15:00:08 #startmeeting ceilometer 15:00:08 Meeting started Thu Jul 24 15:00:08 2014 UTC and is due to finish in 60 minutes. The chair is eglynn. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:09 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:12 The meeting name has been set to 'ceilometer' 15:00:16 o/ 15:00:17 o/ 15:00:20 o/ 15:00:21 o/ 15:00:21 o/ 15:00:28 o/ 15:00:36 hey strangers! :) 15:00:41 :) 15:00:41 * eglynn has been a half-time slacker this week 15:00:48 o/ 15:00:50 o/ 15:00:56 o/ 15:00:58 o/ 15:01:11 o/ 15:01:23 #topic juno-2 retrospective 15:01:32 o/ 15:01:32 so juno-2 is tagged and bagged :) 15:01:39 http://tarballs.openstack.org/ceilometer/ceilometer-2014.2.b2.tar.gz 15:01:48 thanks to all! 15:02:03 for the efforts to get stuff over the line 15:02:12 ... also lots of nursemaiding of approved patches through the gate 15:02:48 BTW the gating situation should improve when the 1.3.1 oslo-messaging release is finally cut 15:02:56 https://review.openstack.org/#/c/107915/1/doc/source/index.rst 15:03:17 these are the laggard backports still holding it up ... 15:03:24 https://review.openstack.org/#/q/status:open+project:openstack/oslo.messaging+branch:stable/icehouse,n,z 15:03:39 * gordc crosses fingers 15:03:58 at least nsaje's and sileht's fixes for the fake driver are now safely landed 15:04:34 sileht: BTW should we thinking about jumping straight to the 1.4.0.next release if that appears first? 15:05:27 sileht: ... k, one to think about :) 15:05:42 so in the end, only two j2 BPs needed last minute bumps to juno-3 15:05:54 dedicated-alarm-database was pretty much approved by Monday 15:05:58 ... but the gating Gods chose not to smile upon the many rechecks 15:06:21 grenade-resource-survivability was also code-complete in time by cdent 15:06:42 @eglynn, IBM needs a http dispatcher (Gorden wrote the initial one), I made some changes and I hope it can get in 15:06:43 ... but the QA PTL is declining any new javelin features until the core of javelin2 is fixed 15:07:19 tongli: can you propose a BP spec? 15:07:23 @eglynn, does every new feature in Juno needs a BP? 15:07:38 @eglynn, ok, I will create one. thanks. 15:07:42 tongli: yes, i think so 15:07:43 eglynn, yep using 1.4.0 permit to land the notification publisher patch, that make ceilometer fastest 15:07:45 patch set will be there soon. 15:07:53 tongli: anything chunky at least, beyond an RFE bug 15:08:00 tongli: you betch yah... unless you manage to confuse everyone and sneak it in as a bug. 15:08:00 tongli: cool, thank you sir! 15:08:42 @gordc, that is the http dispatcher your wrote first, I made some changes, so it is more generic, and I think can be used by anyone. 15:08:44 tongli: i can take a look at it. the poc i did was relativelys simple 15:08:54 tongli: cool cool 15:09:02 thanks folks. 15:09:03 sileht: OK if the successor to 1.4.0.0a3 lands before 1.3.1 does, let's just go striaght onto that 15:09:23 gordc: stop disclosing my tactics publicly! 15:09:43 so the discussion of BPs dovetails nicely onto juno-3 planning ... 15:09:50 #topic juno-3 planning 15:10:03 #link https://launchpad.net/ceilometer/+milestone/juno-3 15:10:15 jd__: lol we've all caught on to your schemes. 15:10:24 :-> 15:10:30 so the cupboard is looking a bit bare still for juno-3 15:10:37 mainly bumps from juno-2 15:10:49 eglynn: isn't there supposed to be some Gnocchi stuff? 15:11:02 like are we sure every bp in ceilometer-specs has a bp on launchpad or what? 15:11:28 eglynn: i can probably tackle sql-part2... i might need some vetting by Mike Bayer. 15:11:30 jd__: yeah, do you wanna start filing placeholder BPs in luanchpad to point to the gnocchi specs? 15:11:40 gordc: excellent! 15:11:46 bureaucracy, ya never got enough 15:12:02 jd__: the placeholder is really just a holder for the 'Specification URL' link 15:12:06 jd__: :) 15:12:14 also, i might have found reason for collector performance issues... i might have broken it... still need to verify but also might need some help from Bayer. 