15:00:41 <eglynn> #startmeeting ceilometer 15:00:41 <openstack> Meeting started Thu Feb 5 15:00:41 2015 UTC and is due to finish in 60 minutes. The chair is eglynn. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:42 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:45 <openstack> The meeting name has been set to 'ceilometer' 15:01:07 <gordc> o/ 15:01:10 <idegtiarov> o/ 15:01:16 <sileht> o/ 15:01:21 <jasonamyers> o/ 15:01:25 <ityaptin> o/ 15:01:37 <_elena_> o/ 15:01:38 <fabiog> o/ 15:02:01 <eglynn> #topic kilo-2 15:02:02 <cdent> o/ 15:02:13 <eglynn> this is tagged and bagged as of yesterday afternoon 15:02:17 <llu-laptop> o/ 15:02:18 <eglynn> https://launchpad.net/ceilometer/+milestone/kilo-2 15:02:24 <prad> o/ 15:03:11 <gordc> good stuff 15:03:12 <ildikov> o/ 15:03:23 <eglynn> self-disabled-pollster, disk-info-meters, ceilometer-configuration-via-data-store all bumped to kilo-3 as the patches weren't going to land in time 15:03:26 <gentux> o/ 15:03:45 <eglynn> but should be able to land pretty early in k3 I think 15:03:56 <fabiog> eglynn: https://review.openstack.org/#/c/142592/ ready for review 15:03:58 <fabiog> :-) 15:04:14 <eglynn> cool 15:04:45 <ildikov> fabiog: I've already added some comments regarding to what we were discussing yestreday 15:05:03 <fabiog> ildikov: saw that will respond later today 15:05:19 <eglynn> #topic kilo-3 plans 15:05:21 <ildikov> fabiog: cool 15:06:04 <eglynn> so I'd really like to do a full-court press on the outstanding gnocchi/ceilometer integration tasks for kilo-3 15:06:29 <eglynn> ... i.e. make those tasks the top priority for the last kilo milestone 15:06:38 <gordc> we have a list of todos? 15:06:59 <gordc> or unassigned items more specifically. 15:07:00 <eglynn> gordc: so I was discussing this jd last week and I came up with this rough list http://bit.ly/gnocchi-ceilo-k3-tasks 15:07:55 <gordc> cool cool. 15:08:08 <eglynn> I'd ask anyone with available bandwidth this milestone thinks about picking up one or more of those tasks 15:08:13 <eglynn> pretty please :) 15:09:01 <jasonamyers> eglynn are any of those a good place for a n00b to join in? 15:09:05 <cdent> I think I'll have at least some 15:09:31 <eglynn> jasonamyers: yeah, I'm sure we could find something newbie-suitable 15:09:43 <jasonamyers> then I'd love to help 15:09:56 <eglynn> excellent :) 15:10:09 <eglynn> we should probably generate a bunch of blueprints from those rough task descriptions for tracking purposes 15:10:18 <eglynn> though I think the launchpad footprint that jd__ created for gnocchi is for bugs only 15:10:39 <eglynn> i.e. no BP support on https://launchpad.net/gnocchi 15:11:14 <ityaptin> eglynn, I will be happy take some tasks :) 15:11:29 <eglynn> ityaptin: excellent, thanks! 15:11:39 <prad> I can help out with a few as well, lemme know 15:13:04 <eglynn> prad: thank you sir! 15:13:10 <idegtiarov> eglynn: I also will be happy to help with that tasks. 15:13:27 <eglynn> idegtiarov: cool :) 15:13:53 <eglynn> so I guess we'll be tracking these items via https://blueprints.launchpad.net/ceilometer if no BPs on launchpad.net/gnocchi 15:14:26 <eglynn> but for now, if any task catches your eye, please just add your nick to the Assignee column of that googledoc sheet 15:15:34 <eglynn> ok, moving on 15:15:41 <eglynn> #topic gnocchi 15:16:32 <sileht> eglynn, this week I have played with autoscaling use-case and gnocchi, and I have a working setup 15:17:00 <eglynn> sileht: excellent! ... did that require much change or any change to heat? 15:17:00 <cdent> yeah, I was going to mention that a) that's very cool b) we should be actively sharing such experiments more visibly so as to create some energy 15:17:16 <eglynn> (to set the requirement user metadata for the instance server_group) 15:17:26 <sileht> I have ~ 10 patches to finish to fix tiny issue I have found and the heat resources 15:18:01 <eglynn> coolness! 