13:06:50 <ndipanov_> #startmeeting sriov 13:06:51 <openstack> Meeting started Tue Mar 1 13:06:50 2016 UTC and is due to finish in 60 minutes. The chair is ndipanov_. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:06:52 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:06:54 <openstack> The meeting name has been set to 'sriov' 13:07:02 <ndipanov_> awesome 13:07:30 <ndipanov_> ok so I thought we should talk about pending patches 13:07:45 <ndipanov_> and any bugs that are bad enough we wanna try and get them in for Mitaka 13:07:59 <ndipanov_> I thought we could use https://etherpad.openstack.org/p/sriov_meeting_agenda 13:08:19 <ndipanov_> to list the reviews that might be pending 13:08:41 <ndipanov_> I don't think a lot of bugfixes are going to realistically get merged this week 13:08:43 <lbeliveau> I have two pending, let me put them on the etherpad 13:08:44 <ndipanov_> as it's FF week 13:08:49 <ndipanov_> lbeliveau, please do 13:09:40 <lbeliveau> the first one is and easy one, you put +2 on it, I guess it just need another core to double check 13:10:28 <ndipanov_> yeah so let me re-arrange that :) 13:12:27 <ndipanov_> lbeliveau, ok I remember the first one - I know johnthetubaguy has looked at it before so we'll ping him once the FF craze is over 13:12:47 <lbeliveau> perfect thanks 13:15:12 <ndipanov_> anything else on the pending patches front? 13:15:33 <lbeliveau> I'm good 13:16:03 <vladikr> I have one 13:16:06 <vladikr> https://review.openstack.org/#/c/283198/4 13:16:19 <ndipanov_> vladikr, yeah I saw it - will try to review that one today as well 13:16:30 <vladikr> great, thanks 13:16:56 <ndipanov_> so there is also this https://review.openstack.org/#/c/242087/ 13:17:11 <ndipanov_> I've seen this refactoring before, and jaypipes mentioned it as well 13:17:12 <lbeliveau> vladikr: have you see my comment about if the fix needs to be applied anywhere else in the compute manager for cleanup ? 13:17:22 <ndipanov_> but haven't really looked at it 13:17:50 <vladikr> lbeliveau, oh, yes, sorry.. 13:18:04 <lbeliveau> vladdikr: just a question. no worries :) 13:18:20 <vladikr> lbeliveau, I was trying to look into that, but I don't see any other places where this would apply 13:18:20 <ndipanov_> anyone else looked at that? 13:18:35 <lbeliveau> I looked at it 13:19:52 <ndipanov_> ok lbeliveau I see you commented quite recently 13:20:17 <ndipanov_> is that one on track for FF? 13:20:45 <ndipanov_> lbeliveau, do you have any thoughts? 13:20:53 <ndipanov_> that you would like to share :) 13:22:13 <lbeliveau> Should ask jay pipes, I think this patch is part of multiple patches for remove pci stats, dunno if it's finished or not 13:22:41 <ndipanov_> lbeliveau, right, but what is your take on the work? 13:22:50 * ndipanov_ hasn't really looked at it tbh 13:23:48 <lbeliveau> I think it makes senses just to keep pci devices and get away from pci stats 13:24:17 <ndipanov_> lbeliveau, but won't that make us query ALL the pci devices ever in the scheduler 13:24:26 <ndipanov_> which is potentially 10s of thousands of rows? 13:25:27 <ndipanov_> I guess I should look at the series 13:26:37 <lbeliveau> it's part of the blueprint for uniform resource provides, haven't follow that closely and wasn't part of the discussions 13:27:59 <ndipanov_> ok I'll try to wrap my head around it today 13:28:29 <ndipanov_> I see that there is a bug that it fixes too - will look at that as well 13:29:34 <ndipanov_> just looking at the list of bugs there seems to be nothing critical other than resize which we know doesn't work 13:29:58 <lbeliveau> ndipanov: what do you want to do about this one https://review.openstack.org/#/c/242573, if you have more comments I can work on it today 13:30:05 <ndipanov_> I was going to look at it if I get a chance after I'm done with some live migration stuff I am trying to figure out 13:30:24 <lbeliveau> great 13:30:41 <ndipanov_> lbeliveau, yeah I have that on my list - I think I was happy with it mostly but it's a big patch so I didn't vote :( 13:30:55 <ndipanov_> moshele, https://bugs.launchpad.net/nova/+bug/1400574 <- can we mark that one as incomplete? 