15:00:12 #startmeeting ceilometer 15:00:12 Meeting started Thu Jun 4 15:00:12 2015 UTC and is due to finish in 60 minutes. The chair is gordc. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:13 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:16 The meeting name has been set to 'ceilometer' 15:00:27 o/ 15:00:43 welcome back folks 15:00:46 o/ 15:00:51 hi! 15:01:23 o/ 15:01:26 <_nadya_> o/ 15:01:33 o/ 15:01:37 o/ 15:01:38 o/ 15:01:55 o/ 15:01:58 let's get going. 15:02:03 #topic vancouver round up 15:02:23 i hope you all had a good time in vancouver a few weeks ago. 15:02:37 i think we covered quite a bit... or at least what we hoped to 15:02:49 o/ 15:02:53 here's the executive summary of what we covered: http://lists.openstack.org/pipermail/openstack-dev/2015-May/064817.html 15:03:12 o/ 15:03:14 the work items and assignees can be found here: https://etherpad.openstack.org/p/liberty-ceilometer-contributors-meetup 15:03:36 for those who have something assigned to them. if you could, can you place an expected target next to the item 15:03:57 just so we can gauge whether we should assign more resources to certain items 15:04:19 i'll keep it on etherpad for now as apparently google docs isn't available everywhere. 15:04:42 yeah, I've got a question on the agents split. I'm the name on that but fabio has started a spec, so what's the plan there. I'm happy either way 15:04:43 will we track this etherpad on the meetings? 15:04:58 that work with everyone? i'll probably ping you individually if i don't see a date so if you don't want to talk to me. put a date. 15:05:23 gordc: :) 15:05:41 ildikov: sure. let's do that... i think first few weeks we can assume people are just ramping up. but we can reference it. 15:05:57 cdent: ... hmm.. take it back.lol 15:06:12 cdub: i'll check with fabiog his plans... haven't seen him around recentlhy 15:06:16 cdent* 15:06:32 gordc: sure, agreed, I just thought that later it will be informative to see where we need more hands, etc. 15:06:56 ildikov: agreed. good idea 15:07:16 it's an honour system as always. don't make my life difficult by deleting the etherpad 15:07:33 :p 15:07:59 cool cool if there's no questions or anything add we can move on... 15:08:02 we got a lot on agenda 15:08:31 #topic Quick vote and decision on https://review.openstack.org/#/c/184307/ 15:08:36 jd__: all yours 15:09:35 yeah I need people to +2a 15:09:36 gordc: are we voting on the idea or just on the proposed names for the split-out repo? 15:09:39 especially on the name of the thing 15:10:18 eglynn: so i think the idea is pretty much agreed upon from the summit 15:10:24 i think we consider names 15:10:28 gordc: agreed 15:10:50 i didn't yet check out the scope of what is split but i think it should be ok. 15:10:57 * eglynn likes Aodh ... but I'm a sucker for silent consonants, being Irish and all :) 15:11:23 eglynn: i was just going to ask you for fancy names... and you did not disappoint 15:11:30 eglynn: lol :) 15:11:50 eglynn: i assume that ends up being pronounced completely different from how it's spelt 15:12:17 like 'A' ... to rhyme with May 15:12:18 jd__: do you want a vote for name? 15:12:39 yes 15:12:41 or an agreement :) 15:12:57 let's give a vote a try to point us in right direction 15:13:09 * gordc wonders how #vote works 15:13:28 just boolean (y/n) options I think? 15:13:41 * aodh https://en.wikipedia.org/wiki/Aodh_(given_name) (Alarm Over Data Heuristically/Helpfully/Hopefully)* Lanterne https://en.wikipedia.org/wiki/Lanterne_(pasta) 50 * Mezzelune https://en.wikipedia.org/wiki/Mezzelune 51 * Ravioli https://en.wikipedia.org/wiki/Ravioli 52 * Spätzle https://en.wikipedia.org/wiki/Sp%C3%A4tzle 53 * Campanelle https://en.wikipedia.org/wiki/Campanelle (a bell-shaped pasta. Alarms, bells…) 15:13:55 eglynn: oh. i guess that doesn't work. 15:14:11 yeah, you can only +1 0 or -1 on a vote 15:14:25 # vote lets do it and call it aodh 15:14:36 (but without the space) 15:14:41 ok 15:14:55 * eglynn concurs 15:14:59 #vote lets do it and call it aodh 15:15:04 don't work 15:15:21 3 15:15:24 let's just assume we're all leaning towords aodh 15:15:31 gordc ... # startvote ? 15:15:34 #startvote lets do it and call it aodh 15:15:35 Only the meeting chair may start a vote. 15:15:39 there we go 15:15:44 #startvote lets do it and call it aodh 15:15:45 Unable to parse vote topic and options. 15:15:50 lol 15:15:53 fdashfasfdaslfkdasf 15:15:55 this is too difficult. 