14:00:59 <johnthetubaguy> #startmeeting nova 14:00:59 <openstack> Meeting started Thu Jul 10 14:00:59 2014 UTC and is due to finish in 60 minutes. The chair is johnthetubaguy. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:01:00 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:01:02 <openstack> The meeting name has been set to 'nova' 14:01:10 <johnthetubaguy> who is around for today's meeting? 14:01:20 <bauzas> \o 14:01:23 <PaulMurray> hi 14:01:27 <n0ano> o/ 14:02:00 <PaulMurray> did I just join the scheduler meeting? :) 14:02:06 <alaski> o/ 14:02:13 <yjiang51> o/ 14:02:19 <mriedem> hi 14:02:20 <n0ano> PaulMurray, nope, this is nova 14:02:24 <johnthetubaguy> #topic Juno-2 is 24th July 14:02:29 <johnthetubaguy> yep this is nova 14:02:41 <johnthetubaguy> same people though, lol 14:02:45 <bauzas> :) 14:02:50 <johnthetubaguy> anyways, on to Juno status 14:03:05 <johnthetubaguy> spec backlog is growing again, lots to review 14:03:19 <johnthetubaguy> Waiting on Reviewer: 40 14:03:20 <adrian_otto> o/ 14:03:32 <johnthetubaguy> Average wait time: 5 days, 3 hours, 16 minutes 14:03:48 <bauzas> #link http://russellbryant.net/openstack-stats/nova-openreviews.html 14:03:51 <johnthetubaguy> so, I kinda want to get this down before we call the specs for Juno closed, but lets see how things are looking later 14:03:58 <johnthetubaguy> I mean nova-specs reviews 14:04:04 <bauzas> oops, bad copy/paste 14:04:06 <johnthetubaguy> #link http://russellbryant.net/openstack-stats/nova-specs-openreviews.html 14:04:14 <mriedem> johnthetubaguy: i thought spec reviews were done after 7/3 or something? 14:04:22 <johnthetubaguy> anyways, today should be freeze day 14:04:26 <sgordon> mriedem, new proposals were 14:04:46 <mriedem> or there are spec reviews, but anything after 7/3 is -2 (spec freeze) unless it's an exceptional case 14:04:50 <johnthetubaguy> mriedem: yeah, that was no new proposals, although lots of people were thinking the opposite, so something went wrong there 14:05:07 * mriedem admits to not reading all of the process stuff in the ML 14:05:08 <johnthetubaguy> #link http://lists.openstack.org/pipermail/openstack-dev/2014-June/038475.html 14:05:18 <johnthetubaguy> Jul 3 (-9): Spec proposal freeze 14:05:18 <johnthetubaguy> Jul 10 (-8): Spec approval freeze 14:05:19 <bauzas> do we have an Exception process ? 14:05:31 <bauzas> haven't seen it 14:05:46 <johnthetubaguy> bauzas: its the same as normal, shout up on the ML, and in the nova-meeting 14:05:48 <bauzas> I mean a SpecFreezeException process 14:05:55 <bauzas> johnthetubaguy: ok 14:06:17 <johnthetubaguy> anyways, I think we will need to take a look at where we are at tomorrow 14:06:29 <johnthetubaguy> we just haven't kept up in the last week 14:06:46 <johnthetubaguy> and I suspect thats because everyone through the reviews were meant to have stopped already :( 14:07:13 <johnthetubaguy> anyways, it is what it is right now, probably thinking we might approve a few tomorrow, if there are stragglers 14:07:18 <johnthetubaguy> any more on blueprint? 14:08:17 <johnthetubaguy> #topic Bugs 14:08:20 <johnthetubaguy> #link http://54.201.139.117/nova-bugs.html 14:08:43 <johnthetubaguy> so we have some new tools cooking 14:08:55 <johnthetubaguy> I am sure tjones would appreciate some feedback 14:09:21 <johnthetubaguy> any more on bugs? 14:09:33 * johnthetubaguy wonders if his IRC is actually working... 