14:10:46 <sheeprine> #startmeeting cloudkitty
14:10:47 <openstack> Meeting started Mon May  9 14:10:46 2016 UTC and is due to finish in 60 minutes.  The chair is sheeprine. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:10:49 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:10:51 <openstack> The meeting name has been set to 'cloudkitty'
14:11:15 <sheeprine> We've got a pretty small agenda for this meeting.
14:11:35 <sheeprine> Summit debriefing
14:11:35 <sheeprine> Gnocchi support stage2
14:11:35 <sheeprine> Road to version 0.6
14:12:08 <sheeprine> #topic Summit debriefing
14:12:24 <sheeprine> About what happened during Austin summit.
14:13:28 <sheeprine> We've got more people interested in CloudKitty, which is great.
14:13:45 <sheeprine> I got in touch with people from rackspace that started working on the project.
14:14:14 <sheeprine> They'll work on ansible roles and try to modify some parts to make it work for their deployments.
14:14:35 <sheeprine> From what I've heard they should have 3 - 4 people working on it for 2 to 4 weeks.
14:14:54 <huats> I have been discussing with some of them during the weekend
14:15:04 <huats> trying to help them out a bit
14:15:31 <sheeprine> Speaking of that, we need to focus more on the doc. Most people don't know how to install or use CloudKitty.
14:15:45 <huats> sheeprine: indeed
14:15:54 <huats> I have been discussing that with one of them
14:16:24 <huats> Actually I was waiting for a patch to land on the trunk to work on that from a full view (installation and configuration)
14:17:00 <huats> and this patch should arrive quite soon from what we discussed last wednesday sheeprine  :)
14:17:02 <sheeprine> Great, I'll try to ping other cores to speed up review process.
14:17:20 <huats> I will make a review of your doc tomorrow
14:17:23 <huats> sheeprine: ^
14:17:27 <huats> for hashmap
14:17:33 <sheeprine> Anything to add about the summit?
14:18:00 <sergio_nubeliu> I agree that we need to improve the doc, and add some doc about gnocchi and cloudkitty
14:18:19 <huats> sergio_nubeliu: actually this is something that I will adress
14:18:33 <sergio_nubeliu> huats: great
14:18:41 <sheeprine> Yes especialy the current limitations around archive policy and the way cloudkitty is fetching data.
14:18:42 <huats> I have discussed that with err earlier today (a guy from the #cloudkitty chan)
14:19:40 <huats> that it was nice to meet Maximiliano :)
14:19:56 <huats> (that is the thing I wanted to add on the Summit)
14:20:32 <sheeprine> And see what you've done with CloudKitty ;)
14:20:33 <sergio_nubeliu> :)
14:21:13 <sheeprine> Let's focus on gnocchi support now, we've got some work about that.
14:21:20 <sheeprine> #topic gnocchi support
14:22:00 <sheeprine> So, what we all agreed on during the summit and session is to improve gnocchi support and to remove the need for an intermediary database.
14:22:33 <sheeprine> We've got 2 ways of doing it, 1 do a 1:1 mapping with gnocchi resources.
14:22:45 <sheeprine> 2 support dynamic resources thanks to the new API.
14:23:01 <sheeprine> I think 1 is pretty easy to attain in a small timeframe
14:23:08 <sheeprine> but 2 might need more work.
14:23:43 <sheeprine> 1 has some limitations, we can't support data collected outside of gnocchi
14:24:05 <sergio_nubeliu> sheeprine: do you agree to use same resources and create a new metric for each metric that have rating rules?
14:24:26 <sheeprine> sergio_nubeliu: Yep, it's the easiest way to do it.
14:24:36 <sheeprine> Just do a metric.cost
14:25:01 <sergio_nubeliu> sheeprine: that's right. understood, the limitations
14:25:22 <huats> We discussed that it would be best to address 1 dring that cycle and then to work on that during the O cycle for the 2
14:25:43 <sergio_nubeliu> sheeprine: what kind of data are you thinking about outside of gnocchi?
14:25:46 <huats> since it would give time to work on things asides that are needed
14:26:20 <sheeprine> sergio_nubeliu: No clear idea, but we're supposed to handle multiple collectors, which means data that is not coming from gnocchi.
14:26:48 <sheeprine> At the moment the only other collector is ceilometer... which is pointless if we are using gnocchi
14:27:07 <sheeprine> But some people might need it. Since at the moment it's not the case we'll first focus on point #1
14:27:15 <huats> sheeprine: isn't it possible to inject data inside gnocchi from the collectors so that there is just 1 data source ?
14:27:59 <sheeprine> Same problem as storing, you are just moving the problem in the collector.
14:28:06 <sergio_nubeliu> huats: i think yes, its totally possible using the gnocchi api
14:28:42 <huats> ok
14:29:13 <huats> I was just mentionning that if someone NEEDS that it is a possibility ...
14:29:54 <sergio_nubeliu> huats: sure...
