21:00:05 #startmeeting Ceilometer 21:00:05 Meeting started Wed Dec 19 21:00:05 2012 UTC. The chair is nijaba. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:05 #meetingtopic Ceilometer 21:00:05 #chair nijaba 21:00:05 #link http://wiki.openstack.org/Meetings/MeteringAgenda 21:00:05 ATTENTION: please keep discussion focussed on topic until we reach the open discussion topic 21:00:06 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:00:08 The meeting name has been set to 'ceilometer' 21:00:10 Current chairs: nijaba 21:00:16 Hello everyone! Show of hands, who is around for the ceilometer meeting? 21:00:16 o/ 21:00:18 hi 21:00:19 o/ 21:00:29 hello world 21:00:32 o/ 21:00:36 o/ 21:00:49 o/ 21:00:56 nice! 21:01:01 let's start 21:01:04 #topic actions from previous meeting 21:01:04 #topic asalkeld prepare a blueprint for v2 api "tightening" changes 21:01:13 http://wiki.openstack.org/Ceilometer/blueprints/APIv2 21:01:24 me too 21:01:39 * nijaba looks 21:01:48 I haven't made a blueprint as there are so many about 21:02:01 wsme v2 one 21:02:05 but I can 21:02:32 also dependant on the pass array args 21:02:33 asalkeld: I think we can link it to an existing bp 21:02:44 asalkeld: I can take the action if you want 21:02:54 looks great otherwise 21:02:56 ok 21:02:57 asalkeld: is there a point to having /resources since that seems to be just /meter with ?resources= ? 21:03:12 not sure 21:03:17 ok 21:03:21 https://groups.google.com/forum/?fromgroups=#!topic/python-wsme/gUJH9rLo5RI 21:03:29 I guess we could have a full review about this next week 21:03:50 once doug is back and everyone has time to review? 21:03:59 sure 21:04:06 fine with me 21:04:12 so I am trying to implement, but running into wsme sort commings 21:04:15 #action nijaba to schedule v2 review for next meeting 21:04:34 #action nijaba to link proposal to appropriate bp 21:04:56 #topic dhellmann experiment with passing arrays of parameters to wsme for queries 21:04:56 #info dhellmann is still traveling, so I am re-actioning this for next week 21:04:56 #action dhellmann experiment with passing arrays of parameters to wsme for queries 21:05:18 #topic eglynn reach out to healthnmon project 21:05:18 #info This was obsoleted later in the meeting last week 21:05:24 didn't get a chance to do anything with healthnmon this week 21:05:32 need to bump that to the next meeting 21:05:33 #topic eglynn, yjiang5, llu-laptop to work on a wiki page for analysis of healthmon duplication, etc.. 21:05:33 #info this was done, there is a full topic about this in a bit 21:05:33 #link http://wiki.openstack.org/Ceilometer/CeilometerAndHealthnmon 21:05:55 so I propose we discuss lianho write up later 21:05:55 that was nice 21:05:59 yep, some interesting observations on that wiki page 21:06:19 #topic jd ask eglynn if https://blueprints.launchpad.net/ceilometer/+spec/nova-independent-virt is implemented 21:06:25 it is 21:06:33 for libvirt 21:06:48 but if anyone is interested in xenapi support 21:06:48 should I change the status then? 21:06:52 that would be good too 21:06:59 yep, status complete 21:07:08 (xenapi is an added extra...) 21:07:24 #info bp marked as implemented 21:07:36 thanks eglynn 21:07:54 #topic russellb raises the question about Oslo notifications versionning on the ml 21:07:55 np! 21:08:09 ooh 21:08:12 I think this was actually proposed as a patch 21:08:15 i did not do that. i wrote a patch instead 21:08:17 yes 21:08:19 which I have reviewed 21:08:25 * nijaba looks for link 21:08:53 #link https://review.openstack.org/#/c/18057/ 21:08:57 #link https://review.openstack.org/18057 21:09:02 duh! 21:09:14 #undo 21:09:20 eheh 21:09:42 other reviews are welcome, I guess 21:09:53 so to clarify, this isn't versioning of notificaitons themselves 21:10:02 just the wrapper part that will eventually allow us to add things like signatures 21:10:06 encryption, other metadata 21:10:19 right 21:10:26 a usefull step! 21:10:30 :) 21:10:36 thanks for reviewing! 