15:00:51 #startmeeting ceilometer 15:00:51 Meeting started Thu Mar 6 15:00:51 2014 UTC and is due to finish in 60 minutes. The chair is jd__. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:51 one last thing 15:00:51 great meeting folks 15:00:52 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:55 o/ 15:00:56 The meeting name has been set to 'ceilometer' 15:01:00 o/ 15:01:02 thanks all o/ 15:01:11 o/ 15:01:15 o/ 15:01:20 o/ 15:01:29 Out of curiosity is anyone intereseted in a mini-summit? I would be willing to try and set something up 15:01:39 before Atlanta of course 15:02:20 #link https://wiki.openstack.org/wiki/Meetings/Ceilometer 15:02:26 I guess I didn't beat the buzzer lol 15:02:31 hello 15:02:34 #topic Milestone status icehouse-3 15:02:47 lets switch to opensatck-neutron to discuss further 15:02:49 so the branch has been cut and i3 is going to be released anytime now 15:03:00 \o/ :) 15:03:08 thanks everyone for helping 15:03:18 we're now in feature freeze, so please don't approve any patch that is not a bug fix 15:03:34 woo! 15:03:50 we've been granted a FFE for the VMware part that is not completely merged but should be soon 15:04:03 https://blueprints.launchpad.net/ceilometer/+spec/vmware-vcenter-server 15:04:28 any question on that topic? 15:05:28 #topic Tempest integration 15:05:35 _nadya_: any news? 15:05:39 <_nadya_> hi all! 15:05:48 I have got merged https://review.openstack.org/#/c/78537/ this morning 15:06:03 <_nadya_> we are fighting with notifications :) 15:06:14 I wait zuul to see if it fixes https://review.openstack.org/#/c/64136/ 15:06:27 _nadya_: did you get debug logging enabled? 15:06:41 _nadya_, did you get my comments about notifications and pipeline interval ? 15:06:49 <_nadya_> nova-notification tests pass on postgress-full but doesn't on mysql 15:07:04 <_nadya_> eglynn: I suggested patch to devstack, yes 15:07:13 cool 15:07:53 <_nadya_> sileht: thanks for comments! but I think that polling interval will not help because it's notification 15:08:20 <_nadya_> and we eventually have reproduced the issue on "the same env" 15:08:55 <_nadya_> and it is reproduced only with all tests. run only our one test is successful 15:09:20 _nadya_: on "the same env" outside of the CI infrastructure? 15:10:03 <_nadya_> eglynn: yes. we've created vm in rackspace and using gating scripts have created the same env 15:10:18 _nadya_: ... so you could control the debug config option for that any way you like? 15:10:40 <_nadya_> eglynn: yep. I think we are close to solution :) 15:10:49 _nadya_: excellent! 15:11:39 <_nadya_> I think that's all from my side. Vadim is preparing new patches for cinder and neutron notifications 15:12:07 <_nadya_> thanks for your help sileht, eglynn and gordc! 15:12:27 np! 15:13:06 cool, moving on then :) 15:13:12 #topic Release python-ceilometerclient? 15:13:25 wait until the patches for new core features are all landed 15:13:37 sounds like a plan 15:13:40 time contrained alarms, selectable aggregates, complex queries etc. 15:13:48 cool 15:14:07 let's try to have a release before/in time for rc1 :) 15:14:18 yes, that would be wise 15:14:43 #topic Outreach Program for Women interns for Ceilometer? 15:14:57 hm I don't see terriyu 15:15:04 #link https://wiki.gnome.org/OutreachProgramForWomen/2014/MayAugust#Schedule 15:15:13 though FWIW I received a mail a few days ago about someone wanting to participate 15:15:26 by that schedule time is getting short for new contributors to rock on up to the project and get something proposed on gerrit 15:15:32 I didn't get any news since a couple of days though 15:15:56 eglynn: we managed that same timeframe and time of the year for terri last year 15:16:02 and it was pretty good, though not that easy 15:16:05 jd__: cool 15:16:09 I think it's never easy anyway :( 15:16:35 I think that if anybody has a something in mind that could be done by intern, i.e. a small blueprint, let me know 15:16:39 <_nadya_> do we need to suggest areas for interns? tasks? 