15:12:34 jd__: ... I think ttx has cooked up a script to auto-file the LP BPs when the specs land 15:12:40 eglynn: haha the spec for the dispatch references https://blueprints.launchpad.net/ceilometer/+spec/dispatcher-for-gnocchi-integration that does not exist 15:12:42 * jd__ blames ildikov 15:12:58 :) 15:12:58 gordc: is that captured in a bug? 15:12:59 I can blame ttx otherwise, let me know 15:13:10 yeah blame ttx :) 15:13:13 jd__, no blames please :) 15:13:14 https://blueprints.launchpad.net/ceilometer/+spec/http-dispatcher 15:13:20 eglynn: it's the bug related to telemetry tempest tests failing. 15:13:29 jd__: me? my last info that the PTL registers the BP on launchpad after it was approved, so I blame eglynn :) 15:13:30 if anyone can approve it, that will be great. 15:13:36 hahahaha 15:13:37 tongli: you'll need to write up spec. 15:13:46 so the workflow is supposed to be land spec first then file LP BP, but TBH it's a bit counter-intuitive 15:13:57 gordc: a-ha, got it 15:14:00 @gordc, where to write a spec for BP? 15:14:04 https://github.com/openstack/ceilometer-specs 15:14:31 I'm gonna grab popcorns and rant about how much time we lose on those form things 15:14:36 gordc: so this one https://launchpad.net/bugs/1336755 right? 15:14:40 Launchpad bug 1336755 in ceilometer "telemetry_notification_api.py test times out waiting for Nova notification" [Critical,Triaged] 15:14:52 tongli: there's a template and a few past examples you can base off of. 15:15:06 eglynn: ildikov: so who's gonna create that one so we put it on j3? ildikov do you still plan on working on it? 15:15:15 eglynn: yeah... i'm not sure why i didn't just copy/paste link in...i have the window open.lol 15:15:16 @gordc, will do that. thanks. 15:15:24 gordc: FYI I've proposed a temporary skip for that in tempest https://review.openstack.org/108011 15:15:40 jd__: yeap, just had some health and internal work interrupt... 15:15:41 ok. 15:15:46 gordc: so you may need to un-skip the skip if you need further diagnostics on the failure 15:16:03 jd__: eglynn: I will register that BP on launchpad 15:16:11 ildikov: thanks! 15:16:17 ildikov: ok cool! let me know if you need help or someone to blame 15:16:33 is ilya around? i was wondering if he could run his fancy performance script again to see if performance issue can be seen? 15:16:37 jd__: thanks :) 15:16:45 gordc, Ilya is not, but I am :) 15:16:46 gordc: good idea! 15:16:52 hi, how about creating a message listener for keystone messags so that these messages turn to meters, that needs a BP and spec as well? 15:16:59 eglynn, gordc, I'll let him know :) 15:17:16 DinaBelova: thanks! 15:17:27 DinaBelova: this patch might be the reason: https://review.openstack.org/#/c/96200/ 15:17:35 gordc, ok, got it 15:17:39 gordc: apparently Ilya has made some efforts to make the loadgen script etc. more consumable outside the mirantis lab 15:17:47 <_nadya_> tongli: 15:17:53 eglynn, still in progress 15:17:59 :) 15:18:00 tongli: the basics of that have already been done by boris-42 15:18:13 <_nadya_> tongli: I've seen the patch about keystone notifications 15:18:17 @eglynn, not really, we talked yesterday. 15:18:27 tongli: a-ha, k 15:18:42 tongli: but that *could* be more in the nature of a bug-fix 15:18:52 boris patch made ceilometer listens to keystone exchange, but no listener really handles any keystone messages. 15:18:54 that is what is missing. 15:19:34 so we've got quite a few still open specs reviews 15:19:40 @eglynn, ok, all right then, no BP or spec for that. 15:20:05 tongli: step1 of jd__'s bug technique: never ask if a bp is needed. 15:20:07 we've missed the acceptance deadline, but we said we wouln't enforce that structly, given vacation and all 15:20:24 however I think we'll have to draw a line by next week's meeting 15:20:28 @gordc, lol. 15:20:31 eglynn, what about the PaaS notification format, do we have any closure on that, please? 