15:16:05 anyone else have a preference? 15:16:17 else let's just say aodh 15:16:20 http://docs.openstack.org/infra/system-config/irc.html#voting 15:16:24 gordc ... example # startvote Should we vote now? Yes, No, Maybe 15:16:25 apparently you can make choices 15:16:49 #link http://docs.openstack.org/infra/system-config/irc.html 15:17:33 #startvote what's the new alarm name? aodh, lanterne, mezzelune, ravioli, campanelle, spatzle 15:17:34 Begin voting on: what's the new alarm name? Valid vote options are aodh, lanterne, mezzelune, ravioli, campanelle, spatzle. 15:17:35 Vote using '#vote OPTION'. Only your last vote counts. 15:17:58 #vote aodh 15:18:12 #vote Maybe 15:18:13 idegtiarov_: Maybe is not a valid option. Valid options are aodh, lanterne, mezzelune, ravioli, campanelle, spatzle. 15:18:17 lol 15:18:18 #vote aodh 15:18:21 #vote aodh 15:18:24 #vote aodh 15:18:24 #vote aodh 15:18:33 #vote aodh 15:18:34 * jd__ does not care 15:18:43 #vote aodh 15:18:46 #showvote 15:18:46 aodh (7): ildikov, jasonamyers, cdent, eglynn, prad, gordc, idegtiarov_ 15:18:51 ok, i think we just burned through time. at least we know how voting works now. 15:18:54 jd__: that's not a valid option :) 15:18:58 \o/ 15:19:00 #vote aodh 15:19:09 #vote aodh 15:19:13 10 sec 15:19:32 #endvote 15:19:33 Voted on "what's the new alarm name?" Results are 15:19:34 aodh (9): ildikov, rjaiswal, jasonamyers, cdent, eglynn, prad, gordc, ityaptin, idegtiarov_ 15:20:02 cool. we learned extreme democracy. 15:20:20 i guess if we can review the spec we should be good to go. 15:20:27 i'll update the spec 15:20:45 I've already started to work on a patch so I might have this new repo done quickly 15:20:55 jd__: cool stuff. 15:21:11 #topic "subscribing" to the guidelines produced by the api-wg 15:21:18 cdent: go for it 15:21:46 just sort of an update to the question of "do we need to follow the guidelines from the working group" 15:22:23 I brought it up at one of the conference sessions that sometimes (e.g. gnocchi) there's a desire to not follow the guidelines (on use of HTTP 501 response code for example) 15:22:40 there's no stick inovolved in the guidelines, the basic statements are: 15:23:02 if you don't follow the guidelines you're not "being openstack" and/or you won't get the "follows-the-api-guidelines" tag 15:23:12 neither of these things seem particularly relevant 15:23:37 until you get branded for real. 15:23:43 but we may want to keep it in mind as we develop api related stuff: there are good reasons for why the guidelines are ending up as they are 15:23:55 what's the reason for exception in gnocchi? 15:24:05 or is it something best left for off meeting? 15:24:23 the debate on when to use 501 is long and complicated 15:24:30 so probably best not to revisit here 15:24:48 cdent: is the "follows-the-api-guidelines" tag for real? 15:24:51 there is a question underlying all this though and that is: do we care? do we want to do something about it? 15:24:54 cdent: cool cool. "the debate on when to use is long and complicated" 15:25:04 cdent: are there other exceptions too? 15:25:07 eglynn: it is "forthcoming" according to lifeless 15:25:31 wow 15:25:32 cdent: i think we care. it's just easier if things happen as expected with common results. 15:25:42 eglynn: yeah... tags... 15:25:44 I'm not currently aware of other exceptions yet, but a) the guidelines are nowhere near complete nor published b) I haven't had a chance to do a full audit 15:25:45 eglynn: this whole tagging story is under discussion as much as I know, but I think it will be soon 15:25:45 tags certainly are the new hammer 15:26:02 yes, eglynn, that's what I said in response... 15:26:41 eglynn: yeap, but we still need to keep in mind to have some... 15:26:42 cool cool. i guess we should try to follow whatever exists so long as it seems logical 15:26:48 ops tags sound brutal on the ops list 15:26:50 tags can only capture a boolean distinction ... whereas there's likely to a wide spectrum from slavishingly following the api-wg guideline to pointedly ignoring them 15:27:01 Something to keep in mind is that there is a relatively small coterie of people validating the guidelines with any real authority (due to experience) and I'm in that group, so we can shape the guidelines quite a bit 15:27:32 * gordc adds a tag rant topic for open discussion. 