14:01:25 <johnthetubaguy> #startmeeting nova
14:01:25 <openstack> Meeting started Thu Oct 16 14:01:25 2014 UTC and is due to finish in 60 minutes.  The chair is johnthetubaguy. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01:27 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:01:29 <openstack> The meeting name has been set to 'nova'
14:01:31 <johnthetubaguy> https://wiki.openstack.org/wiki/Meetings/Nova#Agenda_for_next_meeting
14:01:34 <bauzas> \o
14:01:36 <mriedem> hi
14:01:37 <edleafe> o/
14:01:37 <johnthetubaguy> #topic juno
14:01:38 <dansmith> o/
14:01:38 <leifz> lurking
14:01:40 <boden> o/
14:01:41 <jgrimm> o/
14:01:43 <johnthetubaguy> so juno is shipped
14:01:51 <edleafe> \o/
14:01:54 <johnthetubaguy> stable backports for stuff you spot now I think
14:02:02 <n0ano> o/
14:02:08 <johnthetubaguy> usual backport tags for all that fun
14:02:16 <johnthetubaguy> #topic kilo
14:02:17 * beagles lurks
14:02:32 <johnthetubaguy> so specs are open, we are fast approving some, and slow approving some
14:02:41 <johnthetubaguy> any questions or things to raise?
14:03:01 <mriedem> stable/juno doesn't seem to exist yet
14:03:08 <johnthetubaguy> #info if your code reviews are being blocked on a "procedural -2", please ping the reviewer (in email) and remind them to reconsider
14:03:09 <mriedem> nope
14:03:28 <johnthetubaguy> mriedem: hmm, doh, I guess we only just tagged, so thats next
14:03:41 <mriedem> i'll let it go this time
14:03:48 <johnthetubaguy> :)
14:03:54 <johnthetubaguy> anyways, anything on specs?
14:04:05 <johnthetubaguy> everyone as happy as they can be about the new rules?
14:04:16 <johnthetubaguy> any tweaks needed at the moment?
14:04:24 <bauzas> johnthetubaguy: about priorites
14:04:32 <n0ano> johnthetubaguy, too early to tell, we'll comment when we have issues
14:04:34 <bauzas> johnthetubaguy: it's asked to fill them in
14:04:39 <johnthetubaguy> bauzas: to be agreed at the summit I think
14:04:51 <bauzas> johnthetubaguy: k, so leaving a perhaps in this
14:05:06 <mriedem> bauzas: i just put None if my thing wasn't an obvious priority
14:05:16 <johnthetubaguy> mriedem: +1
14:05:24 <bauzas> mriedem: k, I'm taking the etherpad as a reference
14:05:26 <mriedem> it's a priority for me, but not the project
14:05:35 <johnthetubaguy> we can go back and edit the specs, if we come up with a priority that you fit, after its approved
14:05:45 <bauzas> https://etherpad.openstack.org/p/kilo-nova-priorities
14:06:04 <bauzas> I'm taking these chapters as the ones I'm looking to, if so
14:06:09 <bauzas> johnthetubaguy: k, sounds good
14:06:13 <n0ano> does it make sense to approve a spec that has no priority, I would think that would be required
14:06:27 <ndipanov> so what do the priorities mean now - there was so much back and forth that I am not sure what is agreed on
14:06:39 <johnthetubaguy> n0ano: thats another summit discussion, for now, yes, they are low priority
14:06:57 <bauzas> I was understanding that priorities would be set for juno-3 ?
14:07:02 <bauzas> oops
14:07:03 <johnthetubaguy> ndipanov: right now, they have no power, the idea is to agree at the summit
14:07:04 <bauzas> kilo-4
14:07:09 <n0ano> which means they'll probably get ignored so yes, this needs to be discussed at the summit
14:07:09 <bauzas> kilo-3
14:07:34 * bauzas has to watch the new openstack episodes
14:07:42 <johnthetubaguy> so yeah, my current plan is, kilo-3 nothing except high priority stuff merges, so we kinda have a soft blueprint freeze at the end of kilo-2
14:07:49 <bauzas> k
14:08:00 <johnthetubaguy> but lets see how it goes, is the detailed version of the plan
14:08:16 <ndipanov> johnthetubaguy, and we are not doing runways, right?
