21:02:13 <russellb> #startmeeting nova
21:02:14 <openstack> Meeting started Thu Oct 10 21:02:13 2013 UTC and is due to finish in 60 minutes.  The chair is russellb. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:02:16 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:02:18 <openstack> The meeting name has been set to 'nova'
21:02:18 <russellb> hello, everyone!
21:02:27 <melwitt> hi
21:02:29 <alaski> hi
21:02:30 <dripton> hi
21:02:42 <russellb> #topic havana
21:02:43 <sdague> o/
21:02:49 <russellb> havana release is one week from today
21:03:00 <russellb> we decided to go ahead and do an RC2 for nova
21:03:02 <russellb> #link https://launchpad.net/nova/+milestone/havana-rc2
21:03:24 <russellb> i pulled in some stuff that has been fixed and was on havana-rc-potential
21:03:35 <jog0> was there any one bug that necessitated that
21:03:37 <russellb> if there is anything else that you feel should go in, please let me know
21:03:41 <russellb> jog0: nope
21:03:45 <mikal> Morning
21:03:45 <jog0> russellb: nice
21:03:57 <russellb> i think the release has been pretty solid
21:04:02 <tjones> russleb: via email, or PM, or let you know here?
21:04:04 <russellb> and these were a set of things that were nice to get in
21:04:10 <russellb> tjones: any
21:04:27 <russellb> but we *really* need to get anything that's a nice to have in by sometime tomorrow
21:04:28 <sdague> russellb: the boto unit test patch come back over?
21:04:38 <russellb> anything after that is only absolutely critical stuff
21:04:49 <mriedem> sdague: yeah, dansmith had a cherry pick
21:04:52 <sdague> cool
21:04:57 <tjones> russelb: got it
21:05:17 <comstud> dansmith: is that obj not detecting sys meta changes be something targetted to rc2?
21:05:41 <russellb> it's not targeted, no
21:05:43 <russellb> is there a bug for it?
21:05:45 <comstud> i didn't see what conditions would catch it
21:05:46 <comstud> yeah
21:05:48 <comstud> should be
21:06:09 <comstud> https://bugs.launchpad.net/nova/+bug/1237102\
21:06:11 <uvirtbot> Launchpad bug 1237102 in nova "Conductor does not properly copy objects during change tracking" [Medium,Fix committed]
21:06:13 <comstud> without the \
21:06:36 <comstud> Maybe it's not currently causing a problem
21:06:56 <russellb> comstud: dansmith ok well you guys take a look and let me know if you think its impact justifies going into rc2
21:07:00 <comstud> yeah
21:07:24 <russellb> the other havana todo in the next week is finalizing our release notes
21:07:26 <russellb> #link https://wiki.openstack.org/wiki/ReleaseNotes/Havana#OpenStack_Compute_.28Nova.29
21:07:38 <russellb> I think it's really close to complete at this point
21:07:44 <russellb> we're missing xenapi features and hyper-v features
21:07:48 <russellb> and that's it
21:07:53 <russellb> and anything you guys find when you review it :-)
21:07:57 <russellb> so take a look
21:08:16 <russellb> not just the feature list, but also anything appropriate for upgrade notes
21:08:57 <russellb> i imagine a *lot* of people will see/read this list, so it's worth making sure it's good/complete/correct/etc
21:09:23 <russellb> that's all i've got on havana stuff
21:09:27 <russellb> any comments/questions?
21:10:16 <russellb> #topic icehouse design summit
21:10:25 <russellb> #link http://summit.openstack.org/
21:10:29 <russellb> we already have more proposals than time slots :)
21:10:39 <russellb> #info deadline for proposals, oct 17
21:10:58 <russellb> please get your stuff in by next week, because we'll need time to sort through them and figure out the schedule
21:11:03 <russellb> i talked about making the schedule last week
21:11:19 <russellb> i decided i want to do it with just a small group, to make sure we can get it done quickly, since there really isn't much time
21:11:44 <russellb> but everyone is certainly encouraged to comment on proposals with any thoughts you have, and that will influence the end result
21:12:37 <yjiang5_> russellb: I remember you will prefer no sharing this time , right?