21:10:40 np 21:10:55 what's the upgrade path to the new format? 21:11:22 i.e. is it a big bang, or can the old envelope-less format still be tolerated by consumers? 21:12:23 russellb: ? 21:12:41 just wondering about ceilo remaining compatible-ish with pre-Folsom nova 21:12:45 as it stands, the envelope is optional and not used by default 21:12:53 k 21:12:55 and the code supports receiving both formats 21:13:00 cool 21:13:23 checks to see if it looks like an envelope, and if it doesn't, assumes it didn't have one 21:13:32 nice 21:13:37 great, sounds reasonable 21:13:48 the details of that are in the patch in front of the one linked 21:14:39 any other comments, or shall we move on? 21:14:45 deserialize_msg() in https://review.openstack.org/#/c/17554/7/openstack/common/rpc/common.py 21:14:48 none from me 21:14:53 looks good 21:15:10 #topic need to find a volunteer to organize bug squashing day on Jan 4th 21:15:10 again, I won't be around that week and we need volunteers to organize it. Do we have any? 21:15:30 o/ 21:16:01 we'll also need to divy out some triaging work 21:16:14 start tagging bugs as low-hanging fruit etc. 21:16:26 jd__: can you take care of the organization? I guess we should start "advertising it" soon... 21:16:40 eglynn: good point 21:16:52 nijaba: sure, do you have already some idea of what we could do? 21:16:59 I can announce it this week on the ml at least 21:17:00 (if we're gonna use the bug squashing day to smoke out some new contributors, good to have a nice set of getting starting style bugs ...) 21:17:00 #info everyin 21:17:15 #info everyone should start triaging and tagging bugs 21:17:24 eglynn: I can probably do some of that the day before, but help welcome :) 21:17:30 jd__: cool 21:18:05 jd__: I think we need to have a process to help people that are willing to do some work 21:18:10 this could be: 21:18:14 - bug fixing 21:18:17 - testing 21:18:24 - docco 21:18:26 - bug reproduction 21:19:02 ok, maybe writing this on a wiki page could help? 21:19:18 and we need to organize presence in the IRC chan to help directing people 21:19:40 so find volunteers to greet people throughout the day 21:19:57 which should be easy, since we have people all around the world 21:20:04 for testing, a detailed step-by-step wiki like the Fedora openstack test day, would be good to guide newbies ... 21:20:16 eglynn: pointer? 21:20:18 #link https://fedoraproject.org/wiki/Test_Day:2012-09-18_OpenStack 21:20:22 great 21:20:42 I'll write as much as I could for next week so we can discuss it on next meeting 21:20:48 s/could/can/ 21:20:48 cool 21:20:51 and need resolve the webob/wsme dependency issue before that day, so that trunk works well in devstack. (We should have fixed it at that time) 21:21:04 jd__: action? 21:21:17 yjiang5: agreed, we can discuss this at the end of this meeting btw 21:21:18 yjiang5: good point 21:21:20 nijaba: yup 21:21:48 #action jd__ write a wiki page on how to participate to the bug squashing day 21:21:57 * nijaba will gladly help 21:21:57 #action jd__ announce the bug squashing day on the ml 21:22:06 yjiang5: yep, good point ... the initial experience has to be smooth-ish, otherwise we'll loose the newbies straight away 21:22:32 jd__: we may need to advertise wider than the dev ml 21:22:36 may 21:22:51 nijaba: do you have something on mind? 21:22:59 lkml? :p 21:23:03 maybe on the openstack ml and community on? 21:23:05 jd__: that wiki page can be a good newbie document in the future. 