14:08:36 <mriedem> the policy patch for runways was abandoned
14:08:42 <johnthetubaguy> #info remember, all features need a blueprint, not all blueprints need a spec
14:08:44 <mriedem> which is what i thought we were going to use to get low priority stuff in
14:09:00 <johnthetubaguy> yeah, runways just got shot down too much on the ML
14:09:03 <mriedem> https://review.openstack.org/#/c/115410/
14:09:21 <johnthetubaguy> my hope is to discuss that at the summit, and come up with a comprimise
14:09:37 <ndipanov> johnthetubaguy, ack - that will be a fun discussion :)
14:09:54 <johnthetubaguy> the current one is, kilo-1 and kilo-2 is free, kilo-3 is basically runways
14:09:57 * alaski wanders in late
14:10:19 <johnthetubaguy> cools
14:10:32 <johnthetubaguy> #topic Kilo Summit
14:10:59 <johnthetubaguy> so, I was hoping we would have a session plan, that we could discuss, but we don't really
14:11:13 <johnthetubaguy> I have lobbed a few ideas at the end of the etherpad
14:11:16 <johnthetubaguy> #link https://etherpad.openstack.org/p/kilo-nova-summit-topics
14:11:38 <johnthetubaguy> #info we have 2 days of scheduled slots and 1 day of meetup style
14:11:52 <johnthetubaguy> we need to decide what gets a slot, and what needs meetup style
14:11:53 <ndipanov> I like those ideas - I am not sure we want to do session per BP like we did in the future, meetup can be for that I guess?
14:12:02 <ndipanov> lol future
14:12:03 <ndipanov> past
14:12:05 <ndipanov> in the past
14:12:17 <mriedem> time traveling is fun
14:12:21 <bauzas> johnthetubaguy: slots are 40 mins right?
14:12:27 <johnthetubaguy> bauzas: think so
14:12:42 <johnthetubaguy> ndipanov: agreed though
14:12:53 <johnthetubaguy> so for stuff that is "here is my feature X"
14:12:57 <mriedem> i like meetup style more than slots for everything, but i'd think the process talks would be better for a meetup?
14:13:01 <johnthetubaguy> lets have a spec first
14:13:05 <bauzas> mriedem: +1
14:13:11 <n0ano> johnthetubaguy, you have some big topics on that list, 40 min. is going to be interesting
14:13:24 <bauzas> johnthetubaguy: what if the spec got accepted before the Summit, then violent agreement for 40 mins ? :)
14:13:36 <johnthetubaguy> n0ano: some might get two slots, but yes
14:13:44 <ndipanov> johnthetubaguy, very much agreed
14:13:53 <johnthetubaguy> bauzas: thats the hope, avoid that slot, and re-schedule something else into that
14:13:54 <mriedem> in atlanta a lot of the QA sessions were presentations and those were nicely wrapped up in time for questions within 40 minutes,
14:14:01 <bauzas> johnthetubaguy: awesome
14:14:01 <mriedem> i'd like to see that with blueprint design sessions
14:14:08 <mriedem> and do meetups for big groups of process argument
14:14:20 <dansmith> mriedem: really?
