15:01:00 #startmeeting ceilometer 15:01:03 Meeting started Thu Jul 17 15:01:00 2014 UTC and is due to finish in 60 minutes. The chair is eglynn. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:01:04 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:01:06 The meeting name has been set to 'ceilometer' 15:01:08 hey y'all! 15:01:10 o/ 15:01:11 o/ 15:01:15 o/ 15:01:17 o/ 15:01:30 #topic Juno-2 status 15:01:33 o/ 15:01:33 o/ 15:01:41 #link https://launchpad.net/ceilometer/+milestone/juno-2 15:02:03 let's go thru' the BPs and get a feel for what's gonna land 15:02:46 eeek! 15:02:49 eglynn, https://blueprints.launchpad.net/ceilometer/+spec/mongodb-events-feature has been merged already :) 15:02:51 netsplit :( 15:02:53 :d 15:02:56 wheeeeeeeeeeeee 15:03:05 what's going on? 15:03:12 DinaBelova: updated 15:03:15 cool 15:03:18 llu-laptop: netslpit 15:03:26 tsk, net split again 15:03:38 nsaje: arithmetic transformers looking good? 15:03:57 cdent: ironic notifications looking good? 15:04:28 yup, just want a verify from you and potentially a poke from lucas (or his peers) to make sure the metadata is payload is happy 15:04:36 (I added some more tests) 15:04:51 cdent: great :) 15:04:58 nsaje: around? 15:05:17 eglynn: looking good 15:05:37 nsaje: cool, thanks :) 15:05:43 sileht: around? 15:06:15 cdent: grenade/javelin, what's the thought on that? 15:06:48 jogo is not excited: until javelin is operating in grenade he doesn't see much point adding features on it (he reiterated that on review today) 15:07:08 cdent: so, prolly bump to juno-3 would be sensible? 15:07:25 however Sean is more excited: he's pretty ready to get javelin running and is waiting for I'm not sure what for a patchset that does the intergration to be passed 15:07:32 cdent: working on getting it in grenade as we speak. hoping to get it working this week 15:07:44 cdent, jogo: OK, I'll hold off bumping in that case 15:07:46 speak of the devil! 15:07:49 :) 15:08:00 pradk: around? 15:08:03 I am excited for new javelin features once the basics are working 15:08:19 eglynn, hello 15:08:21 jogo: cool, and EoW is realistic? 15:08:34 jogo have you seen the discussion eglynn and I are having around how to manage semantics in the javelin tests? 15:08:45 eglynn: thats the goal, will know more in a few hours 15:08:53 jogo: cool 15:08:59 pradk: confident on FWaaS/VPNaaS for juno-2? 15:09:03 cdent: right now I just want it to work, so no 15:09:42 it's kind of germane to that actually: our disagreement is because 'create' doesn't fully work (depending on your definition of "work") 15:10:08 eglynn, so far yes.. i should at least be able to get it in partially.. there are a couple of deps from neutron side which we might have to wait on for other metrics. Other than that, so far on track 15:10:52 cdent, jogo: so the key question is whether we chould run javelin/create+check on the old version, or just javelin/create and keep check to post-upgrade? 15:11:25 I think it's agreed that is should be just create upgrade check 15:11:25 pradk: cool, sounds like it'll be enough to declare vistory for the milestone 15:11:32 but until create is "correct" that's not reliable 15:11:33 cdent: cool 15:12:26 yea hopefully, i can get the resource discovery tests working soon.. getting close 15:12:39 eglynn, as for the tempest testing (for j2) - lots of tests are ready to be merged, but due to the QA sprint folks are busy this week.... 15:12:50 llu-laptop: SNMP-improvement isn't on the list because the spec hasn't landed 15:13:05 https://review.openstack.org/104113 15:13:13 eglynn, vrovachev and I will ping all QA folks asap.. 15:13:24 DinaBelova: great, thanks! 