15:27:46 but I don't want to get into a situation where we get lost in the bikeshed here, there's enough of that going on in the api-wg itself 15:27:47 cdent: cool, I just wanted to ask 15:28:47 cdent: i think as long as the services follow the guidelines to some degree, we'll all be in somewhat the same direction. 15:28:47 there's an active process in nova for api developments: if no guidelines exist for somethign that is ambiguous, create a guideline first, then make the nova changes 15:28:56 I don't reckon we need that for us, but it is there as an option 15:29:06 what happened to the idea of the api-wg being a repository of knowledge and enlightenment that projects would naturally buy into, as opposed to holding a big stick? 15:29:12 we have the advantage of being small and friendly 15:29:31 by far the biggest nova API change in recent times was microversioning 15:29:34 IMO at least 15:29:54 yet the api-wg had no input to that? 15:29:59 eglynn: no bug change with microversions :) 15:30:02 that's (education) still the goal by most people eglynn, it's just when the question of "what about when someone doesn't want to follow an important rule" where things enter into discussion of enforement 15:30:21 there' s a pending review on microversioning 15:31:03 well, retrospectively, right? 15:31:23 as in ... not guideline first, change after? 15:31:31 anyhoo, we digress 15:31:46 yes, but all the guidelines are effectively retrospective inclusion of rules that already exist because they are simply recapitulating what ought to be good web behavior 15:31:48 tbh, i don't really care about the tags but cdent you can keep us relatively aligned. 15:32:23 thanks gordc that's pretty much what I'm after is a statement of how much we care and it sounds like "we'd like to be good citizens but aren't going to be uptight about it" which seems sane 15:32:51 cdent: seems like a good politically-correct response. 15:33:11 on to the next? 15:33:27 #topic recurring: ceilometerclient release? 15:33:37 #agreed we'll be good citizens with regard to api-wg but not uptight about it 15:33:47 i am tracking nothing for this... anyone else? 15:34:11 cool, this topic was fun. next. 15:34:15 ha 15:34:17 yay! 15:34:23 #topic recurring: Gnocchi status 15:34:38 jd__: sileht: anything? 15:34:56 cdent: "we'd like to be good citizens but aren't going to be uptight about it" makes sense ... with the caveat, "pretty, pretty please, don't use tags for this" 15:34:58 I'm working on making the API and processing more async 15:35:15 other than that, no big news I guess 15:35:33 jd__: yay! async! 15:35:48 that should make things much faster 15:35:50 I've started experimenting with some gabbi-based functional testing things, just to get a better idea of how things work in a real installed system, I think this will likely result in some exploratory tools 15:35:52 no thing special, I wrote a piece of code to plug gnocchi with grafana 15:36:04 sileht: does it work now? 15:36:22 i think so too. the processing was a bit sluggish. 15:36:23 jd__, I have a trouble with keystone and their proxy stuffs 15:36:30 ok :( 15:36:38 jd__, otherrise I have implemented all your query in their system 15:36:42 awesome 15:36:48 (our queries) 15:36:50 wwwwannnnts 15:36:51 sileht: you need contacts in keystone for help? 15:37:01 "I know people" 15:37:23 i don't know people. i was just going to link irc channel. :P 15:37:25 gordc, my issue is in grafana 15:37:31 ah i see. 15:37:59 not sure they have experience with that. 15:38:40 sileht: let me know if you hit a wall. i can try redirecting issue somewhere... i don't know where yet. 15:38:57 I have published my code here: https://github.com/sileht/grafana2-gnocchi-datasource-plugin 15:40:00 sileht: nice. and it'll exist as something outside for gnocchi? 15:40:04 outside of* 15:40:46 sileht: just stop in to #openstack-keystone if you need help 15:41:01 keystone help that is 15:41:08 dstanek, thx 15:41:23 dstanek: that was the link i was all the help i was going to give. 15:41:28 hmm 15:41:36 i typed two different sentences in one 15:41:40 gordc, this is a plugin for grafana so perhaps the grafana-plugins code tree make sense 15:41:50 sileht: makes sense 15:42:04 gordc: that's because you are efficient 15:42:17 dstanek: thanks!lol 15:42:23 cool anything else gnocchi related? 15:42:44 #topic Midcycle Meetup 15:42:59 not sure who added this. 15:43:02 Do we normally have a midcycle? 