21:12:57 <russellb> yjiang5_: ideally, yes.  in the case of duplicates, we might merge
21:13:06 <russellb> but otherwise, i'd rather avoid merging, yes
21:13:13 <russellb> which means more might get rejected
21:13:18 <yjiang5_> russellb: got it.
21:13:28 <russellb> but i suspect there will be cases where a mailing list discussion is probably OK
21:13:39 <comstud> we won't merge unless we merge
21:13:40 <russellb> something not that controversial, or not as complex, or whatever
21:13:47 <russellb> comstud: that's accurate
21:13:51 <comstud> 100%
21:13:51 <mikal> Heh
21:14:00 <mikal> DB might make sense for a merge, just to be contrary
21:14:08 <russellb> mikal: with what?
21:14:11 <mikal> There are two separate groups working on different DB things, but not enough for two serssions
21:14:49 <russellb> i saw yours, is there another session though?
21:15:05 <mikal> I don't think so. I was goign to trick Boris into using some of mine.
21:15:05 <mikal> '
21:15:06 <russellb> there is this DB one, but it should probably be by itself: http://summit.openstack.org/cfp/details/181
21:15:16 <mikal> But I interpret this conversation as saying you'd prefer that didn't happen?
21:15:21 <russellb> mikal: oh OK, well that would work.  if so, please just update your description
21:15:36 <mikal> ok
21:15:49 <russellb> no it's more for things where it's really different topics ... or kind of related
21:16:07 <russellb> anyway, hard to speak about this in general terms
21:16:08 <mikal> Ahhh, ok
21:16:21 <russellb> i think the thing that triggered this was scheduler folks
21:16:23 <mikal> http://summit.openstack.org/cfp/details/16 is the one I was thinking of
21:16:29 <russellb> i ended up jamming them together in a few sessions last time and they didn't like it :)
21:16:48 <russellb> it didn't really work great either for some of it
21:17:20 <dansmith> oof, sorry, a call went long
21:17:33 <russellb> dansmith: unacceptable
21:17:45 <dansmith> russellb: you know who to blame :)
21:17:49 <russellb> heh
21:17:56 <russellb> that's all i have on the design summit for now
21:18:02 <russellb> unless anyone would like to talk about specific sessions
21:18:36 <russellb> johnthetubaguy: you have 2 that might be duplicates ... "Prestart servers for faster boot times" and "Improve VM boot times, by pre-starting VMs"
21:18:43 <russellb> maybe forgot that you already proposed it?  :)
21:18:50 <johnthetubaguy> russellb: oops, sorry
21:18:54 <russellb> np
21:19:12 <johnthetubaguy> that explains why the blueprint was already created
21:19:54 <russellb> #topic sub-team repots
21:19:58 <russellb> #undo
21:19:59 <openstack> Removing item from minutes: <ircmeeting.items.Topic object at 0x31325d0>
21:20:00 <russellb> #topic sub-team reports
21:20:08 <russellb> any updates from sub-teams?
21:20:25 <johnthetubaguy> no xenapi meeting this week, but the summit session etherpad is getting worked on offline
21:20:45 <mriedem> mrodden: here?
21:20:53 <russellb> johnthetubaguy: sounds good, i saw our sessions start to roll in
21:21:29 <mriedem> i guess we could talk about powervm CI
21:21:36 <russellb> cool, how's that coming
21:21:43 <mriedem> mrodden: ?
21:21:55 <russellb> k, will see if he pops in :)
21:22:07 <russellb> tjones: i saw a message on the -infra list about vmware CI, sounds like that's close?
21:22:07 <mrodden> hi
21:22:11 <russellb> mrodden: hi!
21:22:30 <mrodden> i know last time there was discussion on what exactly the requirements were for CI
21:22:33 <tjones> yep - the name we wanted to use (SmokeSignal) was denied.
