21:00:33 <mikal> #startmeeting nova
21:00:34 <openstack> Meeting started Thu Feb 26 21:00:33 2015 UTC and is due to finish in 60 minutes.  The chair is mikal. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:35 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:38 <openstack> The meeting name has been set to 'nova'
21:00:41 <melwitt> o/
21:00:43 <mikal> Mornnig all!
21:00:50 <mikal> #topic Kilo release status
21:01:00 <tonyb> dims_: Thanks.
21:01:01 <mikal> The next deadline applies to all blueprints: 5th March https://wiki.openstack.org/wiki/FeatureProposalFreeze
21:01:04 <cburgess> o/
21:01:27 <mikal> So that's the next step in our progressive freeze
21:01:32 <mikal> And is disturbingly soon
21:01:38 <mikal> (i.e. next week)
21:01:39 <jogo> o/
21:01:45 <mikal> So just a reminder that its crunch time...
21:02:02 <mikal> John asks that when all code is up, please mark blueprint as NeedsCodeReview, then Implemented when all is merged
21:02:22 <mikal> I also want to note that all the "merge quickly" exceptions from last week merged, so that's cool
21:02:28 <mikal> Anything else on release status?
21:02:53 <mikal> Moving on...
21:02:58 <mikal> #topic Kilo priorities
21:03:09 <mikal> So, a crazily large amount of API code has landed this week
21:03:15 <mikal> How are the other priorities going?
21:03:30 <alaski> cells code could use some reviews
21:03:35 <alaski> it's on the etherpad
21:03:40 <mikal> Cool
21:03:43 <anteaya> nova-net to neutron migration has refocused on L release
21:03:43 <mikal> I will take a look today
21:03:58 <mikal> There was also a spec for scheduler db stuff which was in review when I went to bed last night
21:04:04 <mikal> Not sure what people have said about it overnight
21:04:19 <edleafe> mostly good things :)
21:04:32 <edleafe> Still waiting for your stamp of approval, though
21:04:34 <mikal> anteaya: yeah, thanks for that. So the nova-net migration people are going to take some more time to absorb the feedback from reviews for their prototype, and try for L at this point
21:04:49 <anteaya> obondarev has put up a new patch, if we can get some reviews to figure out if this direction will net results that would be wonderful, patchsize increased from about 350 lines to 700
21:04:51 * mriedem joins late
21:04:58 <mikal> edleafe: so, is there any chance of the implementation getting proposed in gerrit before 5 March?
21:05:10 <dansmith> some of it is up
21:05:11 <dansmith> I tink
21:05:15 <edleafe> mikal: it's been in gerrit for a couple of weeks now as WIP
21:05:16 * mikal notes he is holding up that review
21:05:18 <anteaya> so mostly we need to know, yes on this direction or subclass the api
21:05:42 <dansmith> anteaya: can we add "don't do this at all" to the list of options? :)
21:05:55 <anteaya> dansmith: if that is what you feel
21:05:58 <anteaya> add away
21:06:04 <dansmith> \o/
21:06:12 <anteaya> of course I will follow up with a what do you suggest as a next step
21:06:14 <dansmith> that's definitely what I feel, but I shan't add it
21:06:23 <mikal> So, I'd say we should have a summit session on that bit specifically, except the summit feels too far away
21:06:24 <anteaya> thanks for being honest with me
21:06:26 <anteaya> :)
21:06:28 <edleafe> Dan't has spoken!
21:06:35 <anteaya> ha ha ha
21:06:58 <anteaya> mikal: summit session on edleafe's thing or the migration?
21:07:05 <mikal> YEah, sorry
21:07:09 <mikal> We're talking about two things at once
21:07:12 <mikal> The migration
21:07:13 <anteaya> I don't want a summit session on the migration, but would like a friday table
21:07:20 <mikal> anteaya: ok
21:07:32 <mikal> But I think I'd like to see progress before then, I don't want a stall waiting for the summit
21:07:37 <anteaya> the last thing i want is to be trapped in a room of 400 people who never show up to meetings saying do it my way
21:07:42 <anteaya> agreed
21:07:46 <mikal> anteaya: heh, true dat
21:07:56 <mikal> Ok, so the other topic...