15:20:43 isn't acceptance deadline August 21st, not July 21st? 15:20:46 eglynn: agreed 15:20:49 * jd__ stares at gordc and says "learning you are, young padawan" 15:20:57 jd__: :) 15:21:02 fabiog_: nealph raised that earlier 15:21:13 https://wiki.openstack.org/wiki/Juno_Release_Schedule 15:21:31 * gordc still never seen movie but i get the reference. 15:21:45 nsaje: that's for actual patches, not bp 15:21:53 * jd__ suggests we put more rules so we can have more exceptions to them 15:21:54 nsaje: hmmm, I was going on http://lists.openstack.org/pipermail/openstack-dev/2014-July/039201.html 15:22:26 nsaje: Aug 21st is FeatureProposalFreeze 15:22:54 nsaje: i.e. a deadline for proposal of patches *implementing* the BP 15:23:15 nsaje: whereas SAD was a deadline for the spec being landed in the *-specs repo 15:23:34 ... Kafkaesque, I know 15:23:49 looks like neutron have 83 bps approved by its SAD date for juno-3, terrifying 15:24:07 eglynn: ok, I see 15:24:30 fabiog_: are you planning to update https://review.openstack.org/101282 based on the discussions in Paris? 15:25:01 eglynn: I am working on it 15:25:07 fabiog_: thank you sir! 15:25:17 eglynn, fabiog_ and I are planning to have the discussion about it now and early next week 15:25:25 eglynn: realistically I can have something next week 15:25:30 to define what actually we need to do here and how 15:25:36 DinaBelova, fabiog_: cool! 15:25:43 eglynn: it will probably land code in Kilo-1 15:25:54 What if the implementing code that has been done but haven't been approved by code review? Is it forbidden to check in such code after Aut 21st? 15:26:27 DinaBelova: Hope I could join your discussion! 15:26:32 idegtiarov, ;) 15:26:47 fabiog_: ok, let's look at whether we can expedite that and get something in for juno, maybe without all the bells and whistles 15:27:03 eglynn: I am all for that 15:27:28 fabiog_: (context is that this feature has been on our roadmap since the end of the Havana cycle) 15:27:34 fabiog_: cool :) 15:27:35 :D 15:27:45 for quite a lo-ong time :) 15:28:18 KurtRao: Aug 21st is just a deadline to get the patches proposed to gerrit, not reviewed or landed 15:28:50 KurtRao: it just recognizes the reality that getting patches thru' the review cycle always takes longer than you'd expect 15:28:55 eglynn: great! 15:29:42 k, let's shoot for this day next week to have specs for all the realistic juno BP candidates landed 15:29:45 thanks folks! 15:29:51 shall we move on? 15:30:06 +1 15:30:10 +1 15:30:18 #topic tempest status 15:30:22 a-ha :) 15:30:38 well, folks I have some not really nice news :( 15:30:42 oohhh 15:30:45 https://review.openstack.org/#/c/71259/ - here is the chain of changes 15:30:48 :( 15:30:57 and they still are in the review process 15:31:04 form February to be clear :D 15:31:29 they are really ready to be landed, but they won't land in j2 obviously 15:31:29 DinaBelova, ildikov: so no response on that idea to restart services to temporarily accelerate the polling cycle? 15:31:35 very slow code review process :( 15:31:46 DinaBelova: yeah the juno-2 ship has sailed :( 15:32:12 eglynn: not yet, but this patch is for notifications as I saw 15:32:15 eglynn, as for the restarting - no, no news... this letter was sent, etc 15:32:30 ildikov, yeah - there are ready ones 15:32:34 so part of this may be resolved by this notion being pushed by QA currently 15:32:42 tempest tests not related to the polling 15:32:53 eglynn, they were silent :( 15:32:56 of pushing non-blackbox testing into the functional test suites in the project trees 15:32:58 will continue pinging :( 15:33:02 i.e. removing from tempest 15:33:15 DinaBelova: k, I'll also ping if there's no response bu EoW 15:33:19 DinaBelova: well to be fair that was sent during that midcycle 15:33:30 eglynn: maybe your PTL hat would be useful now to start the discussion about that topic ;) 15:33:30 mtreinish, yes, I know that 15:33:38 it was sent to start the async discussion 15:33:40 I had forgotten about it, until I just saw this 15:33:51 I'll write a reply 15:33:56 mtreinish, as sync one was not available 15:34:09 mtreinish: cool, thanks 15:34:11 