21:00:16 <russellb> #startmeeting nova 21:00:17 <openstack> Meeting started Thu Mar 13 21:00:16 2014 UTC and is due to finish in 60 minutes. The chair is russellb. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:18 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:00:20 <openstack> The meeting name has been set to 'nova' 21:00:21 <russellb> hello, everyone! 21:00:26 <dansmith> o/ 21:00:41 <dripton> hi 21:00:42 <alaski> hi 21:00:42 <sandywalsh_> o/ 21:00:43 <cyeoh> hi 21:01:01 <russellb> #topic icehouse-rc1 21:01:13 <russellb> We're in a freeze, and the focus is on getting to our first release candidate 21:01:15 <mriedem> hi 21:01:16 <russellb> #link https://wiki.openstack.org/wiki/Icehouse_Release_Schedule 21:01:23 <russellb> #link https://launchpad.net/nova/+milestone/icehouse-rc1 21:01:30 <russellb> RC1 can be released when that list ^^^ gets down to 0 21:01:41 <russellb> we should only have things on there that we feel should *block* the release 21:01:55 <russellb> that should be regressions or otherwise high/critical bugs 21:02:06 <mriedem> dansmith: https://bugs.launchpad.net/nova/+bug/1290568 is marked as confirmed rather than in progress but more than half is merged 21:02:09 <russellb> anything that is "nice to have" for icehouse should just be tagged with "icehouse-rc-potential" 21:02:47 <russellb> RC1 process sound good / make sense? 21:02:48 <dansmith> mriedem: you're wrong 21:03:00 <mriedem> o-) 21:03:02 <russellb> you can also find an RC burndown chart here: 21:03:04 <russellb> #link http://old-wiki.openstack.org/rc/ 21:04:29 <russellb> dansmith: so what's the status of that one 21:04:36 <dansmith> the nova-network one? 21:04:39 <russellb> yes 21:04:52 <dansmith> one more +A needed 21:04:55 <dansmith> https://review.openstack.org/#/c/79716/ 21:04:58 <russellb> ok, so should be in progress 21:05:01 <dansmith> I need to kick the last one 21:05:08 <dansmith> russellb: it is, mriedem is just crazy 21:05:25 <russellb> dansmith: approved 21:05:27 <mriedem> 7/8 21:05:48 <russellb> hm, arosen not around 21:05:52 <mikal> . 21:06:08 <russellb> he's got a High bug not in progress yet 21:06:16 <russellb> https://bugs.launchpad.net/nova/+bug/1249065 21:06:26 <arosen2> hiya 21:06:35 <russellb> arosen2: hey, reviewing icehouse-rc1 for nova 21:06:42 <russellb> arosen2: you are assigned to https://bugs.launchpad.net/nova/+bug/1249065 which is targeted 21:06:47 <russellb> arosen2: do you think that should block rc1? 21:07:30 <arosen2> russellb: no i don't think this should be a blocker. I thought i had actually fixed this bug though it seems like it's poping up again. 21:07:31 <russellb> looks like a bug that's just been carried along 21:07:34 <russellb> ok 21:07:56 <russellb> added to icehouse-rc-potential instead 21:07:59 <arosen2> I'll look into it though though in the meantime. But no blocker imo 21:08:00 <arosen2> cool 21:08:04 <sdague> well, it's caused 8 gate resets today 21:08:06 <russellb> so everything else is "in progress" 21:08:08 <russellb> at least in theory 21:08:16 <dansmith> sdague: it did? 21:08:24 <sdague> http://status.openstack.org/elastic-recheck/gate.html 21:08:32 <sdague> 8 fails in 24hrs / 25 fails in 14 days 21:08:38 <comstud> i'd like https://bugs.launchpad.net/nova/+bug/1292185 and https://bugs.launchpad.net/nova/+bug/1292181 targetted to rc1 21:08:39 <dansmith> oh this one 21:08:42 <sdague> it's #4 on gate resets 21:08:48 <comstud> they have the potential tag on them 21:08:55 <comstud> cells regressions due to move to objects 21:09:01 <comstud> and a fix is up for review 21:09:09 <russellb> if only we had better cells gate testing huh? :-p 21:09:12 <comstud> so it'll probably land anyway, but 21:09:14 <comstud> haha 21:09:18 <comstud> well 21:09:26 <comstud> this one requires some interesting testing 21:09:47 <russellb> comstud: if they are regressions, yes, target away 21:09:53 <russellb> should bump to High prio 21:09:58 <comstud> ok 21:10:06 <comstud> i can't seem to target 21:10:13 <russellb> are you a member of the nova-bugs team? 21:10:20 <russellb> when's the last time you touched a bug? :-) 21:10:21 <mriedem> comstud: done 21:10:25 <comstud> nor change the Importance 21:10:29 <comstud> A while :) 21:10:31 <mriedem> cripes 21:10:34 <mriedem> before utah :) 21:10:38 <russellb> comstud: https://launchpad.net/~nova-bugs 21:10:48 <russellb> comstud: join that (open team) and you should be able to 21:10:49 <mriedem> comstud: they are bumped and tagged 21:10:51 * mrodden lurks 21:10:53 <russellb> mriedem: thanks 21:10:57 <comstud> done 21:11:23 <comstud> mriedem: thnx 21:11:36 <russellb> any other RC1 impacting bugs we should discuss? 