21:08:01 <mikal> edleafe: I promise to review your thing today
21:08:05 <anteaya> I would like a friday table session where we can get through any deadlocks
21:08:07 <edleafe> mikal: thx
21:08:10 <mikal> Or you can send me rotten fruit in the mail
21:08:17 <tonyb> edleafe: I promise to annoy mikal until he does ;P
21:08:33 <mikal> Is there anything else on priorities or shall we move on?
21:08:34 <edleafe> tonyb: that should help
21:08:42 <edleafe> :)
21:09:08 <mikal> #topic Liberty release status
21:09:13 <mikal> Ok, so first off...
21:09:24 <mikal> We've been asked what we'd like for the summit in terms of rooms etc
21:09:28 <mikal> Noting we might not get all we ask for
21:09:32 <mikal> Do we want what we had in Paris?
21:09:42 <mikal> Big room for a couple of days and then a big table room on Friday?
21:09:47 <mriedem> that was nice
21:09:50 <mriedem> friday meetup was good
21:09:57 <mikal> I was surprised by how well the Friday worked to be honest
21:10:13 <mriedem> roundtable is good for the meetup rather than audience style
21:10:23 <mikal> Do we see value in doing _any_ audience style?
21:10:30 <mikal> It has the advantage of a schedule and time boxing
21:10:43 <mriedem> shrug
21:10:49 <anteaya> I like a big round table or square formation rather than long boardroom style
21:10:50 <edleafe> Audience is OK, but we need the rings
21:10:52 <mriedem> i think the 2 sessions on priorities on thursday were also good
21:10:56 <mdbooth> You could add a schedule and time boxing to a roundtable
21:11:22 <mikal> I personally think we should keep the mix the same as Paris
21:11:30 <mriedem> agreed
21:11:30 * mdbooth found the audience style all but useless unless you're in the front row
21:11:30 <mikal> Because audience still feels productive to me
21:11:31 <cburgess> You should have some open sessions, but I think we could have 1 more round table day and cut one open session day maybe.
21:11:32 <dansmith> I think that's fine
21:11:37 <mikal> And I worry that all table would go off the rails
21:11:40 <mikal> Sounds like people agree?
21:11:42 <mikal> Or don't care?
21:11:43 <mikal> :P
21:11:48 <dansmith> ==paris
21:11:49 <mriedem> +1 to paris style
21:11:49 <cburgess> +1
21:11:53 <alaski> +1
21:11:55 <anteaya> all table would go off teh rails
21:11:58 <tonyb> I think that the meetup style discourages passer-by input
21:11:59 <dansmith> mriedem: oh you just coiled a term forever
21:12:02 <dansmith> er, coined
21:12:13 <mriedem> only if bauzas is around
21:12:14 <tonyb> that may be a good thing but I feel ike Paris worked pretty well
21:12:15 <cburgess> *giggity*
21:12:20 <mikal> Heh
21:12:27 <mikal> Ok, I shall let ttx know our desires
21:12:28 <cburgess> Speaking of off the rails....
21:12:38 <mikal> And then we'll see how physical reality maps to that
21:12:52 <mikal> So the other thing is...
21:12:57 <mikal> Liberty specs are open
21:13:09 <mikal> For people who got bumped from Kilo, you can retarget your thing right now if you want to
21:13:20 <mikal> Review wont open for a bit, but it means you can do your bits if you desire
21:13:25 <jogo> mikal: maybe a ML post on that soon?
21:13:36 <mikal> I assume we will do a fast track approval thing in L as well, but I note I am not the L PTL
21:13:39 <mikal> jogo: yep
21:13:48 <mikal> So I don't want to promise thigns for L until we know who the PTL is
21:14:07 <anteaya> I should figure out those election dates
21:14:16 <mikal> That would be good!
