15:00:45 <n0ano> #startmeeting gantt
15:00:46 <openstack> Meeting started Tue Feb 25 15:00:45 2014 UTC and is due to finish in 60 minutes.  The chair is n0ano. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:47 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:49 <openstack> The meeting name has been set to 'gantt'
15:01:02 <n0ano> anyone here to talk abou the scheduler (or did I scare everyone off :-)
15:01:50 <mspreitz> I am here
15:01:53 <bauzas> o/
15:02:14 <n0ano> #topic opens
15:02:39 <n0ano> As I said, I don't have any set agenda (I was off sick all last week), anything you guys want to talk about?
15:02:51 <bauzas> https://etherpad.openstack.org/p/icehouse-external-scheduler
15:03:16 <bauzas> we agreed on plan B the previous week
15:03:43 <bauzas> there is a blueprint for task #1
15:04:08 <bauzas> I would like to contribute to task #2, is there any BP ?
15:04:33 <bauzas> and if so, we should amend the etherpad :)
15:04:48 <bauzas> I know we're in FeatureFreeze
15:04:57 <bauzas> sorry, FeatureProposalFreeze
15:05:10 <n0ano> not that I know of, if you have ideas you can add to the etherpad or, if you feel ambitious, add a BP for that
15:05:41 <n0ano> I think the reality is that this work is going to be for Juno, we've pretty much missed the Icehous window by now
15:05:49 <bauzas> n0ano: ok, I'll see if I can fill in some details within another etherpad
15:06:04 <johnthetubaguy> yeah, its juno now for this stuff
15:06:04 <bauzas> n0ano: yup, we all know :)
15:06:13 <n0ano> so I'm not too concerned about timing issues, we just do things as quick as we can
15:06:16 <johnthetubaguy> are we all happy with the approach now?
15:06:30 <bauzas> I'm fine with it
15:06:45 <johnthetubaguy> I think its a good approach
15:06:51 <johnthetubaguy> even after sleeping on it
15:07:05 <n0ano> my only concern is that we make progress on it, just `cleaning up nova' is easy to push off for another day
15:07:15 <bauzas> as gantt is not nova, I think we could still work on delivering a library
15:07:30 <johnthetubaguy> well, I think thats when we make the split though?
15:07:40 <johnthetubaguy> that forces that one, once we get very close
15:08:27 <n0ano> it looks like comleteing both task 1 and 2 is when we can do the true split
15:08:49 <bauzas> IMHO, there are 2 things which can be done while the Nova freze
15:08:52 <bauzas> freeze
15:09:12 <bauzas> 1/ do the cleanup of nova calls within Gantt
15:09:35 <bauzas> 2/ provide an external library for the resource tracker
15:09:47 <johnthetubaguy> you can get the patches up while we are frozen, but the idea of being frozen is to get bugs fixed
15:09:51 <bauzas> both are within Gantt's scope
15:10:00 <n0ano> johnthetubaguy, +1
15:12:12 <johnthetubaguy> OK, so any more for today then?
15:12:22 <n0ano> not from me
15:12:30 <johnthetubaguy> do you want to raise blueprints for the scheduler lib?
15:12:32 <mspreitz> one procedural issue
15:12:34 <johnthetubaguy> just so its tracked?
15:12:41 <mspreitz> Does any wiki page point to the archive of these meetings?
15:12:44 <bauzas> johnthetubaguy: I can handle it
15:13:02 <n0ano> mspreitz, yes
15:13:04 <johnthetubaguy> bauzas: cool, I ping me when you want it approving, etc
15:13:08 <n0ano> #link http://eavesdrop.openstack.org/meetings/gantt/2014/
15:13:20 <bauzas> johnthetubaguy: well OK
15:13:27 <mspreitz> what wiki page points there?
15:13:31 <bauzas> johnthetubaguy: that will be matter of collecting all info
15:13:40 <johnthetubaguy> bauzas: which info?
15:13:44 <n0ano> cool meeting, bauzas got all the ARs
15:13:46 <bauzas> johnthetubaguy: and writing the rationale
15:14:03 <johnthetubaguy> OK, I don't mind drafting something quick, if that helps?
15:14:18 <bauzas> johnthetubaguy: do you prefer me to quickly draft something then ?
15:14:32 <bauzas> johnthetubaguy: because I was preferring to write a wikipage
15:14:43 <Yathi> also regarding the gantt split work and if any blueprints require owners please include me i volunteer
15:14:45 <mspreitz> n0ano: https://wiki.openstack.org/wiki/Meetings/Scheduler does not point to the gantt meetings
15:14:48 <bauzas> johnthetubaguy: but the first one should also good to me :)
15:14:49 <johnthetubaguy> bauzas: OK, I was hoping to keep it short really
15:14:52 <n0ano> mspreitz, my bad, I changed the meeting name and didn't update the meetings wiki with that info, I'll fix it
15:14:59 <mspreitz> thanks
15:15:05 <bauzas> johnthetubaguy: well, OK, will do then
15:15:48 <n0ano> OK, bauzas does all the work, anything else today?
15:16:05 <johnthetubaguy> bauzas: does "Create Scheduler Python Library" work OK?
15:16:15 <bauzas> johnthetubaguy: +1
15:16:22 <johnthetubaguy> cool, thanks
15:17:08 <johnthetubaguy> https://blueprints.launchpad.net/nova/+spec/scheduler-lib
15:17:38 <johnthetubaguy> I will fill it out, feel free to jump on it with ideas
15:17:47 <bauzas> johnthetubaguy: ok, will do
15:19:13 <bauzas> to all, do we consider patching oslo.config for managing different projects ?