14:09:37 <mriedem> no 14:09:55 <johnthetubaguy> #topic Gate Status 14:10:00 <mriedem> it's flaky 14:10:04 <johnthetubaguy> so hows we doing? not well I feel 14:10:04 <mriedem> getting things in batches 14:10:10 <mriedem> http://status.openstack.org/elastic-recheck/ 14:10:15 <johnthetubaguy> #link http://russellbryant.net/openstack-stats/nova-openapproved.txt 14:10:17 <mriedem> the postgresql fix is going through the gate now 14:10:28 <johnthetubaguy> 18 nova patches in the gate queue 14:10:30 <mriedem> if there are any postgresql experts out there, some questions in here: http://lists.openstack.org/pipermail/openstack-dev/2014-July/039800.html 14:10:33 <johnthetubaguy> ah, good news 14:10:48 <mriedem> jogo fixed another gate issue yesterday for quotas and fixed ips in nova-network 14:11:15 <johnthetubaguy> mriedem: got any top bugs that need urgent volunteers? 14:11:16 <mriedem> dansmith made some changes in the virt firewall driver that seems to have helped some on the ssh timeouts 14:11:19 <mriedem> but it's not gone 14:11:40 <mriedem> johnthetubaguy: the two big ones, fixed ips over quota and postgresql have fixes 14:11:43 <mriedem> so not really atm 14:11:53 <mriedem> i'm sure we'll eff that up in no time though... 14:12:00 <johnthetubaguy> awesome, it sounded that way, just double checking 14:12:07 <johnthetubaguy> yeah, true true... 14:12:35 <mriedem> so nothing else on gate stuff 14:12:38 <johnthetubaguy> there was talk about making sure certain changes are rechecked several times before approving, what types of changes were those again? 14:12:48 <johnthetubaguy> resize maybe? 14:13:43 <johnthetubaguy> OK, I guess we are not sure 14:14:00 <johnthetubaguy> #review status 14:14:01 <mriedem> can't remember 14:14:29 <johnthetubaguy> Average wait time: 15 days, 8 hours, 33 minutes 14:14:35 <johnthetubaguy> #link http://russellbryant.net/openstack-stats/nova-openreviews.html 14:14:38 <haypo> hi. is there a gantt meeting here? or is it later? 14:14:46 <johnthetubaguy> lots of very long waiting reviews 14:14:53 <bauzas> haypo: next week on Tues 14:15:04 <n0ano> haypo, tues at 1500 UTC 14:15:04 <johnthetubaguy> haypo: this is the nova meeting, see IRC channel topic name 14:15:28 <johnthetubaguy> anyways, hoping that number gets better next time we check, that is all 14:15:44 <johnthetubaguy> #topic Sub team reports 14:15:48 <haypo> ok, thanks ;) 14:15:53 <johnthetubaguy> OK, people with topics please raise your hands 14:16:04 * n0ano o/ 14:16:25 <johnthetubaguy> n0ano: go go scheduler update 14:16:25 <bauzas> o/ 14:16:41 <adrian_otto> o/ 14:16:54 * bauzas vs. n0ano 0-1 14:17:08 <bauzas> n0ano: up to you :) 14:17:08 * n0ano bloody keyboard froze 14:17:23 <n0ano> anyway, lots of activity on gantt (scheduler) this week... 14:17:36 <johnthetubaguy> o/ 14:17:51 <bauzas> that's worth saying that... :) 14:18:08 <n0ano> concensus seems to be that we want to have a fully functional scheduler after the split (not incremental work) so we have some more work we need to do... 14:18:29 <haypo> n0ano: i wrote to you privately, tell me when you are able to talk 14:18:58 <n0ano> biggest addition is dealing with all aspectes of the resource tracker, big idea is to move the RT into the scheduler, so that will get interesting 14:19:02 <johnthetubaguy> n0ano: OK, so are there plans for a quick prototype to prove things? 14:19:22 <bauzas> johnthetubaguy: n0ano: I left a mail in the -dev list about that 14:19:24 <n0ano> johnthetubaguy, absolutely, bauzas and I are working on creating one 14:19:39 <n0ano> bauzas, do you want to add anything? 