15:00:21 <n0ano> #startmeeting scheduler
15:00:22 <openstack> Meeting started Tue Dec 10 15:00:21 2013 UTC and is due to finish in 60 minutes.  The chair is n0ano. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:23 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:26 <openstack> The meeting name has been set to 'scheduler'
15:00:32 <n0ano> anyone here for the scheduler meeting?
15:00:36 <MikeSpreitzer> yes
15:01:20 <alaski> o/
15:01:25 <trey_h> si
15:01:29 <glikson> hi
15:01:34 <cloudon_> yes
15:01:40 <cfriesen_> yup
15:01:56 <llu-laptop> hello
15:02:21 <n0ano> #topic Scheduler as a service
15:02:37 <n0ano> anyone have a chance to go over the BP I referred to in the agenda?
15:03:01 <alaski> do you have a link?
15:03:14 <jgallard> hi all : https://blueprints.launchpad.net/nova/+spec/forklift-scheduler-breakout
15:03:16 <n0ano> #link https://blueprints.launchpad.net/nova/+spec/forklift-scheduler-breakout
15:03:29 <alaski> thanks
15:03:49 <toan-tran> hi all
15:03:53 <n0ano> note that `as a service' is a little bit of a miss-nomer, the current effor is mainly to split the scheduler code into it's own repository...
15:04:08 <n0ano> after that it should be possible to turn it into a separate service
15:04:16 <MikeSpreitzer> That bp refers to the etherpad we all saw and the ML discussion
15:04:47 <n0ano> MikeSpreitzer, indeed, I want to see if there are any issues/concerns as we are actively working on implementing that BP
15:04:57 <cloudon1> think the agenda referenced the no-db-scheduler BP?
15:05:12 <MikeSpreitzer> that was item 2
15:05:14 <n0ano> cloudon1, second topic for today
15:05:23 <cloudon1> ok, thanks
15:05:56 <toan-tran> MikeSpreitzer: but I think we could fusion the two
15:06:02 <toan-tran> actually there are two approaches:
15:06:28 <toan-tran> one with Robert idea of forflifting first, separate DB latter
15:06:38 <glikson> n0ano: I'm curious, who exactly is "actively working on implementing that BP"?
15:06:47 <toan-tran> one with Boris': separate DB first , forklift latter
15:07:14 <n0ano> glikson, me for one, I'm trying to create the initial scheduler only repo that everyone can then start working on
15:07:27 <glikson> n0ano: cool
15:08:13 <n0ano> I actually have the first cut ready (https://github.com/n0ano/gantt), I'd like to move it to an official place soon and open it up to eveyone
15:08:23 * n0ano git foo is getting better
15:08:28 <cfriesen_> n0ano:  are you on the forklift first, change other stuff later"
15:08:31 <cfriesen_> plan?
15:09:10 <n0ano> cfriesen_, yes, that will cause a little duplicated updating but otherwise we might get delayed waiting for no-db to be ready
15:09:48 <llu-laptop> n0ano: does the new repo means new stackforge projects in Jenkins?
15:09:52 <glikson> I guess it might be useful if someone could outline the work items on launchpad, so that people who are interested to contribute could have a better view of what is going on..
15:09:57 <garyk> n0ano: gantt? as in gantt chart :)
15:10:15 <alaski> n0ano: I agree.  forklift first means we're not waiting, boris can keep working while it's happening
15:10:19 <n0ano> garyk, my alterntaive was pert if gantt doesn't work :-)
15:10:42 <n0ano> glikson, that's what the refterenced BP points to, a detailed etherpad with the forklift steps
15:11:53 <n0ano> llu-laptop, yes, I believe eventually a new project will be created, once we move to scheduler as a service
15:12:14 <llu-laptop> how about putting the TODO list in the workItem section of BP to make it more easily to be found
15:12:49 <n0ano> llu-laptop, I would suggest that the etherpad is the appropriate place for that, easily referenced from the BP
15:13:09 <llu-laptop> n0ano: then before that, what about if someone want to propose a patch to the new repo?
15:13:28 <n0ano> llu-laptop, not sure what your question is
15:14:26 <garyk> n0ano: question is how to get this into stackforge and then having all of the relevant processed in tack
15:15:28 <n0ano> garyk, llu-laptop I'm working with russellb to create the initial, bare bones repo, once we have that my assumption is it gets linked into the infrastructure so that gerrit workflows will work against it
15:15:46 <garyk> ok, thanks for the clarification
15:16:05 <n0ano> we absolutely want the current work flow, with patch approval process, to happen as soon as possible
15:16:31 * n0ano bootstrapping is always a little messy
15:17:52 <alaski> llu-laptop: For now you would propose patches against the current scheduler code, but depending on the scope of the patch you may be asked to wait until the new repo is up and re propose there
15:18:20 <n0ano> alaski, +1
15:18:36 <llu-laptop> n0ano, alaski: got that, thanks
15:19:02 <n0ano> the goal is not to stop development (at the expense of some extra work to update two repos potentially)
15:19:39 <n0ano> sounds like we all understand, if there are no other issues we can move on
15:19:54 <n0ano> boris-42, yt?