14:14:36 <mriedem> i hate the 40 minute thing
14:14:36 <ndipanov> I though that;s exactly why we don't want that in sessions
14:14:38 <mriedem> is my point
14:14:38 <dansmith> mriedem: I don't really want presentations in design summit slots
14:14:39 <ndipanov> go read a bp
14:14:47 <johnthetubaguy> dansmith: +1
14:14:52 <ndipanov> or spec
14:14:54 <dansmith> small slots for small discussions
14:14:55 <bauzas> mriedem: I seriously doubt that some topics can cover both a presentation and a discussion in 40 mins
14:14:57 <edleafe> dansmith: +1
14:15:00 <johnthetubaguy> ndipanov: yeah, if you can read the spec or bp, thats fine, whatever
14:15:03 <dansmith> if it's a longer discussion, then meetup style
14:15:04 <bauzas> mriedem: but hell yeah, planning will be fun
14:15:14 <johnthetubaguy> bauzas: we probably give them two slots then, thats fine
14:15:15 <mriedem> i'm not talking about a demo
14:15:27 <mriedem> basically going through whatever the blueprint is, or sticking points, idk
14:15:33 <dansmith> -1
14:15:40 <mriedem> i just didn't think most of the design sessions seemed to be useful in ATL
14:15:48 <mriedem> the 40 min ones
14:15:50 <bauzas> mriedem: agreed
14:15:56 <johnthetubaguy> so here is the thing
14:16:01 <johnthetubaguy> we don't have much time together
14:16:03 <dansmith> the ATL sessions weren't really well planned
14:16:09 <johnthetubaguy> if we can do something in the specs, then we can avoid it
14:16:35 <johnthetubaguy> dansmith: +1 hope we can avoid the ATL dumb sessions
14:16:40 <dansmith> yeah,
14:16:51 <dansmith> although we're getting pretty close with minimal planning so .. we'll see
14:16:55 <mriedem> ok, i guess i don't have any other summits to compare to so i started with the worst one...
14:16:56 <johnthetubaguy> so any BP session will be because we can't agree on the spec
14:17:15 <johnthetubaguy> mriedem: well the previous one was better, thats all
14:17:19 <bauzas> johnthetubaguy: or because it's wider than a single spec ?
14:17:44 <johnthetubaguy> bauzas: if we can agree half the specs for the scheduler stuff, then it helps us concentrate on the stuff we need to talk about right?
14:17:55 <bauzas> johnthetubaguy: good question
14:17:57 <johnthetubaguy> bauzas: if we agree all the specs, then cool, nothing to talk about
14:18:18 <johnthetubaguy> Some things are about carving up the work between a group of people
14:18:21 <bauzas> johnthetubaguy: well, we at least need to make sure all reviewers stick with the plan
14:18:37 <bauzas> johnthetubaguy: I mean, summits are good for that
14:18:58 <bauzas> johnthetubaguy: specs having 2x+2 don't necessarly mean that the consensus has been reached
14:19:27 <ndipanov> I don;t think we are looking for consensus
14:19:42 <johnthetubaguy> bauzas: if there were no −1s then the system broke, need to get people to vote on there, anyways
14:19:42 <bauzas> ndipanov: then what ? :)
14:19:48 <ndipanov> we are looking for good implementation
14:19:59 <ndipanov> and features
14:20:09 <ndipanov> screw consensus... :)
14:20:11 <johnthetubaguy> anyways, this is not rally going anywhere
14:20:15 <ndipanov> yes agreed
14:20:18 <ndipanov> move on :)
14:20:19 <bauzas> +1
14:20:25 <johnthetubaguy> ndipanov: lets just say, code means a lot
14:20:30 <johnthetubaguy> good code means more
14:20:31 <johnthetubaguy> anyways
14:20:34 <ndipanov> so one more question
14:20:37 <mriedem> was rally a freudian slip there?
14:20:44 <dansmith> I feel like we're ratholing here
14:20:58 <johnthetubaguy> dansmith: +1
14:20:58 <ndipanov> the only perf things on your list johnthetubaguy is the profiler stuff
14:21:14 <ndipanov> oh and Rally
14:21:16 <johnthetubaguy> ndipanov: do we have perf stuff to talk about?
14:21:22 <mriedem> i thought there were db things
14:21:25 <ndipanov> well yes
14:21:26 <ndipanov> db
14:21:33 <mriedem> jogo had something that zzzeeek said wouldn't help
14:21:36 <johnthetubaguy> OK, not seen that proposed
14:21:38 <mriedem> so assuming zzzeeek has alternatives
14:21:57 <mriedem> jogo's was removing the ORM
14:21:59 <mriedem> SA ORM
14:22:04 <mriedem> bayer said that wouldn't really help
14:22:07 <johnthetubaguy> ah, OK, so thats a thing
14:22:08 <ndipanov> so if I could add one thing to the list - it would be talk about perf - with a strong focus on db
14:22:20 <ndipanov> yes that is an old discussion tho
14:22:25 <johnthetubaguy> mriedem: its helped us in some bits, but its probably more about us using it wrongly
14:22:27 <ndipanov> if that's a thing for the "meetup"
14:22:42 <johnthetubaguy> ndipanov: its on the meet up list already actually, in a really bad way
14:22:44 <ndipanov> then cool - but I feel we really should talk about it
14:23:09 <ndipanov> since it is one of the reasons scheduler and other stuff we all want to violently discuss sucks so bad
14:23:21 <ndipanov> :)
14:23:27 <bauzas> ndipanov: :)
14:23:28 <johnthetubaguy> #topic Gate
14:23:31 <johnthetubaguy> so hows the gate
14:23:37 <n0ano> johnthetubaguy, you updated the etherpad, is someone going to create a spec for DB improvements?