14:19:40 <bauzas> johnthetubaguy: https://review.openstack.org/105747 14:19:54 <bauzas> n0ano: yep 14:20:08 <bauzas> n0ano: lemme just find the link for the -dev thread 14:20:10 <johnthetubaguy> n0ano: bauzas: I meant a prototype gantt that fits the suggested client 14:20:45 <bauzas> #link http://lists.openstack.org/pipermail/openstack-dev/2014-July/039863.html 14:21:25 <johnthetubaguy> OK 14:21:26 <bauzas> so, I'm proposing an incremental work for the client 14:21:27 <n0ano> johnthetubaguy, we want to get 3 things done in nova first (client lib, db isolation, RT) and then we can split off a gantt POC 14:22:12 <bauzas> n0ano: my mail is about the last step (RT) which needs to be evaluated with regards to the timeline 14:22:12 <johnthetubaguy> hmm, OK 14:22:27 <johnthetubaguy> I think we could move the RT at a later date, but anyways, lets move on 14:22:34 <johnthetubaguy> adrian_otto: your turn 14:22:55 <bauzas> johnthetubaguy: feel free to leave a comment in the -dev thread then :) 14:23:01 <adrian_otto> tx 14:23:16 <adrian_otto> COntainers team is organizing to make spec proposals for where containers will fit in OpenStack 14:23:31 <adrian_otto> initial plans are to use the spec repo template format 14:23:43 <adrian_otto> should this proceed, even though specs are closed for Juno? 14:23:51 <johnthetubaguy> bauzas: yep, just need to work out what I want to say 14:23:58 <adrian_otto> when will they open for K release? 14:24:11 <johnthetubaguy> adrian_otto: not sure when K will open at this point 14:24:31 <johnthetubaguy> adrian_otto: but current template is a good place to start 14:24:33 <alaski> that's going to be discussed at the mid-cycle 14:24:35 <adrian_otto> is there somewhere we can submit specs that are likely to need a lot of collaboration that will eventually belong in K? 14:24:44 <johnthetubaguy> alaski: +1 14:25:16 <johnthetubaguy> adrian_otto: your own fork of the spec repo might be best at this point, till we get it decided 14:25:25 <johnthetubaguy> adrian_otto: but other may have much better ideas 14:25:37 <johnthetubaguy> s/other/others/ 14:25:40 <adrian_otto> Eric Windisch took an action item to follow up with Mikal for futher guidance as well 14:25:52 <alaski> adrian_otto: nothing prevents you from submitting a spec for K. it just won't get any attention from core/driver reviewers 14:26:00 <johnthetubaguy> adrian_otto: attend the mid cycle summit is a good bet 14:26:19 <johnthetubaguy> alaski: yeah, it might get a −2 put on it, but yeah, you can put it there 14:26:20 <adrian_otto> should I just make a specs subfolder named "K"? 14:26:38 <adrian_otto> next to the juno one? 14:26:42 <johnthetubaguy> adrian_otto: sure, if that works OK for you 14:26:54 <adrian_otto> ok, next subject... 14:27:01 <adrian_otto> on mid-cycle 14:27:04 <johnthetubaguy> there might be a new spec, etc in K, but yeah, lets just see what works 14:27:32 <adrian_otto> do you prefer that containers enthusiast join the Nova mid-cycle meetup in Chicago, or hold another event? 14:28:13 <johnthetubaguy> I have a preference for having containers people talk at the Nova session about nova integration issues 14:28:13 <adrian_otto> there is a group of about 5 who are committed to attend 14:28:32 <johnthetubaguy> I don't know about finding your own room, thats a question for intel folks 14:28:54 <n0ano> adrian_otto, I thought the mid-cycle meetup was in Oregon? 14:28:59 <adrian_otto> might it make sense for us to add one day just for Containers? Would Intel be open to that sort of a request? 14:29:09 <johnthetubaguy> oh, good point, we are in Oregon 14:29:13 <adrian_otto> that way people who are super interested could stay for that, and others could return home 14:29:43 <johnthetubaguy> which Chicago meet up are you thinking about, is that another group of projects? 14:29:48 <johnthetubaguy> like ops or something? 