15:18:34 <sileht> eglynn, Heat part was really simple, I have created two resources, most of the code is shared with already existings heat resource 15:19:19 <eglynn> excellent, the less disruption on the heat side, the better for gnocchi adoption 15:20:59 * eglynn starting looking at what's coming up in the next InfluxDB 0.9.0 release 15:21:48 <eglynn> interestingly they've renamed the "shard space" feature (for efficient expiry of older datapoints) 15:22:09 <eglynn> it's now called: "retention policies" 15:22:18 <eglynn> sounds very familiar :) 15:22:47 <gordc> start trademarking the names you come up with. 15:22:57 <cdent> /kick goodes 15:23:02 <cdent> whoops that was supposed to be 15:23:03 <gordc> new job in advertising 15:23:03 <eglynn> LOL :) 15:23:05 <cdent> /kick gordc 15:23:25 <gordc> ... please don't kick the goods... 15:24:18 <eglynn> anything else on gnocchi status? 15:24:59 <eglynn> k, we're burning thu' the agenda today :) 15:25:06 <eglynn> #topic open discussion 15:25:30 <mmester> Writing an abstract on OSS & Ceilometer 15:25:44 <mmester> wondering the interest level 15:26:37 <eglynn> mmester: an abstract? ... in the sense of a topic proposal for the conference track in Vancouver? 15:26:47 <mmester> yeah 15:26:49 <mmester> sorry 15:27:41 <gordc> is "what is oss?" a stupid question? 15:27:47 <eglynn> yeah I think there are several ceilometer-related proposals already in the works from various team members 15:27:50 <eglynn> ... but definitely the more the better 15:28:13 <eglynn> OSS == open source software? 15:28:13 <mmester> Operational Support Serivces 15:28:18 <eglynn> a-ha :) 15:28:26 <mmester> more Cielometer from an ops point of view 15:28:26 <eglynn> cool 15:28:32 <eglynn> excellent 15:28:45 <gordc> mmester: you can promote it on openstack-ceilometer when you've submitted so we can check it out 15:29:18 <eglynn> mmester: yeah, if you want any upfront reviews on your proposal just shout on IRC 15:29:20 <mmester> ok cool. Ill send it up there later today or maybe tomorrow. Want to fine tune it 15:30:04 <eglynn> mmester: as you're probably aware, there's a weird system for selection of sessions for the conference track 15:30:30 <eglynn> mmester: ... there's a public vote, but that can be overturned by the conference track chairs 15:30:43 <cdent> I would love to see/hear an ops view on ceilometer 15:30:56 <eglynn> me too 15:31:48 <gordc> full disclaimer, have a proposal that is somewhat ops related but it's a pretty broad topic anyways 15:31:49 <cdent> I have a general question about how far down the rabbit hole to go wth gabbi tests of ceilometer and gnocchi? 15:32:24 <cdent> If I chase some stuff that is sort of like but not quite bugs I'll be needing to make patches to wsme and pecan. 15:32:24 <eglynn> cdent: ... in terms of the fullness of the API coverage? 15:34:33 <cdent> eglynn: not in covering the api. mostly in terms of dealing with weird inputs 15:34:41 <eglynn> cdent: I would try to avoid getting too bogged down in wsme & pecan if you can 15:34:54 <cdent> for example if you pass a bogus mime type on a put to ceilo api it will cause an untrapped exception 15:35:24 <cdent> so instead of getting a 4xx back, you'll get a 500 (assuming the server traps the exception). In testing you just get the exception 15:35:35 <eglynn> cdent: a-ha, k ... so if those are simple enough patches that woud be relatively quick to land, then I guess have at it 15:37:09 <cdent> the other main thing I've been finding are missing location headers, but we've talked about that before 15:39:08 <cdent> given unlimited time I'd boot pecan right out the door 15:39:48 <eglynn> cdent: smoking out