mtreinish: np, we realized you guys would be in conclave last week 15:34:12 mtreinish, I'm really ok to have just irc chat :) 15:34:23 mtreinish, or letter responce 15:34:30 whatever is comfortable for you 15:34:55 DinaBelova: either or, I don't want to take up your meeting time though 15:34:56 mtreinish, and some reviews for the https://review.openstack.org/#/c/71259/ cain will be also really appreciated 15:35:09 mtreinish, ;) 15:35:13 thank you sir so much 15:36:13 eglynn, so I hope to see the ready tests asap (they are really almost done - and different comments have been fixed) 15:36:30 DinaBelova: sure, I'll try to take another look at that after I get back fromlunch 15:36:42 FYI mtreinish & I also had a brief discussion yesterday on what's coming up re. pushing out functional testing into the projects http://paste.openstack.org/show/87779 15:36:49 as for the polling tests - let's see what we'll have after mtreinish & co will look on them :) 15:36:56 ... will be followed up on the ML and at the PTLs meeting the week after next 15:37:27 mtreinish, thank you sir so much! I know that's tough to keep in mind so much things you have folks in the QA sphere here 15:37:59 ok, anything else on Tempest? 15:38:35 no, i guess no 15:38:35 moving on ... 15:38:38 #topic TSDaaS/gnocchi status 15:38:56 jd__: the floor is yours sir! :) 15:40:08 sure 15:40:20 so we got merged the new aggregation code using pandas 15:40:26 that now compute things correctly :-D 15:40:39 jd__: \o\ :) 15:40:52 we also have the new indexer feature to have details when listing resources 15:41:24 jd__: will we go ahead and land that with the accept-extension idea? 15:41:52 ... *without the accept-extension idea? 15:42:27 jd__: a-ha, you've got it in the latest :) 15:42:29 https://review.openstack.org/#/c/109315/1/gnocchi/rest/__init__.py 15:42:30 cool 15:43:22 yes 15:43:30 and I'll try to continue otherwise nothing new 15:44:17 jd__: I've given pauldix a few weeks since mid-cycle to work on those influxdb issue we talked about 15:44:34 jd__: ... I'm planning to pick that up again next week with his latest 15:44:51 ok 15:45:14 jd__: ... so we may need to touch base on how the "restricted" rention policies might look like in the API 15:45:27 * jd__ nods 15:45:40 ... i.e. in the case where a driver is not capable of giving infinite degrees on freedom on that 15:45:44 cool 15:45:56 DinaBelova: anythng to report on opentsdb? 15:46:08 eglynn, /me still debugging 15:46:15 hope to end up with it today 15:46:19 DinaBelova: cool enough, thanks! 15:46:26 also Julien said really nice idea to me 15:46:41 oh yeah, wanna share? 15:46:45 I needed to write my own rest client to opentsdb 15:46:47 yeah 15:46:59 and I've placed that code in gnocchi code - like utils, etc 15:47:02 yeah I don't want to bring an opentsdb client in gnocchi as it should be useful standalone 15:47:11 jd__, indeed 15:47:14 yeah I saw that patch being -2'd 15:47:32 so I've posted the change to infa config 15:47:41 to create new stackforge project 15:47:47 not merged yet, but still 15:48:12 now I keep code in gnocchi to debug, but I'll move to the standalone client when the repo will be created 15:48:25 DinaBelova: cool, thanks! 15:48:30 eglynn, np 15:48:49 anything else related to gnocchi? 15:49:26 #topic open discussion 15:50:01 reminder: proposal deadline for summit talks is next Monday July 28th 15:50:36 so if anyone is chewing on proposing something related to ceilo, best to get your skates on :) 15:50:37 just for a quick update, it looks like s.dague is getting pretty closeon the javelin2 stuff 15:50:47 cdent: nice one! 15:51:04 I keep an eye on the various reviews, and they are creeping forward 15:51:09 cdent: ... so that would mean the restriction on new features for javelin would be lifted? 15:51:20 eventually, but can't guess when 15:51:35 cdent: cool, thanks for the update 15:53:25 ... going, going ... 15:53:41 ... gone! :) 15:54:03 thanks all for the productive meeting, let's call that a wrap! 15:54:08 #endmeeting