21:23:10 s/on/one 21:23:10 lkml? ..... 21:23:21 yjiang5: joke: linux kernel ml 21:23:38 maybe some tweets and a blog too? 21:23:50 nijaba: yeah I was thinking about the general openstack list too, there's a community list or something too? I need to check that 21:24:08 jd__: yes, there is one. I'll relay if you want 21:24:09 nijaba: yeah I'll tweet and blog like a crazy cow too! 21:24:16 \o/ 21:24:30 #action jd__ megatweet about the bug squashing day every hour 21:24:43 ok, should we move on? 21:24:43 #undo undo undo 21:24:47 hehe 21:24:49 :-) 21:25:05 That's it for last week action 21:25:15 #topic Meetings during the holiday season 21:25:15 So we have a couple meetings coming up that are falling in the holliday season, so I am wondering if we should still hold them? 21:25:15 I am proposing to do a quick check for presence in the form of a vote 21:25:43 do you guys agree? 21:25:52 should we vote on agreeing? 21:25:56 * jd__ agrees :-) 21:25:57 lol 21:26:18 #startvote present on thu dec 27 15UTC? yes, no 21:26:19 Begin voting on: present on thu dec 27 15UTC? Valid vote options are yes, no. 21:26:20 Vote using '#vote OPTION'. Only your last vote counts. 21:26:24 yep agreed not for next week, we're in shutdown mode ;) 21:26:25 #vote yes 21:26:29 #yes 21:26:30 #vote yes 21:26:32 #vote yes 21:26:34 #vote yes 21:26:41 #vote yes 21:27:03 looks like most of us are avail on the 27th! great 21:27:14 waiting a bit for more votes 21:27:15 #vote no 21:27:26 * eglynn got confused as to what we were voting on 21:27:29 eglynn just took vacation actually 21:27:52 I guess we are done for this one 21:28:00 #endvote 21:28:01 Voted on "present on thu dec 27 15UTC?" Results are 21:28:02 yes (4): yjiang5, nijaba, asalkeld, jd__ 21:28:03 no (1): eglynn 21:28:03 nijaba: I think Doug will be there since he just got vacations 21:28:18 #agreed meeting next week maintained 21:28:32 fair enough, I'll try to make it 21:28:37 #startvote present on wed Jan 2 21UTC? yes, no 21:28:38 Begin voting on: present on wed Jan 2 21UTC? Valid vote options are yes, no. 21:28:39 Vote using '#vote OPTION'. Only your last vote counts. 21:28:47 #vote no 21:28:48 #vote yes 21:28:49 #vote no 21:28:54 #vote yes 21:29:16 #vote no 21:29:45 this one lokks like a looser 21:29:53 any more votes? 21:30:10 #endvote 21:30:11 Voted on "present on wed Jan 2 21UTC?" Results are 21:30:12 yes (2): jd__, eglynn 21:30:13 no (3): yjiang5, nijaba, asalkeld 21:30:30 it's a date then! 21:30:40 #agreed no meeting on jan 2, unless jd and eglynn want to have fun together 21:30:55 yeah, dhellmann might be around too 21:31:11 we'll do a counter-party without you guys, that's it :) 21:31:18 good point 21:31:29 jd__: but that's the day before the bug squash day, possibly someone from community want to join? 21:31:34 so we'll check again next week 21:31:58 #action nijaba to organize new vote next week 21:32:00 yjiang5: yeah, we'll see, we can just start the meeting, chat a bit and end it if it's calm and quiet 21:32:23 jd__: k 21:32:24 #topic Review synaps blueprints 21:32:24 eglynn did a great job at braking this into multiple bluprints and dependencies 21:32:24 #link https://blueprints.launchpad.net/ceilometer/+spec/synaps-integration 21:32:35 This brings a couple questions from me (others may have more): 21:32:35 1/ this seems a lot of work. Do you think we can target this for grizzly with confidence? 