21:14:30 <mikal> Actually, I should figure out the extra-atcs thing before the election too
21:14:38 <anteaya> yes please
21:14:47 <mikal> Do you have a deadline for that being done?
21:14:47 <anteaya> and thanks for checking foundation membership
21:14:55 <anteaya> before teh election roll is fixed
21:14:58 <mriedem> mikal: so who would one talk to about planning mid-cycle meetups if one were so inclined?
21:15:06 <anteaya> which is the end of ptl nominations
21:15:13 <mikal> mriedem: heh, I should answer your email...
21:15:14 <mriedem> i guess next ptl
21:15:22 <mikal> mriedem: it is certainly a next PTL call
21:15:32 <mikal> mriedem: but its also a what the team wants thing I think
21:15:57 <mikal> mriedem: I promise to answer your email today
21:16:04 <dansmith> yeah, I'd avoid punting just because $ptl, since it's a thing that needs to get going
21:16:12 <mikal> dansmith: agreed
21:16:23 <mriedem> is there even a liberty release schedule yet?
21:16:31 <mikal> I just feels dishonest to make a lot of edicts about L when it might be someone else living with them
21:16:33 <tonyb> mriedem:: I promise to annoy mikal until he does ;P
21:16:36 <mikal> mriedem: I don't think so?
21:16:42 <mriedem> don't see a wiki
21:16:43 <mriedem> so nope
21:16:45 * mikal is making a worrying number of promises today
21:16:59 <mikal> Let's move on again unless there is something else here...
21:17:00 <mriedem> promise you'll never let go...
21:17:18 <melwitt> lol
21:17:25 <mikal> mriedem: I will be very strong and not tell you my Rick Astley story of the week
21:17:33 <mikal> #topic Gate status
21:17:38 <mriedem> gate is gr8
21:17:40 <dansmith> mikal: until after the meeting, right?
21:17:43 <mikal> So, there was a library thing, yeah?
21:17:46 <mikal> dansmith: sure
21:17:57 <mriedem> oh right, i was going to release novaclient today
21:17:59 <mriedem> oops
21:18:01 <mikal> A release of something which did something but then soemthing fixed the something?
21:18:05 * mriedem blames jogo for not reminding
21:18:09 <mikal> Heh
21:18:10 <mriedem> it's fixed in tree
21:18:18 <mikal> Oh, I was thinking of the keyring release breaking IIRC
21:18:18 <mriedem> supposed to release 2.21.1 (?) today
21:18:30 <mriedem> there were other stable branch breaks earlier in the week but not us
21:18:31 <mikal> mriedem: do we think releasing on Thursdays is a good idea?
21:18:35 <mikal> Or should we wait until Monday?
21:18:39 <mriedem> probably monday at this point
21:18:44 <mikal> I feel that way too
21:18:46 * jogo points out he was asleep when mriedem should have cut the release
21:18:49 <mikal> In case we horribly wedge the world
21:18:55 <dansmith> oh come on
21:18:58 <dansmith> that could never happen
21:19:00 <mikal> LOL
21:19:07 <melwitt> it definitely didn't happen last time
21:19:08 <mriedem> given it's only one change to revert the thing that broke everything before
21:19:09 <mriedem> it should be safe
21:19:11 <mriedem> but monday it is
21:19:18 <dims_> we can always blame something in oslo :)
21:19:21 <mikal> I need a "what could possibly go wrong" tshirt
21:19:29 <mikal> Let's move on again...
21:19:31 <mikal> #topic Bugs
21:19:33 <dansmith> dims_: we could go ahead and get the blame train moving on o.vo
21:19:41 <dims_> :)
21:19:42 <mikal> There is a critical on the agenda...
21:20:00 * mdbooth put that there. Needs eyes.
21:20:42 <mikal> Fair enough
21:20:59 <mikal> I might be worth -1'ing the other contender for that migration number of avoid a rebase on a critical fix?