21:12:13 <mriedem> i think this is rc1 https://review.openstack.org/#/c/43822/ 21:12:19 <mriedem> but it could probably drop to rc-potential 21:12:29 <russellb> mriedem: it was on rc1, and i moved it to rc-potential 21:12:34 <mriedem> ok 21:12:34 <mriedem> cool 21:12:40 <russellb> mainly because it's not a regression, and has a known workaround 21:12:48 <mriedem> yeah, just got caught up 21:12:52 <russellb> k, cool 21:12:53 <mriedem> agree 21:13:06 <russellb> trying to be aggressive here on what stays on the list, as that's how we'll stay on top of Rc1 21:13:18 <mriedem> totally understand 21:13:32 <russellb> would appreciate reviews against any of the targeted bugs 21:13:36 <russellb> that should be review priority 21:13:54 <russellb> and if you come across one that seems stalled, let me know 21:14:02 <russellb> so we can decide if we need to push it, or just drop it from the blocker list 21:14:15 <tjones> there are a couple with no update in the last 6-7 days 21:14:25 <russellb> tjones: know which ones off hand? 21:14:34 <tjones> *looking* 21:14:37 <russellb> tjones: just needs review? or response to feedback? 21:15:06 <tjones> review not updated i think. still looking 21:15:14 <tjones> it was in that pastebin i sent you earlier 21:15:46 <tjones> i'll keep looking 21:15:51 <russellb> oh 21:16:05 <russellb> well, we can move on then 21:16:07 <russellb> #topic open discussion 21:16:16 <russellb> feel free to bring up more RC impacting issues 21:16:18 <russellb> or anything else at this point 21:16:41 <russellb> mriedem: you had added something to the wiki i think? 21:16:50 <jogo> dims_ has made great progress with libvirt 21:16:51 <mriedem> oh yeah, looking 21:17:00 <russellb> jogo: dims_ awesome 21:17:05 <russellb> how's that going? 21:17:15 <jogo> so we should hopefully be able to close out some long standing gate bugs 21:17:33 <mriedem> russellb: let me know when ready 21:17:46 <tjones> russellb: it looks like you have already followed up on most of the ones i pointed out 21:17:55 <russellb> tjones: yay 21:18:18 <tjones> russellb: ill check them all and let you know 21:18:24 <russellb> tjones: great, thanks! 21:18:27 <russellb> mriedem: sure go ahead 21:18:36 <mriedem> ok so a long while back this happened http://lists.openstack.org/pipermail/openstack-dev/2013-May/008804.html 21:18:49 <mriedem> that's about virt managers in openstack, i.e. vcenter, ovirt, etc 21:18:54 <mriedem> powervc is a virt manager that ibm wants to get something into openstack to talk to it, 21:19:13 <mriedem> and from utah, talking with comstud and dansmith it sounds like the way to do that is via some new architecture involving cells 21:19:23 <mriedem> and then vcenter could be moved there also, 21:19:44 <mriedem> anyway, we have people that want to work on putting powervc driver code into stackforge but they have no idea what this new architecture idea is, nor do i, :) 21:20:03 <mriedem> so am just bringing it up to see if it's something to talk about at summit, because if it's not i don't see it happening 21:20:19 <mriedem> that's it 21:20:30 <mriedem> i've never been to a summit so i'm not sure how this all works 21:20:31 <russellb> dansmith: comstud ^^^ 21:20:42 <russellb> problem is .... we have a really high level idea, but nobody signed up to do it 21:20:43 <dansmith> so, there is no new architecture, there have been proposals and ideas, 21:20:44 <russellb> so we're stuck 21:20:54 <dansmith> however, it seems like we