15:20:28 <n0ano> #topic instance groups
15:20:45 <n0ano> while waiting to see if boris-42 is around, garyk do you have an update on instance groups
15:20:46 <boris-42> n0ano ouch hi
15:20:56 <n0ano> garyk, hold off a sec
15:21:00 <n0ano> boris-42, tnx
15:21:06 <n0ano> #topic no-db scheduler
15:21:12 <boris-42> n0ano so I am here
15:21:24 <n0ano> boris-42, I saw you link to the BP, nice write up, what the status of the changes
15:21:41 <boris-42> n0ano alaski garyk did you read the doc?
15:21:54 <hnarkaytis> let me give update on patch status
15:22:07 <hnarkaytis> design docs were shared yesterday
15:22:10 <n0ano> boris-42, I did
15:22:27 <hnarkaytis> right now work focused on test suite
15:22:48 <boris-42> n0ano yep let hnarkaytis share with us
15:23:00 <alaski> boris-42: I haven't yet, sorry.  let me go find it
15:23:27 <n0ano> cloudon1, this was your interest, do you have any questions?
15:23:33 <hnarkaytis> there is a plan to add two more patches that will complete integration of the no-db schema
15:23:51 <hnarkaytis> that's it - patches are ready for reviews
15:24:07 <n0ano> cool, I like the exmphasis on testing, good point
15:24:12 <cfriesen_> do we have performance numbers?
15:24:29 <hnarkaytis> yes - we prepared extended doc on this
15:24:42 <hnarkaytis> I will share draft tomorrow
15:24:59 <hnarkaytis> 100x gain for memchached
15:25:21 <n0ano> hnarkaytis, one point, sounds like you have more development to do but also have patches ready for review, when do you think you'll be complete?
15:25:49 <hnarkaytis> end-to-end story will be complete till end of this week
15:26:04 <hnarkaytis> right now we presented only a Synchronizer object
15:26:33 <hnarkaytis> two additional patches will replace existing approach on new one
15:26:36 <MikeSpreitzer> BTW, what is URL for "design docs were shared yesterday"
15:26:49 <hnarkaytis> https://etherpad.openstack.org/p/scheduler-design-proposal
15:27:00 <hnarkaytis> https://docs.google.com/a/mirantis.com/document/d/1irmDDYWWKWAGWECX8bozu8AAmzgQxMCAAdjhk53L9aM/edit
15:27:19 <hnarkaytis> I posted into mail thread
15:27:27 <MikeSpreitzer> thanks
15:27:37 <MikeSpreitzer> mind if I add those to the BP?
15:27:53 <hnarkaytis> this mail thread - [openstack-dev] [Nova][Schduler] Volunteers wanted for a modest proposal for an external scheduler in our lifetime
15:27:55 <n0ano> MikeSpreitzer, note, those are the same doc, just different locations
15:28:45 <MikeSpreitzer> which two are the same?
15:29:02 <hnarkaytis> etherpad and google doc are the same
15:29:06 <n0ano> https://etherpad & https://docs.google.com just given by hnarkaytis
15:29:36 <MikeSpreitzer> that's odd, they look different to me
15:29:54 <hnarkaytis> this is copy/paste - read any of them
15:29:58 <n0ano> I just looked an, barring formatting, they have the same content
15:30:10 <MikeSpreitzer> OK
15:30:35 <MikeSpreitzer> So if the BP references just one of those, which should it be?
15:30:51 <hnarkaytis> as soon as last two patches will be ready - we will notify all subscribed reviewers
15:31:01 <n0ano> I'd vote for the etherpad, keep it in the family
15:31:19 <hnarkaytis> yes - lets use etherpad for references
15:31:21 <llu-laptop> hnarkaytis: how about update the BP with the latest URL?
15:31:48 <hnarkaytis> ok - will do
15:31:49 <MikeSpreitzer> Yeah, "Set the URL for this specification"
15:33:27 <n0ano> hnarkaytis, boris-42 - good work, tnx for the effort, anyone have any other questions for them today?
15:34:19 <n0ano> moving on
15:34:26 <n0ano> #topic instance groups
15:34:32 <n0ano> garyk, anything to say?
15:36:24 <hnarkaytis> just a quick note on performance test issue - document with performance numbers is referenced in BP https://docs.google.com/a/mirantis.com/document/d/1_DRv7it_mwalEZzLy5WO92TJcummpmWL4NWsWf0UWiQ/edit
15:36:53 <n0ano> looks like I scared him off, I'll talk to garyk via email
15:37:04 <n0ano> #topic mid-cycle meetup
15:37:31 <n0ano> there will be a meetup in Feb in Utah, I'm going to try to attend, will anyone else be there?
15:37:48 <alaski> I should be there
15:39:14 <n0ano> looks like the scheduler won't have a major presence there but hopefully will be a good meetup anyway
15:39:39 <n0ano> #opens
15:39:49 <n0ano> Any new items anyone wants to raise today?
15:40:50 <llu-laptop> just one suggestion, how about next time we put agenda to #link https://wiki.openstack.org/wiki/Meetings/Scheduler?
15:41:10 <llu-laptop> sorry, https://wiki.openstack.org/wiki/Meetings/Scheduler
15:41:34 <n0ano> llu-laptop, I can try but I'm notoriously lax about updating Wiki, we'll see if I can motivate myself to do that
15:41:41 * n0ano like email :-)
15:42:59 <n0ano> OK, I want to thank everyone and we'll talk next week (probably cancel the two after that but we can discuss that next week)
15:43:15 <n0ano> #endmeeting