21:21:00 <mdbooth> TL;DR Relatively minor misconfiguration of Ironic or VMware can delete all your instances
21:21:03 <jogo> so this shouldn't  be a critical
21:21:12 <dansmith> jogo: greed
21:21:12 <jogo> this doesn't impact all users
21:21:17 <mikal> jogo: there's a discussion of that in the bug
21:21:19 <dansmith> sheesh
21:21:20 <dansmith> agreed
21:21:24 <mikal> I tried and was told I was a liar
21:21:27 <mikal> Well, nicely
21:21:38 <mdbooth> mikal: Glad to be polite :)
21:21:44 <mikal> Heh
21:21:48 <mdbooth> It's critical because it will delete all your instances
21:21:50 <mdbooth> i.e. data loss
21:21:59 <dansmith> we have a definition of critical
21:22:02 <dansmith> and this isn't it, IMHO
21:22:07 <mdbooth> dansmith: Data loss is in it
21:22:10 * mdbooth gets the link
21:22:23 <jogo> mdbooth: isn't there a work around? don't run redundant nova computes
21:22:35 <jogo> may  not be a good work around but not critical
21:22:40 <mdbooth> jogo: You can do it accidentally
21:22:43 <dansmith> IMHO, this is using nova in a not valid way
21:22:58 <jogo> you can do a lot of thing accidentally in life
21:23:01 <mdbooth> dansmith: It does involve a misconfiguration, that's right.
21:23:04 <mikal> For reference, https://review.openstack.org/#/c/122240/ is the other contender for migration 277
21:23:09 <jogo> dansmith: yes, that is the bigger issue
21:23:18 <mikal> So, here's the thing
21:23:18 <dansmith> so data loss due to misconfiguration of one or two drivers != critical, IMHO
21:23:21 <mdbooth> However, this is a relatively small misconfiguration with a very large impact
21:23:27 <mikal> We're having a disagreement about a single piece of metadata
21:23:38 <mikal> Couldn't we just review the fix, land it, and get on with our lives?
21:23:44 <mdbooth> mikal: +1
21:23:56 <mikal> Or reject the fix of course if we think its wrong
21:23:57 <jogo> mikal: we could, but there is a deeper issue which this whole model is just bad
21:24:04 <dansmith> right
21:24:16 <mdbooth> This is why it needs discussion.
21:24:26 <mdbooth> This proposal enforces an existing assumption
21:24:39 <mdbooth> The assumption is already implicit in code
21:24:47 <mdbooth> This proposal makes it explode if you break it
21:24:49 <tonyb> This proposal == https://review.openstack.org/#/c/158269/ ?
21:24:57 <mdbooth> Instead of corrupting your instance
21:25:07 <mdbooth> tonyb: Yea
21:25:08 <mikal> I don't disagree with revisiting the design of this, but I do think we should be protecting current deployments of ironic from face punching
21:25:34 <mdbooth> mikal: Also +1
21:25:50 * mdbooth agrees the design should be revisited
21:25:54 * jogo wants more face punching
21:26:02 <mikal> Sigh
21:26:07 <mikal> Ok, so
21:26:07 <jogo> so what is the hold up between the two options?
21:26:13 * bauzas waves late
21:26:13 <mikal> I'd love some review on that proposed fix
21:26:17 <mdbooth> jogo: Which other option?
21:27:04 <jogo> mdbooth: I thought this was https://review.openstack.org/#/c/154907/1 vs  https://review.openstack.org/#/c/158269 ?
21:27:58 <mdbooth> jogo: Ah, yes. That does a similar thing by storing metadata in vsphere. It doesn't cover Ironic, for eg.
21:28:07 <mdbooth> Also, it's uglier.