all agree that exposing it as a virt driver is really really wrong 21:20:58 <comstud> that's funny 21:21:05 <russellb> so now we decide, block new things until it does get done? 21:21:05 <mriedem> russellb: so i think the ibm people would work it, but they are new to the community and don't know what the ideas are 21:21:10 <russellb> or let stuff in the "wrong" way? 21:21:14 <mriedem> i think it's block 21:21:23 <russellb> i'm worried new members would be in over their head on this 21:21:28 <russellb> it's very much non-trivial 21:21:40 <dansmith> yeah, this is big time stuff 21:21:41 <russellb> my gut anyway 21:21:42 <mriedem> yeah, so i guess i'd have to be a middleman, which i already have been :( 21:22:11 <mriedem> but this team is spending a lot of time trying to clean up their existing driver to put it in stackforge and i've explained they are kind of wasting their time 21:22:15 <russellb> it wouldn't hurt to talk through the idea at summit, and then see what happens 21:22:27 <tjones> russellb: looks like this is the only one (so far) that could be removed. The review is abandoned https://bugs.launchpad.net/nova/+bug/1187679 21:23:02 <mrodden> mriedem: probably have to study how cells works right now 21:23:07 <russellb> tjones: indeed, removed 21:23:14 <russellb> so ... 21:23:15 <mriedem> russellb: comstud: dansmith: so part of the motivation is also that if powervc is the guinnea pig, and something can be worked, then vcenter could move that way also so we don't keep having this discussion, like here https://review.openstack.org/#/c/79190/ 21:23:18 <russellb> it'd be like cells rev2 21:23:19 <dansmith> mriedem: mrodden: doing it with current cells is wrong I think 21:23:20 <russellb> or is it rev3? 21:23:25 <dansmith> revN 21:23:27 <comstud> i'd probably like to go into a summit thing with at least some sort of rough guess at a plan 21:23:50 <dansmith> comstud: could you be a little less specific? :P 21:23:50 <comstud> instead of "hey, we should do this and stuff" 21:24:05 <comstud> hehe 21:24:06 <dansmith> I guess there are two questions I think we need to answer: 21:24:18 * jogo wonders how this is RC related 21:24:19 <russellb> if we're blocking stuff on this, we have to communicate some more detail 21:24:22 <dansmith> 1. Are we going to block new complicated systems acting like a virt driver (I hope so) 21:24:23 <russellb> and at least give people a chance at it 21:24:29 <russellb> jogo: we moved past that topic :) 21:24:29 <mriedem> jogo: this is open discussion 21:24:36 <russellb> re #1, yes 21:24:40 <dansmith> and 2. Are we or is anyone going to work on devising a new plan for how those things fit in? 21:24:47 <russellb> i hope so :) 21:24:54 <dansmith> 1 can be yes and 2 can be no, I think, it just means we're not going to take on those new things 21:25:01 <dansmith> and if 2 is yes, then we can talk about a plan 21:25:23 <russellb> right, at least document an architecture 21:25:31 <russellb> and maybe someone will take it on . .. 21:25:33 <russellb> i'm skeptical 21:25:34 <mengxd> dansmith: looks like cell is one possible answer to 2, right? 21:25:34 <mriedem> the upside to 2 though is if you do make it work for something new, you can theoretically move an existing pain point, vcenter, there 21:25:43 <jogo> so ironic is sorta similar to this whole thing 21:25:53 <comstud> yes, ironic falls into this category as well 21:25:56 <dansmith> mengxd: in it's current form, no 21:26:12 <jogo> and they are looking to get things in nova 21:26:14 <dansmith> right, ironic, vcenter and powervc and then moving existing cells to this could be a possibility 21:26:18 <jogo> devananda: ^ 21:26:24 <mrodden> doesnt ironic re-implement the compute api but just for baremetal? 