21:22:42 <russellb> tjones: heh
21:22:48 <russellb> tjones: how about ... vmware CI
21:22:52 <tjones> that's the last blocker ;-)  so we are proposing vSmoker
21:23:03 <russellb> mrodden: ok, so are you guys waiting on a clearer definition on that from me then?
21:23:07 <mrodden> was there a ML post or wiki page that came out of that?
21:23:08 <dansmith> mrodden: we decided a couple weeks ago that it is "running tempest" I think
21:23:20 <russellb> mrodden: no post, just chatter in the meeting
21:23:22 <mriedem> dansmith: but how scoped is running tempest
21:23:24 <mrodden> yes.. but there are sets of tempest tests
21:23:26 <russellb> running tempest though, effectively, is the goal
21:23:34 <dansmith> mriedem: not disabling things that don't work :D
21:23:55 <mriedem> dansmith: well for example there are tests against APIs that aren't implemented, e.g. nova diagnostics
21:23:58 <russellb> johnthetubaguy: which btw, we need to look at for the xenapi driver in smokestack (running tempest)
21:24:00 <dansmith> mriedem: unless it's a fundamental thing the platform doesn't support, but then upstream tempest should be aware
21:24:25 <russellb> johnthetubaguy: i think we really want to see the CI for all drivers running tempest
21:24:26 <dansmith> mriedem: yep, so I think that's fair game to exclude
21:24:42 <russellb> dansmith: mriedem agreed
21:24:43 <mriedem> i can tell you that right now we're running 1152 tempest tests against powervm with db2
21:24:47 <dansmith> mriedem: maybe publish the tempest config you're using?
21:24:48 <mriedem> so it's not peanuts
21:24:48 <johnthetubaguy> russellb: yep, well looking for more normal based tempest running, outside of smokestack too
21:24:55 <russellb> johnthetubaguy: great
21:25:03 <dansmith> mriedem: for discussion/agreement
21:25:12 <tjones> russellb: vmware CI….  catchy ;-)  sure we can use that.  will have +1 by next meeting, −1 maybe too.  I don't know the exact # we are running but it is > 1000
21:25:25 <johnthetubaguy> russellb: totally agree that tempest is the right goal, not quite there yet, but there is hope yet
21:25:26 <russellb> tjones: vSmoker is cool, too
21:25:30 <russellb> tjones: 1000 nodes?
21:25:36 <russellb> or 1000 tests?
21:25:42 <russellb> johnthetubaguy: *nods*
21:25:53 <tjones> russellb: > 100 tests.   not sure how many nodes.  can find out the config
21:26:00 <tjones> s/100/1000/
21:26:06 <russellb> tjones: doesn't matter, was just clarifying
21:26:10 <tjones> heh
21:26:11 <russellb> tjones: as long as you keep up :-)
21:26:20 <russellb> with the patch flow that is
21:26:24 <mriedem> russellb: dansmith: i think we have a few open questions for powervm that we're still working through like which OS are we going to run on, how are we going to deploy openstack (rdo?), tempest tests, etc
21:26:40 <mriedem> i really need to figure out what smokestack does
21:26:53 <dansmith> mriedem: okay
21:27:03 <johnthetubaguy> why not devstack + tempest as a starting point?
21:27:06 <dansmith> mriedem: I think that smokestack builds packages and then installs them
21:27:26 <russellb> yeah, using puppet
21:27:27 <mriedem> johnthetubaguy: question to me?
21:27:33 <neelashah> dansmith: it builds packages on the fly?
21:27:35 <dansmith> mriedem: johnthetubaguy yeah devstack would be good if it's doable
21:27:38 <dansmith> neelashah: yeah
21:27:49 <johnthetubaguy> mriedem: yeah, its probably easier
21:27:49 <jog0> mriedem: *ideally* the answer to all those questions should be as close to what a user would do as possible.
21:27:53 <mrodden> is the smokestack source available?
21:27:56 <neelashah> is that available someplace to take a look or proprietary?
21:28:00 <mriedem> dansmith: johnthetubaguy: agreed on devstack
21:28:03 <russellb> mrodden: yeah it's all open
21:28:34 <russellb> smokestack.org is not what i wanted ...