15:19:28 <bauzas> that's something I began to work on
15:19:32 <johnthetubaguy> bauzas: doesn't seem worth it right now
15:19:53 <johnthetubaguy> bauzas: gnatt should have its own config right? it can just match what nova has today, I assume?
15:20:06 <n0ano> bauzas, as a separate effort maybe, don't see how we should depend upon that for this work
15:20:19 <n0ano> johnthetubaguy, +1
15:20:20 <bauzas> johnthetubaguy: well, that's pretty hard to import both nova and gantt
15:20:37 <bauzas> johnthetubaguy: I was thinking it as some way to ease the move
15:20:49 <johnthetubaguy> bauzas: but with the cleaner split, it should "just work" TM
15:20:54 <n0ano> bauzas, what do you mean, I though everybody already included oslo.config
15:20:56 <bauzas> johnthetubaguy: agree
15:21:20 <bauzas> n0ano: I mean that all registered opts are duplicated in between Gantt and Nova because of the same scope
15:21:39 <bauzas> n0ano: there is one global cfg object
15:21:49 <johnthetubaguy> yep
15:21:51 * erecio slaps erecio around a bit with a large trout
15:21:53 <johnthetubaguy> and two config files
15:22:18 <bauzas> with possibly same opt names :)
15:22:23 <bauzas> and groups
15:22:25 <n0ano> do we want to create a gantt config name space or would that be too much of a change
15:22:51 <bauzas> nah, I think it's oslo.config duty to handle different namespaces
15:23:22 <Yathi> shouldnt be hard to change the code everywhere to include a new namespace
15:23:27 <n0ano> I believe it does, the problem is, right now, you'd have nova & gantt defining the same names in the same name space, oslo.config gets upset about that
15:23:48 <bauzas> n0ano: yup, and here is why I'm proposing to focus on that side
15:24:08 <bauzas> n0ano: I'm facing same issue with climate-nova
15:24:29 <bauzas> n0ano: the workaround is to deregister all opts from one project, but that's hard to manage
15:24:44 <n0ano> the simple solution is to just create a new name space, be it gantt or climate, but that is a user visible change
15:25:02 <johnthetubaguy> hang on
15:25:11 <johnthetubaguy> in gantt, it will be in gantt.conf
15:25:15 <johnthetubaguy> in nova its in nova.conf
15:25:23 <johnthetubaguy> and gnatt will never access anything in nova.conf
15:25:31 <johnthetubaguy> and nova will not access anything in gantt.conf
15:25:36 <johnthetubaguy> or am I missing something here?
15:25:58 <n0ano> johnthetubaguy, so you are proposing a user visible change, the user now has to change all of her config files
15:26:00 <bauzas> johnthetubaguy: the cfg object is the same for both
15:26:22 <bauzas> when you import libs, it takes the global cfg object
15:26:37 <bauzas> so, within oslo code, it will register some opts
15:26:46 <johnthetubaguy> n0ano: yes, if you use gantt you will need to use gantt.conf
15:26:50 <bauzas> like nova.openstack.common.policy
15:27:08 <bauzas> (well, maybe this example is bad, hold on)
15:27:46 <johnthetubaguy> yeah, it would be gnatt.openstack.common.policy, etc
15:27:52 <n0ano> since we're going for a more major change, not just a code forklift, I like that idea, it's just a bigger change with wide ranging implications
15:29:00 <johnthetubaguy> n0ano: I think anything else is a bit confusing
15:29:08 <johnthetubaguy> we will need upgrade/migrate helper scripts
15:29:11 <bauzas> but provided you call gantt.openstack.common.whatever
15:29:12 <johnthetubaguy> but that is true anyways
15:29:31 <bauzas> all the opts will get registered in the same CONF object
15:29:45 <johnthetubaguy> yep, they will
15:30:04 <johnthetubaguy> but when its an issue, they are in separate processes and code trees, just like Cinder vs Nova
15:30:05 <bauzas> so, they will lead to DuplicateOptError
15:30:12 <n0ano> johnthetubaguy, do we  have precedent for helper scripts that change config files, I'd prefer not to be the first one to do this
15:30:48 <johnthetubaguy> I think cinder did something like this
15:30:53 <johnthetubaguy> we would have to check with them
15:31:03 <johnthetubaguy> russellb: might remember what cinder did for config migration
15:32:21 <n0ano> I'll see if I can see what was done, cinder would obviously have needed to do something
15:33:05 <n0ano> it will certainly make splitting things a lot easier if the config moves to a new space, DuplicateOptError has been the bane of my life recently
15:33:57 <Yathi> is there a list of blueprints for gantt split work ?
15:34:33 <n0ano> Yathi, the etherpad is the best place for that, it already has the first list of task & BPs
15:34:58 <Yathi> ok thanks
15:36:10 <n0ano> OK, anything else for today?
15:36:38 <johnthetubaguy> not from me
15:36:44 <johnthetubaguy> https://blueprints.launchpad.net/nova/+spec/scheduler-lib
15:36:47 <n0ano> looks like we talked out for today
15:36:47 <johnthetubaguy> is up there now
15:36:58 <n0ano> tnx everyone, we'll talk again next week
15:37:11 <n0ano> #endmeeting