21:26:29 <dansmith> no 21:26:30 <comstud> no 21:26:33 <mrodden> oh 21:26:39 <jogo> its a virt driver today 21:26:42 * mrodden should do his homework 21:26:50 <comstud> it's a virt driver that talks to ironic API 21:26:54 <comstud> (or it will be when it merges) 21:26:57 <mrodden> so it is similar 21:27:02 <dansmith> and nova-bm brought us the 1:n node concept, which is what vcenter does, and what I don't like 21:27:08 <comstud> I think ironic can have an exception because there's a bare metal driver already 21:27:14 <comstud> which it is replacing 21:27:24 <dansmith> an exception for now maybe, yeah, but if we end up with a new thing... 21:27:25 * jaypipes always gets a kick out of the name nova-bm. 21:27:30 <dansmith> vcenter has the same exception by that default 21:27:33 <comstud> jaypipes: :) 21:27:45 <dansmith> jaypipes: it's funny because it's true 21:27:54 <comstud> nova is one big bm 21:27:54 <mrodden> lol... 21:27:55 <jaypipes> isn't it ironic? 21:28:00 <comstud> haha 21:28:04 <dansmith> aaand, we lost it. 21:28:06 <mriedem> right, so there are exceptions for 2 drivers right now 21:28:10 <comstud> aaaand scene. 21:28:20 <mriedem> if we want to move those, it sounds like this is the opporutnity 21:28:36 <mriedem> there are people willing to do the work if they can get guidance on an agreed to architecture 21:28:52 <dansmith> well, and if we do this right, we might be able to get this to help with our cells problem too 21:28:58 <comstud> mriedem: Are these people going to write shitty code? 21:29:08 <mriedem> comstud: possibly 21:29:08 <dansmith> mriedem: people just willing to code isn't going to do it 21:29:12 <comstud> sorry, maybe that was too blunt 21:29:20 <dansmith> mriedem: we can't just give them an architecture and expect them to do it 21:29:20 <mrodden> comstud: not really 21:29:29 <mriedem> dansmith: yeah i know 21:29:39 <mriedem> dansmith: but you and comstud know the most about this i think 21:29:41 <jogo> dansmith: I agree about the exception, for ironic I don't think we should require this change. but rather require a rooadmap for it in the future 21:29:56 <russellb> based on history, i have a hard time seeing that this is going to actually get done 21:29:56 <dansmith> I think it is totally okay for us to say that nobody has the ability and drive to do this right now 21:30:08 <dansmith> jogo: not going to force them unless we have an alternative, of course 21:30:16 <dansmith> jogo: and always a roadmap, for them and vcenter of course 21:30:24 <jogo> dansmith: yup yup 21:30:30 <mrodden> i think that there is definitely a case for accommodating virt managers in OpenStack, and i think nova would be the place 21:30:34 <russellb> long term goal with no real plan to get there 21:30:36 <dansmith> mriedem: yeah, and I'm real busy and comstud works like two hours a week, so.. :) 21:30:50 <mriedem> i know :( 21:30:53 <mriedem> i blame comstud 21:30:55 <comstud> russellb: i could probably get it done, but under no timeframe commitments. it'll happen as 2 all nighters some random time 21:31:00 <dansmith> mriedem: yep 21:31:00 <comstud> come on, it's like 3 hours at least. 21:31:18 <mengxd> dansmith: i'd like to spend good time on this topic. 21:31:18 <comstud> 2 of them spent in meetings 21:31:30 <dansmith> comstud: :P 21:31:30 <russellb> comstud: heh, well would love for you to if you're able to make it a priority 21:31:36 <mriedem> i'm basically just looking for some agreement that this is something worth discussing at summit 21:31:38 <mriedem> baby steps 21:31:46 <comstud> russellb: doubtful.. i'd probably be overcommitting 21:31:51 <comstud> you've seen what I've done for icehouse 21:31:56 <russellb> comstud: heh, thanks for being honest 21:31:59 <comstud> ie: nothing 21:32:02 <dansmith> mriedem: if we do, I feel like we'd be dedicating an hour to agree to what we just did 21:32:04 <russellb> comstud: you get a patch in? 21:32:08 <mriedem> comstud: we'll wake you up when there are any eventlet patches 21:32:09 <dansmith> mriedem: and time slots are really valuable at summit, 21:32:10 <comstud> i did 21:32:13 <russellb> yay 21:32:16 <dansmith> mriedem: so it needs to be well-justified I think 21:32:18 <russellb> so summit 21:32:20 <comstud> i have some from november 21:32:23 <comstud> and 1 yesterday :) 21:32:24 <russellb> we're likely going ot have 1 day less than in the past 21:32:31 <russellb> so bar will go up 21:32:34 <sdague> dansmith: well, the question is would it tend to bring in other folks that might actually pitch in 21:32:41 <comstud> bar tabs 21:32:54 <dansmith> sdague: yes, we think it will, then that'd be worth it.. I'm very skeptical 21:33:02 <dansmith> s/we/if we/ 21:33:05 <mikal> russellb: so that's two days at the summit now? 21:33:07 <sdague> I think that would be the only reason for a summit session, if we thought it would actually kick start something here 21:33:15 <sdague> mikal: 3 days 21:33:16 <russellb> mikal: 3 instead of 4 21:33:17 <dansmith> sdague: yeah 21:33:19 <comstud> sdague: agree 21:33:21 <russellb> mikal: since we're doing a cross project day 21:33:21 <mriedem> sdague: +! 21:33:36 <mriedem> because this is not only nova, it's also cinder and neutron 21:33:41 <mriedem> powervc would be in them just like vcenter is 21:33:49 <mriedem> summit would have cross-project people around 21:33:52 <cyeoh> so we may think that the people who are capable of doing the cells work don't have the time to do it now. But can we find a path where we train some new people who are interested to get the skills required? 21:34:01 <russellb> we're also ignoring the fact that i think powervc is silly in the first place 21:34:09 <comstud> i'm not sure cells needs to be the first thing 21:34:13 <russellb> even if we had this new thing ... 21:34:19 <mriedem> russellb: so replace with ovirt 21:34:22 <mriedem> or whatever virt manager 21:34:25 <mriedem> foo manager 21:34:26 <dansmith> comstud: well, if it ends up not solving the cells thing, then that would suck 21:34:27 <cyeoh> comstud: what is the first thing? 21:34:31 <russellb> nobody is trying to put ovirt in openstack 21:34:34 <comstud> true true 21:34:50 <dansmith> mriedem: I'm not sure I really think there *is* a strong place or need personally 21:34:52 <comstud> cells maybe has the most requirements 21:34:53 <mriedem> russellb: the idea was floated awhile back from this http://lists.openstack.org/pipermail/openstack-dev/2013-May/008804.html 21:35:23 <russellb> mriedem: yeah, those folks disappeared after that summit session 21:35:24 <dansmith> mriedem: ovirt was suggested/presented at a recent summit and very few people had any interest in such a thing, other than for migration 21:35:43 <dansmith> like, before the mic hit the floor 21:35:55 <dansmith> and the room was packed with like four people in the audience 21:35:57 <russellb> we have stuff that works, mostly 21:36:01 <russellb> could be better, but it works 21:36:10 <russellb> and i think that's why it's not high on anyone's priority list 21:36:15 <mriedem> this keeps coming up with vcenter blueprints and patches too, i feel like we just keep kicking this can down the road 21:36:21 <russellb> and i don't think we're missing support for a system lots of people are dying to have 21:36:22 <mriedem> i realize it's not high priority 21:36:32 <mriedem> but i feel like we're ignoring something that everyone knows isn't going away 21:36:50 <dansmith> mriedem: well, vcenter is a real tough fit and it's only going to get worse as things like the races in the aging patch keep coming up 21:36:56 <dansmith> mriedem: so I'm with you there 21:36:58 <mriedem> right 21:37:04 <mriedem> that's what i meant by something 21:37:11 <russellb> but who's motivated to fix it? 21:37:11 <dansmith> mriedem: and I'd like it to be fixed, for sure 21:37:11 <sdague> yeh, the whole idea of putting a DLM inside the virt driver... ooof 21:37:15 <mriedem> as in even during this meeting we're talking about regrets on vcenter and ironic 21:37:21 <dansmith> mriedem: but there just doesn't seem to be inertia 21:37:36 <mriedem> i'm trying :) 21:37:38 <mengxd> i think we need to figure out a correct architecture and implemetation to solve this for the long run. 21:37:40 <mriedem> oh lord i'm trying 21:37:43 <dansmith> mriedem: sure, but not taking on any new ones and managing what we have is one strategy... not ideal, but we have people to do that at least 21:38:02 <russellb> dansmith: yes