15:43:07 I did 15:43:11 jasonamyers: ah i see. 15:43:30 jasonamyers: we had one and the next one we cancelled 15:43:33 jasonamyers: so we skipped the midcycle in kilo because we couldn't get enough attendees 15:43:50 ahh okay 15:43:50 jasonamyers: generally we need someone to offer as host 15:44:01 we can talk about it though. 15:44:10 on the beach near my house 15:44:15 i don't know if anyone has concerns about budget 15:44:16 Cisco will gladly host at RTP 15:45:02 jasonamyers: cool. rtp might be ok... might be an issue for devs in russia/china. 15:45:11 thats 4 miles from where i live.. so +1 from me :P 15:45:25 but yea the rest of the team is far far away 15:45:27 so we have cdents house and rtp. 15:45:32 we have other offices, but rtp was the first one they offered 15:46:00 if their is another office that works I can see about that as well 15:46:03 jasonamyers: maybe you can get a more euro based option. and possible time? 15:46:36 for all the contributors we can start asking our managers for pots of gold... 15:46:46 will do 15:46:54 ask for something really big so it makes a plane ticket seem more reasonable 15:47:03 jasonamyers: gordc: the Nova folks had an eventbrite registration or smth like just to see how many attendees they can expect 15:47:16 maybe we could do smth similar, it was announced on the dev list 15:47:29 ildikov: good idea 15:47:45 ildikov: i'll look into that. 15:48:03 I would prefer Ireland, I still have a corporate plane ticket to there... 15:48:21 gordc: cool, tnx 15:48:23 we do have a nice office outside of dublin 15:48:49 jasonamyers: cool, sounds nice to me :) 15:49:04 ... ildikov always gets what she asks for. 15:49:16 Ill get on it 15:49:30 cool stuff... thanks for taking lead on that jasonamyers 15:49:36 * ildikov tries hard at least :) 15:49:43 jasonamyers: anything else you need? 15:50:09 assume no. 15:50:14 #topic Short term solution vote and decision on https://review.openstack.org/#/c/171826/ 15:50:19 rjaiswal: floor is yours 15:50:30 i practiced my voting skills 15:51:23 Im good 15:51:35 gordc: thx, yes, we need a vote on the short term approach on this - file polling or HUP-signal to achieve pipeline reloading 15:51:56 short term is solving the config reload on pipeline level? 15:52:04 rjaiswal: ... that seems too complex for a irc vote.lol 15:52:40 gordc: assuming the options and pros/cons were discussed in the summit 15:52:46 i guess we did agree on some form of file based refresh. 15:53:03 other potential options are using olso to reload config file 15:53:18 I think the file based approach does not prevent us to switch to datastore later 15:53:28 jd__: want to remove your -2? https://review.openstack.org/#/c/171826/ 15:53:53 gordc: I need to re-read it I guess? 15:54:05 jd__: sure. 15:54:18 rjaiswal: did you refresh spec with options? i didn't read it tbh. 15:54:21 so I think we can start there and then if the pipeline rebuild is done with file reload, then we still have ways forward 15:54:49 i don't know if you have quick pitch on what pro/con of each option is? 15:54:58 rjaiswal: or we can just read the spec 15:54:58 gordc: i havn't, i was thinking of another spec for the short term approach 15:55:37 rjaiswal: ah i see... i think that might be a good idea. i don't think we all have a sense of your proposal and what the alternatives are 15:55:52 rjaiswal: i think you can present an option you feel is best and we can go from there. 15:55:56 gordc: sure, so the file reloading is easiest to achieve, but error-prone 15:56:19 since there is no coordination among agents to use a single pipeline 15:57:05 rjaiswal: kk. yeah. i think if we drop that in a short spec we can all take a look and provide feedback. i don't think we'll be able to come to conclusion in 3 minutes 15:57:25 rjaiswal: we can keep this on the meeting for next week if we haven't decide by then? sound good? 15:57:34 gordc: yes, i can work on a spec and pull in feedback from etherpad 15:57:44 rjaiswal: cool stuff. 15:57:52 gordc: yes, sounds god, shud have it ready by then 15:58:03 rjaiswal: awesome 15:58:08 gordc: thanks 15:58:21 #topic open discussion 15:58:29 ildikov, __jd: thanks 15:58:33 1 minute to do whatever 15:58:57 tags throwdown 15:59:11 rjaiswal: np, add to the review please, if you have a new spec 15:59:19 cdent: ... don't ask me about tags... i will rant... and i will get fired.lol 15:59:41 cdent: don't ask me about ... i will rant... and i will get fired. 15:59:48 * cdent would like to see that 15:59:53 ok. 16:00:02 end of time 16:00:06 #endmeeting