15:13:36 eglynn - I still hope to see and least all notifications tests merged in j2 15:13:47 at least* 15:14:10 DinaBelova: did Vadim make any headway with the intermittent nova notification test timeout, do you know? 15:14:12 llu-laptop: yes i'm still doing the coding, but it's risky for j2. My hope is to get the code ready before j2, and get it merged early j3 15:15:01 eglynn, that's no reproducible for him :( 15:15:02 sorry, last msg should be directed to eglynn :( 15:15:19 llu-laptop: having patches up up by j2 would be great, thanks for the update 15:15:47 eglynn`- that's really strange thing - he tried several times to catch it - no results yet :( 15:15:54 eglynn: as for the polling testing in tempest, some mails already landed on the dev list trying to solve the issues there 15:16:13 DinaBelova: yeah, it's happening pretty regularly in the gate according to logstash 15:16:20 eglynn: hopefully we will get some response after the qa sprint 15:16:30 ildikov: what's the $subject? 15:16:39 ildikov, eglynn - yep, as QA folks have meetup - no way they will answer quicker... 15:17:01 eglynn - [openstack-dev] [qa] Test Ceilometer polling in tempest 15:17:15 #link http://lists.openstack.org/pipermail/openstack-dev/2014-July/040434.html 15:17:44 eglynn: yeap, that's it :) 15:17:47 eglynn, although polling tests were not supposed to be part of j2 (too much work even with basic ones) 15:17:59 ildikov, DinaBelova: did you discuss with on IRC also with the qa folks? 15:18:12 eglynn, they went to the sprint :) 15:18:12 eglynn: we couldn't reach them 15:18:17 a-ha, got it 15:18:28 they are lost for this week :) 15:18:33 eglynn: that is why we started with the mail as async version 15:18:38 Sean travelled to Germany on Friday 15:18:52 ildikov, DinaBelova: fair enough 15:18:55 so even previous week was a little bit lost 15:19:09 so other stuff for juno-2 ... 15:19:22 this bug would be great to see fixed https://bugs.launchpad.net/bugs/1336193 15:19:25 Launchpad bug 1336193 in pecan "Timeout test_coordination_assign gate error" [Undecided,Confirmed] 15:19:37 nsaje: has a potential fix in the olso-messaging fake driver 15:20:03 we were waiting on a oslo-messaging 1.3.1 release to get our hands on that 15:20:12 hi there, probably someone can help me - I can't upload image to Glance and see the following error in Glance logs: http://paste.openstack.org/show/86954/ - how to fix it? 15:20:20 however that's now looking like it may not happen 15:20:51 tnurlygayanov: redirect to #openstack-glance 15:21:19 tnurlygayanov, wrong chat :) 15:21:45 instead we may have to just straight to the 1.4 series ... i.e. >=1.4.0.0a3 15:22:19 possible as part of an overall dependency upgrade a la https://review.openstack.org/103029 15:22:32 anyone see anything problematic in that? 15:23:13 eglynn, no from my side 15:23:32 eglynn - just silent nice update :) 15:24:21 ok anything else relevant to juno-2? 15:24:45 ildikov: how is the admin guide looking? 15:25:05 eglynn: kind of on track :) 15:25:21 eglynn: but I do not think that the whole story will fit in j-2 15:25:40 ildikov: ok, shout if you need extra eyes and/or content contributions 15:26:03 eglynn: sure, will do 15:26:24 there's some bugfixes for python-ceilometerclient in gerrit that need core reviews (seems only llu-laptop looks at that repo these days :-) ) 15:26:35 I guess it'd be good if they landed for juno-2? 15:26:57 nsaje: yeah we'd need to cut a new python-ceiloclient release 15:27:24 nsaje: the client releases are decoupled from the main release cycle 15:27:38 eglynn: oh right, I remember that now 15:28:39 just on juno-2 admin in general ... 15:28:48 I need to take half-days off next week for family reasons 15:28:52 but will be contactable 15:28:59 (so just shout on email re. anything relevant to the milestone) 15:29:16 ... also ttx will not be doing the release this time 15:29:24 (it'll be russellb instead pushing the buttons) 15:29:47 ok shall we move on? 15:29:59 o/ 15:30:09 some praying is also involved 15:30:25 ttx: LOL :) 15:30:30 ttx has scripted and documented the process well :) 15:30:43 haha :) 15:30:52 #topic gnocchi update 15:31:00 * eglynn channels jd__ 15:31:09 hehe :) 15:31:22 so work is progressing on the aggregation/truncation fixes discussed at midcycle 15:31:46 fyi, /me still working on opentsdb driver draft - hope to post it tomorrow (at least, its parts) :) 15:31:53 this patch is up for review https://review.openstack.org/105185 15:32:22 DinaBelova: cool! ... by draft, do you mean a WIP/strawman patch? 