14:23:39 <mriedem> i put some links in the agenda
14:23:47 <johnthetubaguy> ndipanov: do try a single caching scheduler, thats actually quite fast
14:23:50 <mriedem> the gate is....kind of unknown
14:23:59 <mriedem> categorization of gate bugs is at like 58% which sucks
14:24:12 <mriedem> so we have something out there, i suspect the cinder volume delete lvm/vgs hangs, that aren't tracked
14:24:16 <mriedem> i'm working on tracking that today
14:24:21 <johnthetubaguy> lots of new bugs, or just lots of broken code up for review?
14:24:30 <johnthetubaguy> OK, cool, your digging
14:24:37 <mriedem> two bugs to mention
14:24:42 <mriedem> 1. https://bugs.launchpad.net/nova/+bug/1353962 - that's the fixed ips quota one
14:24:44 <uvirtbot> Launchpad bug 1353962 in nova "Test job failes with FixedIpLimitExceeded with nova network" [Undecided,Confirmed]
14:24:48 <mriedem> still need eyes on that, we have better logging now
14:25:02 <mriedem> 2. this is the cinder volume hang bug we're tracking https://bugs.launchpad.net/cinder/+bug/1373513
14:25:04 <uvirtbot> Launchpad bug 1373513 in cinder "Lvm hang during tempest tests" [Critical,Confirmed]
14:25:14 <mriedem> vgs/lvs calls are taking sometimes 2 minutes causing cinder volume deletes to timeout
14:25:18 <johnthetubaguy> n0ano: I should reach out on the ML for owners and spec drivers for some of this stuff
14:25:22 <mriedem> i need to get a better query up in e-r for that one today
14:25:27 <johnthetubaguy> mriedem: is this a review call to action, or a help thing?
14:25:36 <mriedem> johnthetubaguy: help
14:25:42 <n0ano> johnthetubaguy, OK, it's just that that's a rather nebulous topic
14:25:51 <mriedem> we have a patch up for cinder to get some strace on the vgs/lvs calls but it needs some help i think
14:25:54 <johnthetubaguy> n0ano: agreed
14:26:02 <mriedem> https://review.openstack.org/#/c/126735/
14:26:13 <mriedem> someone with rootwrap knowledge to fix that would be nice
14:26:38 <johnthetubaguy> cool, so we are done here I guess, since no one is jumping up to discuss things?
14:26:43 <mriedem> i guess
14:27:01 <mriedem> moving on
14:27:03 <johnthetubaguy> #help the gate still needs more love
14:27:08 <johnthetubaguy> #topic Open Discussion
14:27:26 <johnthetubaguy> OK, anyone got things they want to talk about?
14:27:28 <boden> I have a general question I was not able to get a response to on nova channel
14:27:40 <johnthetubaguy> boden: OK, fire away
14:27:41 <boden> what is our stance on ec2 parity... e.g. https://bugs.launchpad.net/nova/+bug/1370384
14:27:43 <uvirtbot> Launchpad bug 1370384 in nova "Cannot expand root volume by EC2 API" [Medium,In progress]
14:27:56 <johnthetubaguy> boden: patches welcome, I think thats the stance
14:28:12 <johnthetubaguy> we had a sub group pushing on that, but not heard a lot recently
14:28:21 <boden> johnthetubaguy -- the point is we can do that workflow using existing nova resize
14:28:37 <boden> do we really need to support the ec2 "manual workflow"?
14:28:50 <johnthetubaguy> boden: duno, do people want to use it?
