14:05:20 <sheeprine> #startmeeting cloudkitty
14:05:21 <openstack> Meeting started Mon Nov 23 14:05:20 2015 UTC and is due to finish in 60 minutes.  The chair is sheeprine. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:05:23 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:05:26 <openstack> The meeting name has been set to 'cloudkitty'
14:06:14 <sheeprine> I wanted to talk first about the new contributors.
14:06:52 <sheeprine> That's awesome, we've got multiple people contributing to CloudKitty and submiting bug reports. I hope the numbers will keep growing.
14:07:16 <sheeprine> I think 0.5 will help us gather new people around the project.
14:07:36 <sheeprine> #topic gnocchi support
14:07:47 <sheeprine> Ok first topic, gnocchi.
14:08:02 <sheeprine> We've talked about this during sessions at the summit in Tokyo.
14:08:12 <sheeprine> And people are waiting for us to integrate support for gnocchi.
14:08:57 <sheeprine> We've started working on this with gpocentek. And we plan on providing a first iteration ASAP.
14:10:21 <sheeprine> And then make the integration evolve with new version (new collector format, full gnocchi storage support)
14:11:13 <sheeprine> Some parts need work in CloudKitty others from Gnocchi. That's why we started implementing support new and we'll incrementally add features as they are available
14:11:14 <sheeprine> Our plan is to bundle gnocchi support with 0.5
14:11:14 <sheeprine> which should hit git in a few weeks max.
14:11:43 <sheeprine> We implemented an hybrid storage for gnocchi to leverage capabilities of gnocchi and remove information from the current storage driver.
14:11:58 <sheeprine> It should greatly improve performances and reduce the amount of duplicated data.
14:12:24 <sheeprine> Anyone got suggestions so far?
14:12:41 <gpocentek> sounds good to me so far
14:12:51 <sheeprine> great
14:13:08 <sheeprine> Should we move to next topic or anyone got something to add?
14:14:06 <sheeprine> I guess we can move to next topic.
14:14:17 <sheeprine> #topic pending reviews
14:14:57 <sheeprine> Ok, to make it simple... We've need to validate some code pending review
14:15:16 <sheeprine> There are some patches that in my opinion should hit git and be integrated in the master.
14:16:19 <gpocentek> it's on my todo to review/validate a few patches
14:16:51 <gpocentek> today or more likely tomorrow
14:17:19 <sheeprine> Most are bug fixes about v3 breaking cloudkitty or data duplication in the storage driver
14:17:19 <sheeprine> Without them I doubt people will keep using CK.
14:17:19 <sheeprine> which brings us to next point
14:17:20 <sheeprine> #topic road to version 0.5
14:18:11 <sheeprine> We need to define what should be integrated in next release.
14:18:50 <sheeprine> During Tokyo summit we decided to move to multiple release during a cycle, and that's a good thing because we can easily deprecate code and move forward.
14:19:29 <sheeprine> I think the last points waiting for 0.5 to go live is the integration of the bugfixes and some features.
14:20:24 <sheeprine> Gnocchi support is a plus, and since 0.5 is supposed to be compatible with Liberty we can have users in real prod environments.
14:20:51 <sheeprine> And then we should transition to Mitaka, to implement new features and integrate changes from libs.
14:21:45 <sheeprine> One last thing, we need to validate the Horizon integration. I'm not 100% sure that it's working flawlessly due to js compression.
14:22:15 <sheeprine> We've changed the way we setup cloudkitty in Horizon enabled files. And it looks like only applications can use compressed js.
14:22:28 <sheeprine> We need to be sure that 0.5 bundles a working version of Horizon integration.
14:22:50 <sheeprine> gpocentek: huats: Other thoughts?
14:23:00 <gpocentek> not here
14:24:46 <sheeprine> #topic next version goals
14:25:05 <sheeprine> Ok last topic, we need to define what new version will feature.
14:25:47 <sheeprine> We've got a list of pending features, some needs CloudKitty code rewrite or refactor. And potentially to deprecate old code.
14:26:47 <sheeprine> I'm working on a blueprint to implement rating rule validity times and all middleware code needed so it is rating module agnostic.
14:27:29 <sheeprine> We can implement this feature without the need of major refactor, or backward compatibility breaking changes.
14:28:13 <sheeprine> So we can integrate it in whatever release as soon as it's ready
14:28:29 <sheeprine> Next we've got the collector model.
14:29:09 <sheeprine> Here we are breaking old CloudKitty format, we'll need to implement a compatibility layer to expose data as it was in previous versions.
14:29:50 <sheeprine> The new format will decompose metric from metadata "a la gnocchi", it will reduce duplicated data from the pipeline and speeds calculations up.
14:30:20 <sheeprine> We can imagine implementing new collector format and have a flag in configuration to "flatten" the metadata in a dict like before.
14:30:40 <sheeprine> New storage will benefits from the new format and store the metadata only once.
14:31:10 <sheeprine> Here we don't really have choice, it will be a new version. So you can still use the old one or transition to the new one when you are ready.
14:31:47 <sheeprine> With these changes we'll have a really robust and scalable solution. It will even improve modularity.
14:32:12 <sheeprine> I've got a patch that I'll send in review as WIP in a few days to handle dynamic resource typing from the API or collector.
14:33:32 <sheeprine> Gnocchi needs the same kind of stuff so we might chop some code from it and move it in a lib. I'm looking towards oslo.db, it will make sense.
14:34:20 <sheeprine> So I guess we should have 2 more release, one related to collectors and one to storage.
14:34:37 <sheeprine> With this implemented we should be close to a version 1.0 :)
14:36:35 <sheeprine> Some features I am missing?
14:36:40 <sheeprine> gpocentek: huats: ^
14:37:11 <gpocentek> still nothing here, sounds good, and enough work ,à
14:37:12 <gpocentek> ;)
14:37:51 <sheeprine> Huge patches incoming indeed.
14:38:56 <sheeprine> #topic possible next topics
14:39:05 <sheeprine> Do we have other topic ideas?
14:39:09 <sheeprine> gpocentek: huats: ^
14:39:31 <sheeprine> I'll do a mail summary on the mailing. Some people might be interested.
14:39:36 <gpocentek> I don't think so
14:40:35 <sheeprine> I'll wait a few minutes just in case and if nobody speaks then I'll stop this meeting.
14:43:55 <sheeprine> Ok, thanks guys.
14:43:58 <sheeprine> #endmeeting