that's how i've been feeling 21:38:16 <dansmith> so, I guess my feeling is, 21:38:17 <russellb> vcenter is the biggest most obvious one we need to support 21:38:27 <russellb> so forget the rest, why do we need them? 21:38:34 <mriedem> ok, i guess i'm just trying to see if people are even willing to consider talking about this at summit 21:38:35 <dansmith> I'd rather manage what we have now, maybe let comstud iterate on some ideas with no timeline and if something looks promising we can adopt a plan 21:38:51 <dansmith> mriedem: I don't think it's worth an hour, personally 21:38:59 <dansmith> mriedem: how about the next mid-cycle meeting? 21:39:00 <russellb> mriedem: if we have time sure, i'm worried the outcome is just the same old "we can see a better way, but nobody willing/able to do it" 21:39:20 <dansmith> the mid-cycle ones are much easier to plan a talk, and if it's five minutes, move on to something else 21:39:23 <dansmith> harder to do that at summit 21:39:25 <russellb> heh, not even an hour 21:39:26 <russellb> 40 minutes 21:39:28 <mriedem> mid-cycle is better than nothing 21:39:29 <russellb> IIRC 21:39:43 <dansmith> russellb: 40 minutes that always runs into 59 :) 21:39:43 <comstud> looking forward to gutting nova-compute. 21:39:49 <russellb> but yeah 21:39:50 <mriedem> juno would be a wash, but it sounds like it probably would be anyway for this idea 21:39:55 <russellb> until there's something concrete, not sure what to talk about 21:39:56 <dansmith> yep 21:39:59 <russellb> "we should do this" "yep" 21:40:00 <sdague> mriedem: I think if you can find someone to sign up to spear head this in advance, it would sway things 21:40:11 <mriedem> sdague: that's mengxd 21:40:11 <dansmith> s/w/c/ 21:40:18 <mriedem> mengxd: ^ 21:40:28 <mriedem> welcome 21:40:28 <russellb> orly 21:40:29 <mengxd> yes, i will be happy to take this task 21:40:50 <russellb> well, write up a proposal and post to the upcoming nova-specs git repo for review :-) 21:41:08 <mengxd> ok 21:41:15 <mriedem> he's gonna need some kind of input from comstud and dansmith though on what their existing ideas are 21:41:34 <mriedem> i mean i tried following in utah on the topic and was lost 21:41:39 <dansmith> mriedem: well, we can try, but between now and summit I have nary a free minute planned I think 21:41:43 <mriedem> plus there were breadsticks 21:41:50 <russellb> dansmith: i guess you guys could write up a summary of that topic from utah 21:41:51 <comstud> there were breadsticks 21:41:55 <comstud> tasty breadsticks 21:41:57 <russellb> nom 21:42:12 <mriedem> dansmith: i understand, i wouldn't expect this to be a top priority, would just be good to brain dump a bit 21:42:18 <mriedem> and then let people pick at it 21:42:23 <sdague> I think the important thing to remember about summit sessions is they are the middle of the conversation, not the beginning 21:42:28 <beagles> man that sounds really gross 21:42:38 <dansmith> the picking and discussion is time-consuming too, you know 21:42:47 <sdague> so some real pieces need to be done first, otherwise the summit session isn't really useful 21:43:25 <dansmith> the rush here only comes from the newly proposed integrated systems, right? 21:43:28 <mriedem> dansmith: yeah, i'm saying anything is appreciated 21:43:30 <russellb> dansmith: right 21:43:40 <russellb> dansmith: ones that i'm not sure we actually care about 21:43:46 <russellb> so where's the motivation? 21:43:48 <dansmith> I mean, the thing is, if we're already skeptical that we'd take anything powervc-related, then the pressure to spend time on this goes totally away 21:43:51 <dansmith> right, exactly 21:43:53 <mengxd> :sdague agree, the key thing is to make progress on the right path 21:44:17 <mriedem> russellb: dansmith: i think it would also apply to zvm 21:44:24 <mriedem> ibm tried doing something with that in havana and it dropped out 21:44:28 <russellb> the way vcenter works today kinda sucks, but honestly, i think we have more important fish to fry 21:44:30 <dansmith> mriedem: which I have no interest in taking