21:32:35 2/ it seems that the symaps team is quite busy. Will you be covering all of this by yourself? 21:32:46 yea a lot of stuff there ... 21:33:03 and yep, I'm not counting on a great deal of input from their side 21:33:24 eglynn: wow 21:33:32 so I've started working on it and will get as much done as I can 21:33:50 when is the g3 date? 21:34:03 (heaps of distractions this time of year) 21:34:06 so, how should we target this? Mark them as approved but with no milestone until you complete them? 21:34:18 eglynn: Feb 21st, I think 21:34:33 verified 21:34:34 right be aiming most of it for g3 so I think 21:34:50 g2 is way too soon given the holidays etc. 21:35:10 and changes may be breaking compat with folsom too? 21:35:38 well, it'll depend on post-Folsom ceilo features 21:35:56 e.g. multi-publish, nova-virt-inspector etc. 21:36:07 so, do everyone agree that I approve those bp? 21:36:21 sure 21:36:40 eglynn: we can priotize these bps 21:36:48 yjiang5: cool 21:36:55 eglynn: some are fundamental, while some are add-ons 21:36:59 nijaba, there are a lot of overlapping bps atm 21:37:03 #action nijaba to approve synaps's bp 21:37:12 yjiang5: agreed 21:37:20 asalkeld: I know 21:37:39 asalkeld: should we have a bp cleanup session next week? 21:37:48 could do 21:37:49 agree 21:38:15 you won be around though. maybe we can start together on thu (my morning) 21:38:32 ok 21:39:01 #action nijaba to organixe next thu as bp cleanup day 21:39:35 thu means thursday, right? 21:39:36 let's move on? 21:39:44 yjiang5: yes 21:39:50 nijaba: thanks 21:39:57 #topic Discuss heathmon/ceilometer duplication and differences 21:39:57 lianhao did a great job at analysing the differences and duplication: 21:39:57 #link http://wiki.openstack.org/Ceilometer/CeilometerAndHealthnmon 21:40:09 From what I read, it seems that the hard points are 21:40:09 1/ very different approach on the data model 21:40:09 2/ healthmon does not use standard openstack events 21:40:09 so it seems that at best we could recommend to integrate healthmon with ceilometer through the use of the multi-publisher blueprint 21:40:09 what do you think? 21:40:30 nijaba: lianhao is not on the meeting, can we do it in next meeting? 21:40:44 was their host monitoring something we could consider pulling in? 21:40:52 yjiang5: we can still gather some feedback and conclude next week 21:40:58 nijaba: sure 21:41:07 asalkeld, won't be around next week 21:41:28 maybe - I'll see 21:41:59 #action nijaba to reschdule topic for next week when lianhao is around 21:42:08 any comments for now? 21:42:21 agree with eglyn on host monitoring 21:42:30 well I'd like to see the devs on irc 21:42:43 and actively supporting this 21:42:51 +1 21:42:59 not another case of us pulling code in 21:43:09 yep agreed 21:43:16 +1 21:43:36 we could be guided though by their approach to host monitoring 21:43:36 asalkeld: shall you start a thread on the ml, asking what they think? 21:43:40 eglynn: but host monitor is a feature we need support in future, either from healthnmon, or restart from scratch 21:44:08 yjiang5: yep, just thinking we could borrow ideas that they've battle-tested 21:44:18 nijaba, maybe - or just get them to the meetings 21:44:23 yjiang5: approaches they've proven to work at scale etc. 21:44:24 there is already some effort on https://blueprints.launchpad.net/ceilometer/+spec/monitoring-physical-devices 21:44:47 asalkeld: that could be the topic of the mail :) 21:44:57 eglynn: agreed 21:45:19 sure nijaba 21:46:53 ok, I guess we can move on then... 21:47:02 #topic Open discussion 21:47:30 * dragondm waves 21:48:20 I have nothing 21:48:30 one thing I discussed with asalkeld on IRC, current get_meters need looking for the whole table , possibly we need enhance that? 21:48:49 you mean for sql? 21:48:55 asalkeld: yes. 21:48:55 I'd like to send a patch to move away the v2 api as a second citizen and get back the v1 API as the main one for now, and remove dep on pecan/wsme (unless for tests), wdyt? 21:49:08 yes totally agree yjiang5 21:49:16 jd__: agree 21:49:47 jd__, agree until we get to one api server 21:49:47 jd__: second class citizen = disabled by default? 