condition
13:17:39 <pkoniszewski> that might be a good way to move forward
13:17:44 <moshele> that what we did with the cold migration, we still have races there
13:17:44 <pkoniszewski> given how long this patch is up for review
13:17:48 <pkoniszewski> and it covers a lot of cases
13:18:19 <moshele> I am not core, but I will +1 if the basic is working :)
13:18:31 <lbeliveau> anyway if you try to fix too many things in a patch, cores will most likely ask you to split it
13:18:52 <moshele> yes that right
13:19:18 <pkoniszewski> yeah, but Jay already said that it doesn't make sense to split this patch as he couldn't even find how the patch can be split
13:19:34 <pkoniszewski> so the resource tracking part works, that's sure
13:20:01 <pkoniszewski> I still have some issue with setting up SR-IOV, I will do some more NUMA testing
13:20:31 <moshele> pkoniszewski: if it work with NUMA that good start, we can fix the other later
13:20:36 <pkoniszewski> by setting up SR-IOV I mean that it's my env issue
13:20:39 <pkoniszewski> not the patch
13:20:56 <lbeliveau> pkoniszewski: let me know, I can help with that
13:21:35 <pkoniszewski> ok, thanks, once I check all the stuff I can I will ask for help
13:21:47 <moshele> pkoniszewski: can you rebase the patch and add TODO comment on the race conditions
13:21:53 <pkoniszewski> sure
13:23:07 <moshele> so after the new patchest let review it and if it ready I will ask jaypipes to review it again
13:23:16 <pkoniszewski> okay
13:23:49 <moshele> #topic Specs for Review
13:24:10 <moshele> none of the specs got approved for SR-IOV
13:24:28 <moshele> except the whitelist regex
13:24:29 <lbeliveau> :(
13:24:48 <moshele> so I was thinking to do a bi weekly meeting
13:25:01 <jaypipes> moshele, pkoniszewski: morning guys.
13:25:19 <moshele> hi jaypipes
13:25:41 <jaypipes> moshele, pkoniszewski: which reviews do you need most urgently that I have not yet done?
13:26:55 <jaypipes> moshele, pkoniszewski: as you may know, dan berrange is no longer working on Nova, so we need more than ever to get our PCI/SRIOV review load under control and maintain a prioritized list of patches in this team
13:27:36 <moshele> jaypipes: so first we need a subteam review but we are talking about the live migration calim patch https://review.openstack.org/#/c/244489/
13:27:49 <moshele> jaypipes: yes I saw it on the ML
13:28:22 <jaypipes> moshele: ok, I can review that (though it needs a rebase looks like)
13:28:31 <moshele> jaypipes: I don't think will have so match patches for review most of the pci spec rejected
13:29:09 <moshele> jaypipes: it not argent I want the sub team to review it first
13:29:11 <jaypipes> moshele: even though pci specs were rejected, I still very much will need your reviews on the PCI and SRIOV-related nested resource providers work.
13:29:18 <lbeliveau> moshele: there are still bug fixes in the pipeline
13:29:39 <pkoniszewski> yeah, I noticed that, I will try to focus some more on PCI/SRIOV related stuff
13:30:06 <moshele> jaypipes: yes  of course
13:30:25 <lbeliveau> jaypipes: when do you think you'll have something ready for review ?
13:30:49 <jaypipes> lbeliveau: https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:bp/nested-resource-providers :)
13:31:22 <jaypipes> lbeliveau: and spec is here: https://review.openstack.org/#/c/386710/
13:31:33 <lbeliveau> jaypipes: ok will have a look
13:31:50 <jaypipes> lbeliveau: will be pushing a patch today that adds child providers for PFs and NUMA nodes.
13:32:28 <moshele> jaypipes: are you planing to do the change that the compute node will use report them as nested resources? is this for this cycle?
13:32:42 <jaypipes> moshele: yes and yes
13:33:04 <moshele> jaypipes: ok cool
13:33:39 <moshele> anyone have anything to else add?
13:33:55 <sfinucan> nothing from me
13:34:16 <moshele> #endmeeting