either 21:44:31 <cyeoh> perhaps we can aim for something even lower - a path to help get more people who are interested in this area to get the skill/knowledge they need? 21:44:33 <mriedem> i think they are looking for the same path forward 21:44:34 <russellb> mriedem: yeah not sure we care about that either 21:45:03 <dansmith> mriedem: it is a force-fit at best, and the last few rounds weren't even close to reasonable, implementation-wise 21:45:04 <mrodden> ibm will expend the effort either way i'm sure 21:45:04 <russellb> cyeoh: gain a deep understanding of how nova works, and what problems we face today with existing systems 21:45:08 <mrodden> in community or out 21:45:17 <dansmith> we need to be pretty careful about taking on drivers to satisfy ticky marks 21:45:20 <russellb> the same way we got that knowledge, diving in 21:45:41 <russellb> mrodden: it's not just IBM effort 21:45:54 <russellb> there's a significant -core drain when we consider something for inclusion 21:46:01 <sdague> yeh, there will be a ton of core review time pulled into something like this 21:46:02 <russellb> review, ongoing maintenance 21:46:07 <dansmith> we told hawai (sp?) no to their driver too, 21:46:15 <russellb> and oracle 21:46:18 <dansmith> because we don't think there is a large enough user base to take on the responsibility 21:46:19 <dansmith> and oracle 21:46:22 <russellb> and freebsd 21:46:25 <mrodden> ic 21:46:25 <dansmith> and that 21:46:34 <russellb> for different reasons, but yeah 21:46:41 <sdague> so I think if someone is spear heading this that's new, they also need to bring additional help to the rest of the problems we're trying to solve in nova 21:47:00 <russellb> same ol' gain deep influence and understanding of the project story 21:47:04 <dansmith> having someone plan to jump into nova, learn enough to just implement this thing is far from realistic 21:47:06 <sdague> which is also a great way to understand nova 21:47:11 <dansmith> right, that 21:47:28 <mriedem> sure that's fair 21:47:39 <russellb> spend a year+ doing other stuff so you learn this crazy beast 21:47:45 <russellb> and then we can talk, probably 21:47:48 <russellb> can't spoon feed it 21:48:00 <mrodden> oh, but we have tried... 21:48:03 <dansmith> the same way anyone gets a major feature into nova is spend a year building familiarity, influence, context and then push for $BIG_RADICAL_CHANGE 21:48:07 <mriedem> yeah so for now the powervc stuff will just go up in stackforge and be worked like a virt driver i think, then i guess we see where things go from there 21:48:33 <mriedem> that was the plan when we removed powervm anyway 21:48:49 <russellb> dansmith: yup 21:48:51 <mriedem> maybe as that team works on it more they'll get into nova and it'll be magic 21:49:08 <dansmith> mriedem: yep 21:49:11 <russellb> right, and i personally think 3rd party addons on stackforge are perfectly reasonable 21:49:12 <russellb> have at it 21:49:20 <mriedem> ok, i'm satisfied 21:49:24 <russellb> use CI to make sure you stay current with unstable APIs 21:49:24 <dansmith> woo 21:49:25 <dansmith> progress 21:49:30 <mriedem> yup, 3rd party CI is planned 21:49:38 <russellb> it's highly visible in the community, uses the same dev tools 21:49:44 <russellb> doesn't train the core of nova 21:49:44 <mriedem> yup, gerrit, etc 21:49:50 <russellb> really a good setup 21:49:53 <mriedem> yeah 21:49:53 <mriedem> agreed 21:49:58 <mriedem> ok, thanks for all the time on this 21:49:59 <russellb> s/train/drain/ 21:50:01 <russellb> sure 21:50:06 <russellb> 10 minutes left, anything else? 21:50:25 * mrodden wonders what the restrictions on putting things in stackforge is... 21:50:32 <mrodden> seems to be pretty loose 21:51:31 <dansmith> I think so, 21:51:37 <dansmith> but this is definitely in the realm of reasonable things 21:51:47 <jogo> russellb: yeah one last thing: it would be nice if we start trying to get rid of stacktraces for expected failures 21:51:55 <mrodden> yeah "related OpenStack projects" or something is in the description 21:52:01 <jogo> definitly not