those kinds of inconsistencies in the API (e.g. lack of expected location header) sounds like an excellent secondary benefit to this program of gabbi testing 15:40:24 <cdent> I'm intentionlly making the tests in a sort of black box kind of way: I make requests and see what happens 15:40:57 <cdent> and I reckon its the primary benefit as most of the actual data stuff is covered by the existing api tests 15:41:15 <eglynn> cool 15:41:21 <cdent> secondary seems to be that its an excellent learning tool 15:41:31 <eglynn> so that kind of thing would seem to make sense to fix ... as it's under our direct control, as long it can be done in a backward compatible way 15:41:32 <cdent> I recommend it to anyone who wants to understand the api better 15:41:40 <cdent> where it => writing gabbi tests for some api they don't know 15:41:48 <eglynn> (e.g. without changing the status codes etc.) 15:41:53 * cdent nods 15:42:04 * eglynn digs out jay's old API change guidelines ... 15:42:17 <gordc> cdent: how terrible is the api currently? 15:42:20 <eglynn> https://wiki.openstack.org/wiki/APIChangeGuidelines 15:42:39 <eglynn> specifically https://wiki.openstack.org/wiki/APIChangeGuidelines#Adding_New_Header_OK.3B_Changing_Response_Code_Not_So_Much 15:43:07 <cdent> gordc: are you asking my opinion on its behavior and practical aspects or on its style and cachet? 15:43:45 <gordc> the former? 15:44:56 <cdent> I haven't touched stats or events yet, but meters and the half of alarms I've covered do the job they describe without being too terribly wrong. 15:45:26 <gordc> probably too overloaded a question... but i guess it's good to know you're getting roughly what you expect. 15:45:52 <cdent> oh I'm not getting what I _expect_ but I am getting what's _advertised_. 15:46:05 <cdent> I trust you get the difference? 15:46:50 <gordc> what are we advertising?... "you'll get something"? 15:47:16 <cdent> there is an API and if you use it exactly like this it will do some stuff 15:47:29 <gordc> ah gotcha. 15:47:36 <cdent> the issues are around: if you throw random crap at it, the behavior goes weird 15:47:42 <cdent> and in my expectations that's not good enough 15:48:16 <gordc> not as bad as i was thinking... always expect the worse = always pleasantly surprised. 15:48:48 <cdent> well presumably it has to at least work somewhat or else it wouldn't be usable and people do use it? 15:49:45 <jasonamyers> I have a question about how to get started contributing to ceilometer. I know how to do it for openstack, but I'm trying to find the right task to pick up here to get my feet wet. 15:50:31 <jasonamyers> and perhaps this is better directed in the normal chat channel 15:50:59 <gordc> jasonamyers: any particular interest? db collection? metrics? stats? 15:51:10 <jasonamyers> I like the db and metrics components 15:51:37 <eglynn> jasonamyers: would helping cdent out to extend the API coverage of the gabbi tests be something interesting to you? 15:51:50 <jasonamyers> I'd give it a whirl 15:52:08 <eglynn> cool 15:52:08 <cdent> yeah, feel free to come looking for me and I can point you in the right direction 15:52:33 <eglynn> excellent :) 15:52:34 <jasonamyers> sounds good 15:52:44 <cdent> I'm sort of doing it in three phases: phase one is get the basic framework in, phase two is get basic tests for each type of resource, phase three is more complex tests that check query strings and the like 15:52:48 <cdent> phase two is just getting started 15:53:01 <jasonamyers> excellent I'll reach out outside of the meeting 15:53:08 <cdent> cool 15:54:15 <eglynn> ok, sounds like we're out of topics to discus 15:54:41 <gordc> back to openstack-ceilometer 15:54:56 <eglynn> cool, thanks for your time folks! 15:55:01 <cdent> o/ 15:55:05 <eglynn> #endmeeting ceilometer