14:29:08 <johnthetubaguy> is anyone going to maintain it?
14:29:12 <mriedem> boden: the other thing is we rarely get new/good working ec2/boto tests in tempest
14:29:20 <dansmith> boden: I think the answer here is, it's important if someone thinks it's important and wants to work on it
14:29:25 <dansmith> boden: if not, then not
14:29:25 <mriedem> the last new boto test broke the gate and was reverted
14:29:33 <ndipanov> johnthetubaguy, (stepped away for a bit...) so yeah - caching is definitely one of the solutions to the db suckage :) still we should mention it at least in the meetup
14:29:39 <johnthetubaguy> dansmith: +1 you said what I am failing to express
14:29:51 <dansmith> the lack of enthusiasm here indicates that it's not important, IMHO :)
14:30:05 <boden> dansmith -- I dont mind working on it, but hanve't seen many asks
14:30:19 <boden> anyway fair enough answer
14:30:21 <boden> thanks
14:30:23 <dansmith> boden: if you think it's important, then propose a spec to describe what you think needs changing and why it's important
14:30:28 <n0ano> dansmith, your're probably correct but that goes back to are we devloper focused or user focused, I think we've tilted too far to the developer side right nwo
14:30:31 <dansmith> until then, there's really not much to discuss :)
14:30:45 <boden> dansmith +1
14:30:59 <dansmith> n0ano: no developers being asked by their paying users to add that thing is a good indication to me :)
14:31:00 <johnthetubaguy> anyways, lets not go down that whole
14:31:06 <johnthetubaguy> any more questions?
14:31:20 <johnthetubaguy> dansmith: +1 I am afraid
14:31:32 <johnthetubaguy> seems like we are done early
14:31:35 <n0ano> dansmith, maybe, I won't argue that here
14:31:36 <mriedem> is cells meetup or slot?
14:31:44 <johnthetubaguy> mriedem: slot right now
14:31:51 <mriedem> not seeing it at the bottom of https://etherpad.openstack.org/p/kilo-nova-summit-topics
14:31:53 <dansmith> we can talk about cells for days,
14:31:54 <johnthetubaguy> mriedem: based on whats in my head
14:32:03 <dansmith> so I think that bounding it is fine
14:32:09 <dansmith> maybe a double slot, if we're considering such things
14:32:14 <mriedem> yeah i just want to see the work items/plan
14:32:14 <ndipanov> johnthetubaguy, were you looking at improving tests for cells - i kinda saw something last week or so
14:32:22 <johnthetubaguy> mriedem: its at the top of the "must have slots" list
14:32:30 <bauzas> I was seeing a discussion about some kind of cells for the crossproject track
14:32:38 <mriedem> johnthetubaguy: ha, i wasn't scrolled up high enough apparently
14:32:54 <johnthetubaguy> mriedem: no worries
14:33:02 <bauzas> wouldn't it be interesting to discuss about cells at a general openstack level, if we consider rebooting them ?
14:33:25 <dansmith> bauzas: this conversation is not about that,
14:33:33 <dansmith> bauzas: it's about fixing what we have, not rebooting cells entirely
14:33:38 <alaski> bauzas: the first priority is making them a first class citizen in Nova
14:34:01 <bauzas> dansmith: https://etherpad.openstack.org/p/kilo-crossproject-summit-topics L114
14:34:13 <bauzas> alaski: dansmith: ack
14:34:21 <johnthetubaguy> bauzas: yeah, its because of an email thread
14:34:41 <dansmith> okay, I'm talking about the cells discussion on our list
14:34:42 <bauzas> johnthetubaguy: but I totally understand that cells can be discussed in both sessions
14:34:44 <johnthetubaguy> bauzas: but yeah, nova one is talking about different things really
14:34:52 <bauzas> johnthetubaguy: got it
14:34:52 <dansmith> I think the cross-project one is in  response to the cascading stuff
14:35:00 <johnthetubaguy> yeah
14:35:05 <bauzas> indeed
14:35:22 <johnthetubaguy> anyways, seems like we are done, we can take other things to the usual channel
14:35:26 <johnthetubaguy> thanks all
14:35:35 <johnthetubaguy> #endmeeting