14:29:55 <sergio_nubeliu> so, are we agreed to develop this 1:1 option at first?
14:29:55 <sheeprine> Might, but I really doubt we'll ever do something like that
14:30:02 <sheeprine> Yes
14:30:27 <huats> sounds good for me too
14:30:31 <sheeprine> I'm fixing a bug and doing some code cleanup in the current collector.
14:30:53 <sergio_nubeliu> sheeprine: are you ok if i work on this?
14:31:03 <sheeprine> Yeah sure.
14:31:18 <sheeprine> I'll finish the collector part, but you can start the work on the storage.
14:31:47 <sheeprine> This way we can have good gnocchi support faster.
14:32:00 <sergio_nubeliu> great! i think it shouldn't take more than two weeks to have a first version?
14:32:04 <sheeprine> I think that we should create a new driver.
14:32:20 <sergio_nubeliu> sheeprine: agree
14:32:24 <sheeprine> You can duplicate the gnocchihybrid but name it gnocchi
14:32:45 <sheeprine> As it's pretty different from the way it's working.
14:33:09 <sheeprine> And then we implement #2 as new features.
14:33:52 <sheeprine> If you're working on storage I'll try to focus on new collector and internal objects.
14:34:27 <sheeprine> Does anyone as more to add about gnocchi?
14:34:59 <huats> not from me
14:35:01 <sergio_nubeliu> ok, i'll be working on 1:1 storage for the next two weeks
14:35:07 <huats> great to hear sheeprine
14:35:09 <huats> sergio_nubeliu:
14:35:13 <sheeprine> good.
14:35:14 <huats> ^
14:35:15 <huats> :)
14:35:23 <sheeprine> sergio_nubeliu: try to push you patch in review as work in progress
14:35:49 <sheeprine> This way I can have a look at the architecture and tell you if something should be modified earlier.
14:35:53 <sergio_nubeliu> sheeprine: i'll do it
14:35:56 <sheeprine> great.
14:37:19 <sheeprine> Let's move on.
14:37:29 <sheeprine> #topic road to version 0.6
14:37:59 <sheeprine> Nothing really new here, we'll try to keep focus on what we planned for CloudKitty till now.
14:38:10 <sheeprine> The full roadmap is in the previous meeting logs.
14:38:27 <sheeprine> We worked on what we'll do (objectif libre) during the next spints.
14:38:50 <sheeprine> We'll try to get this published ASAP so everyone can have a clear picture of what is soon to come on what's on the map.
14:39:17 <sheeprine> sergio_nubeliu: Any specific point you want to include in version 0.6?
14:41:16 <sergio_nubeliu> sheeprine: for this version 0.6, having the native gnocchi it fine by me
14:41:40 <sergio_nubeliu> sheeprine: when do you plan to release?
14:42:22 <sheeprine> for 0.6? improved gnocchi support, new collector resources modeling.
14:42:31 <sheeprine> Not the awful list of dicts
14:42:40 <sheeprine> You'll have an object with all the parameters
14:42:54 <sheeprine> But it will still behaves as a dict some legacy code still works.
14:43:08 <sergio_nubeliu> sheeprine: i think, the native storage will be a major performance improvement
14:43:08 <sheeprine> I guess it's pretty important for you.
14:43:23 <sheeprine> I'm pretty sure yes.
14:43:49 <sergio_nubeliu> sheeprine: yes, you are right
14:43:52 <sheeprine> At some point of time it might be usefull to plug the fake collector and a dummy storage to profile rating code of cloudkitty.
14:43:53 <sergio_nubeliu> :)
14:43:58 <sheeprine> But not for this release ;)
14:44:36 <sheeprine> version 0.7 will focus on storage
14:44:37 <sergio_nubeliu> sheeprine: do you have an estimated date for the release?
14:44:52 <sheeprine> 0.6? I hope less than a month
14:45:11 <sergio_nubeliu> sheeprine: thats great!
14:45:14 <sheeprine> So if you say you are done with the storage in 2 weeks it's perfect
14:45:23 <huats> it depends on you too sergio_nubeliu  ;)
14:45:23 <sergio_nubeliu> I hope so
14:45:35 <sheeprine> sergio_nubeliu: No presure ;)
14:45:57 <sergio_nubeliu> hahahaha, no presures :)
14:47:44 <sheeprine> Are we ok about what's v0.6?
14:47:53 <huats> I think we are sheeprine  !
14:48:14 <sheeprine> #topic last minutes
14:48:16 <sergio_nubeliu> yes, i think so!
14:48:30 <sheeprine> We have a few minutes left, anything to add?
14:48:44 <huats> Not for me !
14:49:32 <sergio_nubeliu> no, i'll contact you on irc while i develop the storage to discuss if needed
14:50:03 <sheeprine> Ok great.
14:50:07 <sergio_nubeliu> thank you guys!
14:50:13 <sheeprine> We're done then, thanks everyone.
14:50:15 <sergio_nubeliu> have a nice week!
14:50:23 <sheeprine> #endmeeting