21:28:41 <mriedem> internally we're building packages too and running against rhel 6, naturally i felt like i had to duplicate that for CI but i don't think so
21:28:47 <russellb> mrodden: https://smokestack.openstack.org/about
21:28:56 <mrodden> russellb: cool
21:29:37 <mriedem> i'd like to work db2 into this as a backend so we can kill 2 birds with one stone there, but that's probably more of a stretch at this point, although db2 is a plan for icehouse
21:29:40 <dansmith> mriedem: well, being close to how users use your platform is a good plan for sure
21:30:08 <dansmith> mriedem: i.e. building packages in the same way you build your real ones
21:30:15 <mriedem> dansmith: yeah, so i'm actually thinking fedora 19 + smokestack + db2
21:30:27 <mrodden> +1
21:30:32 <mriedem> but mysql would work too at first
21:30:47 <dansmith> I'd rather you use db2
21:30:50 <dansmith> so I can laugh at you
21:30:50 <mriedem> me too
21:30:52 <dansmith> but either way
21:30:53 <mrodden> lol
21:30:54 <dripton> mriedem: wouldn't hurt to have both.  That way you could tell your db2 problems from general problems.
21:31:05 <mriedem> dripton: totally agree
21:31:09 <dansmith> dripton: surely, but just more resources that way
21:31:21 <mriedem> anyway, this is my thought process right now, we still have to tackle the scale issue
21:31:23 <dansmith> dripton: I think these guys are using surplus Powermac G5s, so...
21:31:31 <russellb> eep
21:31:37 <dansmith> (just kidding)
21:31:45 <dripton> so by the time smokestack finishes the patch is already approved
21:31:58 <dansmith> that's a good point
21:32:17 <dansmith> russellb: we probably need in the guidelines a "a vote 48 hours later is not good enough"
21:32:21 <neelashah> dansmith: actually, are there guidelines o nthat?
21:32:33 <neelashah> how long before the tests have to report back in?
21:32:34 <dansmith> neelashah: we need some for sure
21:32:43 <mrodden> thats something that i would like to see
21:32:50 <neelashah> +1
21:32:52 <mrodden> we need to kind of figure out what we need to be shooting for
21:32:58 <johnthetubaguy> smokestack is going to take longer if it starts doing tempest instead of torpedo, so there is some work there too I think
21:32:59 <dansmith> neelashah: have you guys done any back of the envelope calculations to figure out how fast you'll be able to report?
21:33:29 <dansmith> neelashah: assuming you have an idea on number of nodes you'll have, which maybe you don't...
21:33:36 <neelashah> dansmith: we haven't yet - starting to put the hw/infrastructure plans so this will be good info to have to plan for
21:33:40 <dansmith> neelashah: okay
21:33:54 <dansmith> russellb: anyway, we can probably work this in the background if you want to continue
21:34:01 <mriedem> agreed
21:34:08 <russellb> OK, well, good progress i think
21:34:13 <russellb> and it's clear we need some details written up on the wiki
21:34:16 <dansmith> yeah
21:34:17 <russellb> what tests need to run
21:34:23 <russellb> and some speed requirements
21:34:35 <russellb> so i've got that on my todo
21:34:46 <russellb> #topic open discussion
21:34:56 <russellb> beagles: you around?
21:35:03 <russellb> last week you brought up your neutron vs nova-network doc
21:35:07 <russellb> any updates on it?
21:35:14 <russellb> #link https://docs.google.com/document/d/1E9CiG8fmcn0FMI15zpE0dRQ1CcFhXV8TxL7mU7vgWv8/edit
21:35:33 <russellb> i failed and didn't get you feedback this week ... sorry
21:35:47 <beagles> hi guys
21:35:48 <beagles> sorry
21:35:54 <beagles> was interrupted
21:35:58 <russellb> np!