14:29:50 <mriedem> adrian_otto: most nova people probably already have their trips and flights booked 14:29:57 <adrian_otto> oh, Oregon 14:30:03 <adrian_otto> sorry to Confuse you with Chicago 14:30:04 <n0ano> adrian_otto, NP, we have 3 rooms right now (nova, ironic, breakout) containers people can take the break out room or I can reserve a 4th room 14:30:11 <johnthetubaguy> no worries 14:30:34 <adrian_otto> ok, n0ano I will follow up with you, and we can sort that out 14:30:38 <johnthetubaguy> cool, so n0ano and adrian_otto can sort that out, cools 14:30:43 <johnthetubaguy> any more? 14:30:48 <mriedem> maybe we should just stop doing design summits... 14:30:52 <n0ano> adrian_otto, NP 14:31:08 <adrian_otto> that's all for my update. Strong attendance in the Containers meetings, and lots of good discussion. 14:31:12 <adrian_otto> *end update* 14:31:37 <johnthetubaguy> mriedem: or do design summits every three months, once with the conference, once without, yeah, duno. 14:31:43 <johnthetubaguy> adrian_otto: awesome, thanks 14:31:52 <bauzas> johnthetubaguy: strong +2 here 14:31:55 <johnthetubaguy> I put my hand up, but nothing really exciting 14:32:24 <johnthetubaguy> XenAPI CI seems to be keeping up and running, some more improvements communing soon, with a little bit of luck 14:32:29 <johnthetubaguy> nothing more from me 14:32:35 <johnthetubaguy> #topic Open Discussion 14:32:42 <mriedem> johnthetubaguy: mostly joking 14:32:47 <johnthetubaguy> OK, so there is a note about a nova-client release 14:32:57 <mriedem> o/ 14:32:59 <mriedem> novaclient release 14:33:06 <mriedem> mikal hasn't said anything in a week on that 14:33:09 <mriedem> came up in IRC last week 14:33:27 <johnthetubaguy> mriedem: yeah, but its getting such a pain with these midcycles, maybe we should club together with the other integrated projects 14:33:30 <mriedem> i think if nobody objects we should just have russellb do it 14:33:39 <johnthetubaguy> russellb: I am good with you doing that 14:33:48 <johnthetubaguy> seems like we need that doing 14:33:52 <russellb> have asked mikal a couple times, no response 14:33:53 * russellb shrugs 14:34:06 <johnthetubaguy> well anyone against russellb doing a release in here? 14:34:35 <johnthetubaguy> I think mikal is locked in a room writing an openstack book, or on a plane on the way to the room, so I duno 14:34:44 <johnthetubaguy> anyways, sounds like we are all for it here 14:34:55 <johnthetubaguy> if that helps at all 14:35:03 <russellb> k, will do 14:35:12 <johnthetubaguy> awesomeness 14:35:23 <johnthetubaguy> so, good news, thats the end of the agenda 14:35:32 <mriedem> russellb: thanks 14:35:33 <johnthetubaguy> any thing people want to talk about? 14:35:56 <mriedem> there is a ML thread on nova virt driver requirements re: containers 14:36:06 <mriedem> comes back to the fact we don't have a 'core' virt driver API defined 14:36:17 <mriedem> were we going to talk about that at the meetup? 14:36:21 <dansmith> that was supposed to be a task for this cycle 14:36:30 <johnthetubaguy> dansmith: +1 14:36:34 <sgordon> johnthetubaguy, mikal isn't at the book sprint 14:36:34 <dansmith> we talked about doing it in ATL, but nothing has come of it, AFAIK 14:36:39 <sgordon> johnthetubaguy, could well be in transit tho 14:37:30 <johnthetubaguy> dansmith: that was defining the tempest tests every "good" driver should pass, among other things? 14:37:38 <dansmith> johnthetubaguy: no 14:37:48 <dansmith> johnthetubaguy: the functions that every supported virt driver must implement 14:37:49 <johnthetubaguy> dansmith: OK, you mean the virt driver itself? 