21:49:50 +1 (for now and until g2 is past) 21:50:13 asalkeld: I will talk with you on implementation in IRC 21:50:26 mv ceilometer-api ceilometer-api-v2 21:50:29 eglynn: that means rename bin/ceilometer-api => bin/ceilometer-api-v2 and rename bin/ceilometer-api-v1 => bin/ceilometer-api 21:50:31 mv ceilometer-api-v1 ceilometer-api 21:50:40 yip 21:50:41 jd__: got it, cool 21:50:41 * jd__ high fives asalkeld 21:50:58 I'll do that tomorrow then 21:51:06 action? 21:51:13 jd__: will this bring jenkins back? 21:51:22 #action jd__ rename api to api-v2 and api-v1 to api 21:51:35 yjiang5: Jenkins is already back, I sent a patch earlier today 21:51:47 I need to do a bunch of recheck on everything, but didn't have time to do this today 21:51:48 and I fast track approved it 21:51:59 on my evil plan for tomorrow, before the world ends 21:52:06 jd__: thanks. 21:52:14 haha 21:52:17 jd__: hehe 21:52:33 we should meter that... 21:52:50 hehe 21:53:33 anything else? 21:53:36 jd__: BTW, please have a check on the multiple-publish patch when you are available, it has been changed 21:53:36 Folks here @ rackspace planning to contribute more to Ceilometer. We would like to use it for some of our usage. 21:53:40 anyway pecan/wsme looks nice and the upstream author is helpful 21:53:50 yjiang5: yeah I see, I'll look tomorrow too! :) 21:53:59 yjiang5: me too 21:54:07 jd__: eglynn: thanks 21:54:08 dragondm: great! 21:54:12 dragondm: yep. I have been discussing this with sandy and dan 21:54:14 dragondm: do you have an idea of what you want to tackle first/ 21:54:39 We have a list of things we'll need to run it at our scale in our infrastructure 21:54:41 dragondm: but I would like them to propose some concrete plan on the changes they think of 21:54:50 so we can discuss it here first 21:55:18 Yes, We will break out our stuff into some blueprints. 21:55:23 coolio 21:55:39 nijaba: definitely agree. sandywalsh is not available for this meeting and holidays mean we'll probably need to wait until first of year. 21:55:45 Plus, we would like to bring in some of the features of our stacktach too. 21:55:49 er tool. 21:55:59 we're certainly interested in working with the ceilometer group though. 21:56:14 all goodness to me. but needs to be planned carefully against the dealines of the project 21:56:31 Yah, our stuff is going to be longer range. 21:57:28 We're not going to drop it on you for g2 or such :-> 21:57:40 dragondm: pfeww ;) 21:58:14 We are in the planning stage now, testing some ideas, etc. 21:58:33 with synaps, healthmon and multi-publisher, we already have qui te a bit for g3 21:59:54 ok, we are just at the top of the hour 22:00:01 any last comments? 22:00:11 clear 22:00:15 nothing from me ... 22:00:18 no 22:00:22 happy holidays! 22:00:30 thanks a lot everyone, another great meeting 22:00:31 yep, ditto! 22:00:38 thanks for having us. we'll meet back up after the beginning of the year to discuss the rackspace stuff. 22:00:46 Yup. 22:00:47 #action enjoy your hollidays 22:01:03 danspraggins: dragondm: happy to have you join 22:01:16 #endmeeting