21:36:09 <beagles> I'm still working on the tempest/testing section
21:36:23 <beagles> along with a survey of current outstanding issues that are arguably related
21:36:50 <jog0> beagles: btw we have basic perforamnce testing and am working on getting it up on neutron
21:37:03 <beagles> jog0: nice!
21:37:06 <jog0> https://review.openstack.org/#/c/50449/
21:37:53 <russellb> speaking of new gate jobs ... we need to get that cells job going
21:37:59 <russellb> comstud: alaski ^
21:38:09 <beagles> there was some chatter about getting me to Hong Kong to have a session amongst the Neutron crowd.
21:38:17 <russellb> beagles: that would rock
21:38:18 <beagles> I suspect that is not going to happen though
21:38:27 <beagles> but you never know
21:38:29 <russellb> beagles: i was about to ask, since you're going somewhere else far soon right?
21:38:48 <beagles> russellb, yes, Tel Aviv on Saturday for a week
21:38:54 <russellb> safe travels :)
21:39:05 <russellb> well regarding the summit, i will at least try to do my part to catch up with neutron folks
21:39:07 <johnthetubaguy> russellb: I know there are plans around fixing the cells job, I may help out too, depending on how stuff goes
21:39:15 <russellb> johnthetubaguy: excellent
21:39:26 <beagles> russellb, thks!
21:39:29 <russellb> johnthetubaguy: i just want to get the bare minimum running and gating, and then we can expand from there
21:39:30 <johnthetubaguy> russellb: cells issue is also getting all the features working, so it can be full tempest!
21:39:32 <alaski> russellb: yeah.  I can get a PR for that together
21:39:44 <russellb> johnthetubaguy: just devstack exercises would be OK to start with, i think that's how it's configured now
21:39:51 <johnthetubaguy> russellb: agreed, simple tests would be good first
21:39:54 <russellb> alaski: not sure what a PR is, but OK!  :-)
21:40:15 <alaski> russellb: sorry, pull request, non gerrit terminology :|
21:40:16 <dansmith> pull request?
21:40:19 <russellb> ha
21:40:21 <dansmith> heh
21:40:23 <russellb> sorry.
21:40:48 <comstud> i need to check experimental to see if the cells gate still passes
21:40:49 <comstud> heh
21:40:56 <comstud> but it does only run devstack exercises and should pass
21:41:09 * johnthetubaguy was wondering why we needed a press release
21:41:30 <russellb> if it's working, now that we're past the big freeze rush, should move it back to running on everything for a little bit, and then can switch to gating
21:41:31 <russellb> johnthetubaguy: hm?
21:41:38 <russellb> oh
21:41:41 <russellb> gah i'm slow.
21:41:43 <comstud> haha
21:42:20 <russellb> alright, any other topics?
21:42:24 <jog0> In other news, as of a few minutes in the future(https://review.openstack.org/#/c/49853/) novaclient will be python33 compat
21:42:28 <russellb> nice work on havana everyone, coming together nicely
21:42:30 <dripton> woot!
21:42:43 <russellb> jog0: nice!
21:42:47 <jog0> and gating will happen shortly after  https://review.openstack.org/#/c/51058/1
21:43:01 <russellb> jog0: i need to do a novaclient release soon anyway, for the final havana bits
21:43:06 <jog0> also got the keystone token caching working again in novaclient
21:44:25 <jog0> so less round trips to keystone needed
21:44:32 <russellb> good stuff
21:44:41 <dansmith> jog0: now you're just bragging
21:45:06 <jog0> dansmith: I stayed up late last night doing it ... so  you cought me
21:45:10 <dansmith> heh
21:45:18 <russellb> well your work is very much appreciated :-)
21:45:55 <russellb> i noticed our review turnaround averages are looking really good right now ... http://russellbryant.net/openstack-stats/nova-openreviews.html
21:45:57 <russellb> so yay
21:46:14 <russellb> not to say that some things haven't waited a long time ... but overall, better than we were doing toward the end of the cycle
21:46:24 <russellb> perhaps not surprising, but yay anyway
21:46:47 <russellb> ok, thanks everyone for your time!
21:47:02 <russellb> #endmeeting