14:38:04 <johnthetubaguy> right 14:38:05 <dansmith> johnthetubaguy: i.e. "it must support create, destroy, start, stop, resize, etc" 14:38:10 <mriedem> so i can add it to the meetup etherpad as a topic 14:38:19 <johnthetubaguy> yeah, I guess the attach volume is another 14:38:26 <dansmith> right, those sorts of things 14:38:26 <johnthetubaguy> mriedem: thanks, thats a good plan 14:38:31 <mriedem> i just don't want to see the meetup turn into several 45 min sessions like the design summits 14:38:48 <russellb> right, and every big topic shouldn't have to wait for in person to make progress 14:38:49 <mriedem> especially with nova/ironic/containers all being there at the same time 14:39:01 <dansmith> yeah, that was my concern about having the co-located groups :( 14:39:11 <johnthetubaguy> hmm true 14:39:27 <dansmith> then again, I feel like we talked about a lot of things in ATL and nothing came of them, 14:39:36 <dansmith> so maybe we need design summit MK-II :/ 14:39:48 <mriedem> russellb: yeah, i guess we could always vote on a core virt driver api in ML and meetings 14:39:52 <mriedem> and then expand as needed 14:40:09 <mriedem> yeah i feel like summit design sessions were mostly unproductive 14:40:13 <mriedem> but it was my first rodeo 14:40:14 <russellb> i mean, it will take some work, need to do some good digging around on what various platforms support etc 14:40:14 <johnthetubaguy> mriedem: well what is the core list today, (I should read the ML thread)? 14:40:16 <mriedem> 40 min isn't much 14:40:25 <russellb> something i'd expect the PTL to drive, honestly 14:40:27 <dansmith> johnthetubaguy: we don't have one 14:40:46 <johnthetubaguy> dansmith: sorry, I meant, what do all virt drivers do today 14:40:52 <dansmith> johnthetubaguy: the list of things supported by the divers varies widely 14:40:54 <mriedem> johnthetubaguy: http://lists.openstack.org/pipermail/openstack-dev/2014-July/039295.html 14:40:57 <russellb> support matrix wii page is closes we have 14:40:59 <russellb> closest* 14:41:03 <dansmith> yeah 14:41:14 <johnthetubaguy> yeah 14:41:51 <johnthetubaguy> I am thinking, we pick a very small set, we map to tempest tests, then we start expanding till people raise their hand, but thats just the first thing that comes into my head 14:41:58 <dansmith> see, 14:42:02 <dansmith> I think that's not the way to do it, 14:42:18 <dansmith> because we wanted to use this to *drive* some of the drivers to implement things they've been punting on 14:42:24 <russellb> let's please be lighter weight than defcore, heh 14:42:32 <dansmith> so I'd rather look at what we want to get to, set that as the core and aim for getting there 14:42:37 <dansmith> like we did for the first round of CI requirements 14:42:39 <johnthetubaguy> dansmith: OK, thats a good point, so driver for a consistent API goal? 14:42:51 <russellb> dansmith: needs to also be set as a reasonable line we think can be reached by what we want to be able to fit 14:42:52 <dansmith> (which is another thing we wanted to expand this cycle, which we discused, and which didn't happen) 14:43:04 <russellb> dansmith: yes :( 14:43:05 <dansmith> russellb: surely, pause is clearly out of scope forever because: vmware :) 14:43:07 <johnthetubaguy> dansmith: +1 14:43:36 <johnthetubaguy> dansmith: right, pause, thats a good one to exclude, and thats probably cool 14:43:40 <dansmith> yep 14:43:45 <dansmith> it's the example we keep coming up with 14:43:54 * johnthetubaguy remembers the conversation now 14:44:10 <johnthetubaguy> volume attach is the container issue 14:44:13 <dansmith> yes 14:44:23 <johnthetubaguy> I guess whats ironic got, pause is bad there too I supose 14:44:32 <dansmith> and I really don't think we can exclude volumes, so we need some sort of tiered approach I guess, 14:44:45 <dansmith> to exclude volumes for containers, or require/allow them to host-mount them or something 14:44:46 <adrian_otto> johnthetubaguy: https://github.com/openstack/cinder/blob/master/cinder/tests/compute/test_nova.py 14:44:50 <dansmith> and ironic is a problem, yeah 14:44:54 <adrian_otto> to pass you just need to attach and snapshot 14:44:54 <johnthetubaguy> dansmith: yeah, I think we need a "relatively" consistent storage model 14:45:05 <russellb> there has been good volume progress on the docker side btw 14:45:16 <adrian_otto> unless there is a func test somewhere that shells in to a guest do a mount 14:45:19 <russellb> there has been work to get it supported in docker itself 14:45:31 <dansmith> russellb: yeah, but still requires the host to mount them, AFAIK 14:45:38 <russellb> not my understanding 14:45:38 <johnthetubaguy> russellb: awesome, dansmith: that sure validates your approach 14:45:52 <adrian_otto> russellb: yes, but those efforts need to be balanced with security concerns that are rooted in Linux kernel implementation of CAP 14:45:54 <russellb> thought they were looking at exposing block devices to the container *shrug* 14:46:03 <russellb> adrian_otto: *nods* 14:46:09 <russellb> hasn't been accepted into docker yet anyway 14:46:15 <dansmith> russellb: yeah, not sure that can be done, by my understanding 14:46:17 <dansmith> but anyway 14:46:25 <dansmith> johnthetubaguy: yes, this is why we shoot for what we want, not what we have :) 14:46:45 <johnthetubaguy> dansmith: yes, makes total sense now, agreed 14:47:11 <russellb> but of course, these discussions are just wasted time unless someone takes it and runs with it 14:47:16 <dansmith> yar 14:47:21 <russellb> and i don't have the time for stuff like this this cycle 14:47:26 <johnthetubaguy> I was just going to ask 14:47:34 <johnthetubaguy> anyone able to take this? 14:47:50 <johnthetubaguy> I wish I could, but its not happening right now 14:47:51 <dansmith> not like I have time either, but if nobody else will, 14:48:02 <dansmith> I guess I'll try to own the virt driver requirements thing 14:48:17 <johnthetubaguy> dansmith: I am happy to validate a list from a XenServer side 14:48:44 <dansmith> we'll see how far I get before I ignore it too long and we have to find another owner :P 14:49:12 <dansmith> but, I'll try to have a proposed list and/or discussion points for the mid-cycle 14:49:18 <dansmith> I'll do that on the plane 14:49:36 <johnthetubaguy> dansmith: awesome, we the meeting attendees, hand you the time bounded mutex of leadership on the virt driver interface 14:49:50 <johnthetubaguy> sounds like a good plan 14:49:54 <dansmith> johnthetubaguy: do I get a certificate or trophy for that? 14:50:05 <johnthetubaguy> I think we need to get people to own all the suggested topics 14:50:07 * dansmith imagines a trophy shaped like a steaming turd 14:50:30 <russellb> on the plane, huh? 14:50:34 <johnthetubaguy> dansmith: sure some sort of beer like thing 14:51:00 <johnthetubaguy> I hear I can't use my laptop on the plane, else I can't get through security again, or something 14:51:05 <johnthetubaguy> anyways, that was a good topic 14:51:07 <russellb> lol. 14:51:11 <johnthetubaguy> any more of those, or should we stop? 14:51:17 <russellb> can't turn on? must be a bomb 14:51:37 <russellb> i'm sure we could think of plenty of big things we've talked about but aren't making a lot of progress on :) 14:51:39 <dansmith> I need to run anyway 14:51:41 <johnthetubaguy> russellb: quite, one way to force people to repair the batteries 14:51:52 <johnthetubaguy> russellb: agreed 14:51:57 <adrian_otto> Exide cheers. 14:51:57 <russellb> what to do with cells ... what we're doing with v3/v2.1 ... 14:52:17 <russellb> what we're doing to improve CI ... 14:52:23 * johnthetubaguy general sigh noises 14:52:33 <russellb> yup, heh 14:52:38 <dansmith> oye 14:52:43 <johnthetubaguy> yeah, I will start making a list in the meeting thing 14:52:46 <johnthetubaguy> bye all 14:52:50 <russellb> o/ 14:53:00 <johnthetubaguy> thanks for comming 14:53:02 <johnthetubaguy> #endmeeting