21:28:09 <dansmith> so,
21:28:10 <sdague> mikal: so, I think https://review.openstack.org/#/c/122240/ should get 277, johanes is good with it, redo this one as 278 and ask people to review
21:28:29 <mikal> sdague: I would be ok with that
21:28:32 <dansmith> I like a driver-specific bandage for this right off
21:28:44 <mikal> sdague: but if we're going to land 240 let's pull the trigger now
21:28:46 <mriedem> +W on https://review.openstack.org/#/c/122240/ - so 277 is taken now
21:28:46 <mdbooth> dansmith: 2 driver-specific bandages
21:28:47 <dansmith> I'm still trying to digest the supposedly global fix
21:28:49 <mikal> sdague: my -1 there is no longer valid
21:28:58 <dansmith> mdbooth: driver specific implies "in each driver affected" yes
21:29:13 <jogo> dansmith: yeah, I like the driver specific thing as well
21:29:15 <mdbooth> dansmith: Also, the assumption is in Nova, not in the driver
21:29:23 <dansmith> mdbooth: I'm aware
21:29:25 <mdbooth> i.e. Nova core
21:29:30 <dansmith> I also think the assumption is fine
21:29:45 <mdbooth> All I'm doing is proposing that Nova's existing assumption is enforced by Nova
21:29:53 <mdbooth> Seems safest and cleanest
21:30:01 <dansmith> I know
21:30:17 <dansmith> I think we should (a) make this non-critical and (b) review/continue off the meeting
21:30:29 <jogo> mdbooth: the bigger issue here is we have drivers that are designed where this is even an issue
21:30:36 <jogo> dansmith: already took it out of critical
21:30:42 <dansmith> cool, then (b) ? :)
21:30:42 <mikal> Ok, so
21:30:47 <mikal> Yeah, let's review this code
21:30:53 <mikal> And discuss there
21:31:01 <mikal> I can't see us resolving this in this meeting
21:31:46 <mikal> #topic Stuck reviews
21:31:52 <mikal> So, there are none on the agenda
21:31:56 <mikal> And I don't know of any
21:32:17 <mikal> Anythign where we will never reach a concensus on gerrit?
21:32:37 <dansmith> how to spell concensus?
21:32:37 <mikal> No?
21:32:37 <sdague> well... I could troll with this - https://review.openstack.org/#/c/150929/
21:32:41 <dansmith> I can't see us ever agreeing on that
21:32:45 <mikal> dansmith: you're dead to me
21:32:53 <dansmith> heh
21:32:54 <mikal> sdague: you're in danger as well
21:32:58 <dims_> ha sdague!
21:32:58 <tonyb> sdague: :D
21:33:02 <dansmith> sdague: take it out of -W
21:33:05 <dansmith> sdague: my +2 is already on it
21:33:12 * mikal rage quits this topic
21:33:17 <mikal> #topic Open Discussion
21:33:21 <mikal> Ok, now discuss openly
21:33:25 <sdague> mikal: so it was only a half troll
21:33:31 <mikal> Or I call tell you about the shirine to Rick Astley I visited this week
21:33:44 <dims_> any feed back on trivial patch monkey process is welcome. https://etherpad.openstack.org/p/kilo-nova-priorities-tracking
21:33:55 <mikal> dims_: I find it annoying
21:34:00 <sdague> but I want to make sure we're good that if we get the change in the ec2 folks need for out of tree, we can deprecate
21:34:02 <mikal> I go to review those things, and they're already merged
21:34:02 <melwitt> there is this review for working around a libvirt bug where there's some uncertainty about whether a workaround config option is needed https://review.openstack.org/151953
21:34:10 <jogo> dims_: been happy with it
21:34:17 <dims_> mikal: we update it twice a day at least
21:34:22 <mikal> sdague: oh, sorry. I thought that was the vmware / ironic cluster review again
21:34:25 <dims_> will try to do it before you wake up :)
21:34:30 <mikal> dims_: everyone else should review slower
21:34:35 <sdague> I figure I'll wait until after ops summit to un WIP it just so we can get some face reactions with ops on it
21:35:01 <bauzas> dims_: provided monkeys have time :-)
21:35:03 <mikal> sdague: oh, so that review is me sucking to be honest
21:35:09 <dims_> bauzas: haha
21:35:10 <jogo> dims_: can we remove the list of merged things now? I don't think we need that list to show it works
21:35:16 <mikal> sdague: I haven't had a chance to sit ttx down and work out a way forward which meets his needs
21:35:25 <dims_> jogo: happy to
21:35:27 <mikal> sdague: its on my todo list
21:35:34 <jogo> dims_: it would make it easier to scroll to the bottom
21:35:37 <jogo> I am lazy
21:35:41 <sdague> mikal: ok, no prob. I just wanted to keep it on the front burner so it doesn't get lost before k3
21:35:53 <mikal> sdague: agreed, thanks for keeping me honest
21:35:58 <dansmith> sdague: mikal we really need quick work on this
21:36:12 <dims_> jogo: i am going to add a counter instead
21:36:15 <jogo> do we have anyone attending the ops summit to take notes?