15:16:59 _nadya_: small tasks/bugs are good to have them starting, and then a small blueprint for the actual intership is great 15:17:14 and having a mentor is great too 15:17:37 <_nadya_> jd__: tests for tempest is not ok? :) 15:17:54 _nadya_: it's more a infinite set of tasks :) 15:18:14 yeah something more self-contained is better 15:18:25 it could be though, but I'm not sure it's that thrilling and it's also not directly contributing to Ceilometer, but rather Tempest 15:18:27 with a definite start, middle and end ;) 15:18:56 so just keep that in mind an feel free to fill the OPW wiki with ideas 15:20:37 #topic change semantics of first & last sample timestamps in API resource representation? See LP 1288372 for details 15:20:48 #link https://bugs.launchpad.net/ceilometer/+bug/1288372 15:20:54 ok so the basic problem is that GET /v2/resources is just too darn slow :( 15:21:04 big O(db.meter.count) as opposed to big O(db.resource.count) 15:21:12 ... where db.meter.count >> db.resource.count 15:21:28 ... /me butchers the notation ... ;) 15:21:39 previously it was too darn greedy for memory 15:21:45 (when based on the mongo aggregation framework) 15:21:59 but after I rebased it on a map-reduce, the fundamental problem remains 15:22:19 i.e. walking the entire meter collection to retrieve data about a much smaller number of resources 15:22:33 so we have a couple of related bugs ... 15:22:40 mongo - https://bugs.launchpad.net/ceilometer/+bug/1288372 15:22:48 sqlalchmey - https://bugs.launchpad.net/ceilometer/+bug/1264434 15:22:53 * jd__ stares at eglynn 15:23:03 banana ? 15:23:41 eglynn: is there a solution to that problem rather than dropping this piece of info? 15:24:08 <_nadya_> hehe, in HBase we do vice versa. get meters from resources 15:24:10 and other than "API v3" :) 15:24:12 not so much dropping 15:24:28 more *redefining* the semantic of {first|last}_sample_timestamp 15:24:36 (to be the timestamps for the 1st and last samples ever seen for this resource) 15:25:02 it's a more subtle change to the API that just cutting out those fields 15:25:13 eglynn: I think that would work for me 15:25:21 cool :) 15:25:21 I mean I can't see any downside as an API consumer 15:25:48 in any case the first and last timestamps can't have been that important to callers as they were mistakenly left out of the Resource representation initially 15:25:48 if first_timestamp is < to the timestamp I passed in the query, I'm smart enough to understand that my timestamp is included 15:26:06 timestamp and computer 101 15:26:25 so the aspect I wasn't so sure about was the migration question 15:26:58 we don't currently have any formal migrations for mongo 15:27:08 it's NoSQL, we don't need it 15:27:16 it's all don thru name-mapping, old versus new etc. 15:27:54 but we could derive a first sample timestamp for each resource doc in the resource collection using a once-off query on the meter collection 15:28:35 for new resources this would set when the resource doc is upserted 15:28:54 but for pre-existing resources, there would be no good default 15:29:20 similarly for the last timestamp on old resources that we don't see any more fresh samples for 15:29:54 a once-off batch update could set all those values for all existing resource docs 15:30:05 ... is that a wacky idea for mongo? 15:30:14 not sure it's really worth it imho 15:30:33 jd__: fair enough 15:30:50 I'll go ahead and target https://bugs.launchpad.net/ceilometer/+bug/1288372 for icehouse-rc1 then 15:31:22 ... that's all I have on that topic 15:31:35 sounds like a plan :) 15:31:39 thanks eglynn 15:31:55 #topic Open discussion 15:32:02 hi guys, I have some concerns of the patch https://review.openstack.org/74702 of the FFE BP VmSphere and want to hear your opinions. 