a high priority but a good thing for icehouse 21:52:14 <mikal> Juno you mean? 21:52:15 <dansmith> jogo: you wanna say something about partial-ncpu? 21:52:17 <russellb> Juno sure 21:52:25 <russellb> icehouse focus needs to be the RC blocker list :) 21:52:27 <jogo> so if people have cylces to squash staktraces 21:52:41 <jogo> dansmith: sure 21:53:01 <cyeoh> jogo: should we have a holding blueprint for those sorts of fixes? I do them in a sort of an adhoc individual bug basis 21:53:03 <dansmith> basically, jogo is awesome, EOM 21:53:06 <jogo> as many of you have noticed we have had this test grenade*partial-ncpu that as been failing for a long time 21:53:23 <jogo> the test upgrades everything but nova-comptue to make sure we support the rolling upgrade process. 21:53:40 <jogo> the final patches for that job are in flight and we hope to turn it on early next week 21:53:43 <jogo> on=gating 21:53:52 <russellb> cyeoh: https://blueprints.launchpad.net/nova/+spec/clean-logs 21:53:54 <dansmith> +1x10^6 21:54:09 <russellb> +1 :) 21:54:10 <russellb> thanks guys! 21:54:12 <russellb> that's huge 21:54:20 <sdague> yeh, nice job jogo 21:54:22 <cyeoh> russellb: ah, thanks! 21:54:37 <jogo> if the gate didn't get backed up we would have this in already 21:54:55 <jogo> so icehouse should be the first release where we can say 21:55:04 <comstud> jogo: nice! 21:55:05 <jogo> you can run havana and icehouse nova-computes side by side 21:55:21 <dansmith> yes, I'm very much looking forward to that 21:55:21 <russellb> dansmith: you should do a blog post that describes the upgrade process we support now 21:55:24 <jogo> which is pretty mind boggling sweet 21:55:29 <mrodden> nice 21:55:34 <jogo> russellb: ++ to dan writting something 21:55:36 <dansmith> russellb: was planning on it, thanks for spoiling! 21:55:39 <russellb> heh 21:55:49 <russellb> so i think work toward this started in folsom 21:55:54 <mikal> Quick, everyone write one before dansmith 21:55:55 <russellb> 2 years 21:56:00 <jogo> so at the summit hopefully we can start thinking about what the next step is 21:56:06 <jogo> for gating upgrade support 21:56:14 <dansmith> agreed 21:56:25 <jogo> perhaps the DB data migrations in the background 21:56:27 <dansmith> we have other things nova needs to do to make this even better, but more gating would be great too 21:56:33 <dansmith> yes, that for sure 21:56:47 <mikal> I'd like to see multinode gating in Juno as well as an aside 21:56:54 <dansmith> me too 21:56:56 <russellb> mikal: would be nice, yes 21:56:56 <dansmith> a lot. 21:56:57 <jogo> mikal: and I would like a ferrari 21:57:03 <dansmith> heh 21:57:06 <mikal> jogo: don't be so negative 21:57:17 <sdague> yeh, we have a loose idea on how to do multinode allocations 21:57:29 <dansmith> sdague: make it happen man 21:57:29 <russellb> i'd like what we test now to be reliable too :) 21:57:32 <sdague> but it will need a couple of folks 21:57:34 <comstud> ferrari on wishlist isn't negative 21:57:35 <jogo> sdague: oh exciting, also tirpleo is doing it the hard way too 21:57:37 <dansmith> sdague: honestly, I have no idea what you do all day 21:57:41 <sdague> and a bunch more quota 21:57:43 <sdague> :P 21:57:52 <mikal> sdague: I can lend you at least one person for that 21:58:00 <sdague> mikal: awesome 21:58:07 <dansmith> human trafficking now.. awesome 21:58:35 <sdague> that should be a good summit session actually 21:58:45 <jogo> sdague: human trafficking? 21:58:57 <russellb> heh, inappropriate, guys. 21:59:02 <mrodden> well this degraded quickly.... 21:59:05 <dansmith> hah 21:59:10 <russellb> anything else? 21:59:11 <russellb> 1 minute 21:59:11 <dansmith> my fault 21:59:29 <comstud> for shame 21:59:36 <russellb> quick, end on something positive 21:59:42 <dansmith> ICEHOUSE 21:59:45 <russellb> YEAH! 21:59:46 <mrodden> go team 21:59:49 <comstud> 42 22:00:01 <russellb> nova rocks, love working with you all 22:00:04 <jogo> \o/ 22:00:14 <mikal> You have nice hair 22:00:17 <russellb> thanks for your time everyone :) 22:00:17 <comstud> lol 22:00:21 <russellb> #endmeeting