21:36:19 <jogo> dims_: ++
21:36:29 <mikal> jogo: sdague is
21:36:34 <jogo> mikal: nice
21:36:36 <sdague> yeh, I'll be there
21:36:53 <mikal> dansmith: yep I know, I already feel guilty
21:36:54 <bauzas> dims_: counter on what ?
21:37:13 <dims_> bauzas: count of things merged
21:37:22 <mikal> dims_: I think a counter is useful, because the board occassionally asks how we're fixing this. Metrics are therefore always nice.
21:37:59 <bauzas> dims_: ack. Please don't provide name stats
21:38:37 <mikal> Name stats?
21:38:50 <dansmith> individual stats
21:38:55 <bauzas> I mean, count on reviews/monkey
21:39:04 <mikal> As in who is the best predictor of mergable monkey patches?
21:39:07 <tonyb> mikal: dims_ (11'ity billion) tonyb (0) etc
21:39:16 <dansmith> tonyb: right
21:39:21 <sdague> right, agreed. But bugs merged or reviews closed, or something
21:39:21 <mikal> Huh, why not?
21:39:24 <mikal> Not that I need that stat
21:39:32 <mikal> Just curious why we explicitly don't want it
21:39:33 <dansmith> mikal: gamification
21:39:47 <mikal> Oh, ok
21:39:47 <bauzas> mikal : because I think it would corrupt the process
21:39:47 <dansmith> mikal: and because some people are pretty bad at it :P
21:39:50 <mikal> Sure
21:40:42 * dims_ spies 2 deer sauntering by his window
21:40:47 <mikal> So... in other news most of the electrical bits of my car don't work. Are we done here so I can go and get it fixed?
21:41:03 <mikal> What I would normally call an "early mark"?
21:41:32 <dims_> nova is now oslo namespace free
21:41:40 <sdague> dims_: \o/
21:41:44 <bauzas> \o/
21:41:51 <sdague> and we have novaclient functional tests for a regression
21:41:57 <sdague> those are nice wins this week
21:41:59 <dims_> nice
21:42:10 <dansmith> and o.vo!
21:42:20 <dims_> yay
21:42:22 <mikal> Yay team!
21:42:25 <edleafe> woohoo!
21:42:27 <dansmith> and I'm about to make dhellmann eat his napkin
21:42:28 <sdague> and dansmith drinks ovoltine
21:42:35 <mikal> I shall eat a cookie as a reward
21:42:37 <dims_> hahaha
21:42:38 <dansmith> heh
21:42:47 <dansmith> sdague: oh, t-shirt.
21:42:49 <jogo> mikal: isn't it breakfast time by you?
21:42:53 <mikal> So, I think that means we're done for this week
21:43:00 <mikal> jogo: yes, breakfast cookie of victory
21:43:05 <mikal> Its been lovely typing near you all
21:43:09 <anteaya> dims_: nice, I had a turkey fly past the other day
21:43:09 <mikal> #endmeeting