13:30:55 <openstack> Launchpad bug 1400574 in OpenStack Compute (nova) "Create VMs sometimes failure when use mellonax nic as SR-IOV" [Low,Confirmed] - Assigned to KaiLin (linkai3) 13:30:57 <lbeliveau> I understand, let me know 13:31:21 <moshele> yes 13:32:18 <moshele> ndipanov_: he didn't respond to my question 13:32:23 <ndipanov_> moshele, yeah 13:32:27 <ndipanov_> so I was looking at this 13:32:29 <ndipanov_> http://bit.ly/1QjJuVf 13:32:40 <ndipanov_> this is basically a search for pci in nova bugs 13:32:54 <ndipanov_> but that might not get all the bugs we care about 13:33:24 <ndipanov_> for example neutron integration 13:33:43 <jaypipes> lbeliveau: in progress 13:34:02 <ndipanov_> so if there are some bugs that you are aware of that we should bump in priority and try to get fixed in M - that would be awesome 13:34:10 <moshele> so here are all neutron sr-iov bugs https://bugs.launchpad.net/neutron/+bugs?field.tag=sriov-pci-pt 13:36:30 <moshele> ndipanov_: we should fix the migration issue so I guess lbeliveau https://review.openstack.org/#/c/242573 is important 13:37:07 <lbeliveau> right, after it get merged, I proposed myself to look at the resize issue 13:37:08 <ndipanov_> moshele, yeah I guess so - it's really annoying that it's there 13:38:42 <ndipanov> ok well if that's all we have 13:39:12 <ndipanov> let's get back to reviewing/coding and hopefully get these in soon 13:39:49 <ndipanov> we could add an sriov tag to bugs 13:39:55 <ndipanov> to make it easier to track 13:40:05 <ndipanov> I might do that and go back and tag useful bugs with it 13:40:44 <vladikr> ndipanov, looks like that series https://review.openstack.org/#/c/242087/ 13:41:00 <ndipanov> vladikr, yeah 13:41:03 <vladikr> ndipanov, completely removes all of the vf-pf stuff 13:41:08 <vladikr> :) 13:41:13 <ndipanov> lol nice 13:41:24 <ndipanov> back to the drawing board 13:41:31 <lbeliveau> jeez 13:41:40 <vladikr> yeah, going to look at this closely today 13:41:50 <ndipanov> vladikr, please do - I will too 13:41:57 <vladikr> maybe I'm missing something 13:42:34 <lbeliveau> I couldn't follow your work on that (was busy on prodcut stuff), but let me know if I can help 13:43:39 <ndipanov> lbeliveau, well the main drawback of that approach is that it's not used in consume_requests which means there will be races 13:43:48 <ndipanov> that approach = vladik's series 13:44:00 <ndipanov> otherwise it's very contained 13:44:05 <ndipanov> and straightforward 13:44:27 <ndipanov> lbeliveau, but please if you get a chance take a look at it and even take it for a spin 13:44:33 <ndipanov> and report any bugs that you find 13:45:02 <lbeliveau> vladikr: what is your series in question ? 13:45:12 <ndipanov> lbeliveau, it's merged already 13:45:34 <vladikr> lbeliveau, I should have a link somewhere.. sec 13:45:54 <ndipanov> but I think gerrit doesn't keep links between patches once they merge 13:46:11 <ndipanov> vladikr, but if you have all the links... 13:46:38 <vladikr> ndipanov, yeah, it's messed up 13:46:51 <vladikr> https://review.openstack.org/#/q/topic:pf-neutron-port 13:47:40 <lbeliveau> thanks, will have a look, will be helpful to get perspective for https://review.openstack.org/#/c/242087/ 13:48:05 <vladikr> lbeliveau, thanks 13:49:44 <ndipanov> ok cool - let's give ourselves 10 min back 13:50:20 <ndipanov> chat in 2 weeks - I am assuming the topic will be the same - so keep an eye out for bugs :) 13:50:31 <ndipanov> bye folks 13:50:34 <vladikr> ndipanov, thanks 13:50:36 <lbeliveau> bye guys 13:50:42 <vladikr> bye 13:51:00 <ndipanov> #endmeeting sriov 13:51:26 <ndipanov> err.. 13:52:07 <ndipanov> that didn't work 14:01:15 <openstack> dprince: Error: Can't start another meeting, one is in progress. Use #endmeeting first. 14:01:19 <d0ugal> o/ 14:01:26 <shardy> o/ 14:01:37 <jdob> o/ 14:01:48 <michchap> o/ 14:01:50 <adarazs> \o 14:02:07 <derekh> the last meeting needs to be ended first 14:02:26 <derekh> #endmeeting 14:02:37 <trown> ruh roh 14:02:48 <dprince> derekh: I think it needs to come from the person who started the meeting 14:03:09 <gfidente> o/ 14:03:16 <hrybacki> o/ 14:03:19 <slagle> ndipanov: can you #endmeeting? 14:03:21 <derekh> ndipanov_ ^^ 14:03:24 <jdob> this is the openstack equivalent of turf wars 14:03:38 <dprince> slagle: I don't think it was him either 14:03:44 * leanderthal starts the west side story soundtrack 14:03:48 <derekh> ndipanov: I guess it didn't end because you started it as "ndipanov_" 14:03:50 <leanderthal> *snap snap* 14:04:26 <ndipanov_> #endmeeting sriov