15:32:29 llu-laptop: why didn't you put it in the review? 15:32:30 it introduced a new MemoryUsagePollster and its corresponding inspector method which is not supported by libvirt inspector now. 15:32:39 ok, I'll do that 15:33:15 <_nadya_> I have a question about "fix order in assertEqual" 15:33:40 ah yes, we were supposed to get that landed last thing before i3 15:33:44 <_nadya_> As I understood we have a plan to merge it after feature-freeze? 15:34:17 _nadya_: mayyyyybe 15:34:22 I think fix order is just code refactoring 15:34:30 can't land in rc1? 15:34:32 yeah the idea was to leave it last on the i3 queue to avoid more featureful patches needing to be rebased 15:34:36 eglynn: it'll be in rc1 not i3 15:34:56 jd__: yep the i3 ship has sailed 15:35:01 sorry I was late to the meeting, did you already talk about interns? 15:35:25 terriyu: yes 15:35:34 terriyu: backlog here http://eavesdrop.openstack.org/meetings/ceilometer/2014/ceilometer.2014-03-06-15.00.log.txt 15:35:45 <_nadya_> jd__, eglynn, let's merge it :) 15:36:01 terriyu: feel free to add more content if you have more to say :) 15:36:03 _nadya_: yes I think it would be worth getting in for RC1 15:36:20 eglynn: I was talking about vmsphere sorry 15:36:28 jd__: sorry again for being late. I was sleeping in and forgot about the early meeting. 15:36:30 I don't care about this assertEqual stuff 15:36:38 jd__: a-ha, ok 15:36:47 terriyu: np :) 15:37:10 eglynn: I don't want it to be /my/ waste of time and energy :) 15:37:22 cool enough 15:37:40 <_nadya_> eglynn: ok, deal :) 15:39:01 <_nadya_> jd__: I don't want to change your mind but once I spent 40 min beuse of wrong ordering 15:39:50 <_nadya_> and just a plea to decide smth about https://review.openstack.org/#/c/69644/ 15:40:35 * jd__ pats _nadya_ 15:41:04 _nadya_: but I won't block it I just don't want to fight for it neither :) 15:41:26 <_nadya_> jd__: ok, np :) 15:42:35 so about the Outreach Program for Women, I already know one applicant who is interested in Ceilometer specifically 15:42:47 She is trying to set up Devstack and Ceilometer right now 15:43:19 Also, sphoorti who applied the last round for Ceilometer and didn't get in, she is also thinking of applying again 15:44:13 terriyu: so I guess they're both looking for ideas as to a topic? 15:44:15 anyways, if anyone has ideas, there are applicants who are interested! 15:44:43 eglynn: yeah, they would be interested in ideas 15:45:09 cool, let's all get our thinking caps on 15:45:09 the application process closes on March 19, which is pretty soon 15:45:35 o/ i missed this entire meeting... jd__, thanks for taking care of vmware bp 15:46:23 gordc: oh, if you have any ideas for projects that would be suitable for interns, that would be great! 15:46:40 gordc:I think the theme today is "this meeting came really early today" 15:47:07 terriyu: i can ask around. 15:47:34 gordc: cool, the Outreach Program for Women deadline this round is March 19 15:47:40 nealph_: yeah... i hit snooze too many times today 15:49:45 anything else? 15:50:03 we're stuck in migration hell 15:50:12 meaning? 15:50:28 nealph_: sqlachemy-migrate hell? 15:50:56 eglynn: yep. probably take it to the cm room, but is there anyone specifically familiar with that? 15:51:19 looking at the notes it seems like jay pipes was working this a while back... 15:51:33 probably with discuting in #openstack-ceilometer indeed 15:51:44 jd__: k, will take it there. 15:51:51 s/with/worth/ 15:51:55 discussing 15:51:57 * jd__ tired 15:52:22 well, I hereby declare this meeting ended 15:52:45 have a nice day and happily hack 15:52:48 #endmeeting