15:32:45 eglynn, no, I mean normal change - but I think I'll need to add smth there 15:32:53 I'm not sure 15:33:12 DinaBelova: cool enough 15:33:14 some of my efforts there might require some opensdb patching :) 15:33:34 a-ha, you're finding opentsdb issues? ... nice 15:33:50 eglynn, like retention thing - it's not fully implemented there via rest api - just through hbase directly 15:34:13 but as we use happybase+thrift agent to work with it from python we can't set ttl there :) 15:34:23 as opentsbd is java one - I may fix it there :) 15:34:35 and it'll work without any issues :) 15:34:51 a-ha, k 15:34:59 in other gnocchi news ... 15:35:07 ildikov's ceilo/gnocchi dispatcher spec landed 15:35:10 https://github.com/openstack/ceilometer-specs/blob/master/specs/juno/dispatcher-for-gnocchi-integration.rst 15:35:23 \o/ :) 15:35:47 just a link to the gnocchi+opentsdb ML: https://groups.google.com/forum/#!topic/opentsdb/9O57MfpRXI0 15:36:01 trying to start working on it now, j-2 is not a realistical target, aiming for early j-3 15:36:42 ildikov, hehe :) 15:36:47 ildikov: cool 15:36:55 k, move on from gnocchi? 15:37:00 eglynn, +1 15:37:03 #topic trusty switchover in the gate 15:37:31 so as you've probably noticed, clarkb has enabled trusty nodes in the gate 15:37:54 so finally the ceilo py27 is actually running the mongodb scenario tests! :) 15:37:58 \o/ 15:38:06 \o/ 15:38:21 against mongodb version 2.4.9 IIRC 15:38:26 yeah~ 15:38:32 eglynn, :) 15:38:49 so the question is, are we ready to enable a tempest variant running against mongo? 15:38:53 great! 15:39:21 seeing as running against sql-a was always just a workaround for mongodb of the correct version not being available on precise 15:39:28 eglynn, actually we need to try - because I see no serious issues here to be met 15:39:36 DinaBelova: cool 15:39:49 eglynn, only if it'll be found in the process of trying :) 15:39:57 does anyone feel like proposing such a change to infra? 15:40:13 eglynn, me or Vadim may do that :) 15:40:23 Is the idea to run both sql-a and mongo, or just one? 15:40:24 DinaBelova: excellent! :) 15:40:33 cdent: both probably 15:40:35 good 15:40:37 cdent, I guess for the first time both 15:40:50 eglynn, oh 15:40:51 https://review.openstack.org/#/c/96316/ 15:41:26 eglynn - it was a while ago, I need to remember that change 15:41:27 ;d 15:41:34 DinaBelova: yeah f20 was run into issues with skew in the nodepool allocation algorithm 15:41:40 a-ha 15:41:45 yes, exactly 15:41:49 ok, good 15:41:54 DinaBelova: (effectively discriminates against less-used distros) 15:42:01 :D 15:42:41 though ianw had a priority-aging patch up that addressed that 15:42:58 not sure if it actually landed in the end or was abandoned 15:43:12 landed :) 15:43:14 https://review.openstack.org/#/c/101110/ 15:44:01 eglynn, a-ha, /me looking 15:44:29 even so, resistence might be reduced for a tempest-mongodb variant running on trusty as opposed to f20 15:44:41 eglynn, yes, for sure 15:44:51 because of the duration of support question with fedora 15:44:53 I just found that old change in my reviews 15:45:06 and tried to get what it was about :D 15:45:16 the qa folks want >=18 months support for any distros they gate on 15:45:31 eglynn, yep, I remember your ML thread about that 15:45:51 DinaBelova: yeap, so I think the path would be smoother if the new tempest variant was proposed for trusty 15:46:07 k, move on? 15:46:11 eglynn, agreed, ok, /action for me 15:46:24 +! 15:46:26 +1 15:46:43 #action DinaBelova propose tempest-mongodb variant on trusty 15:46:48 :) 15:46:52 #topic opens discussion 15:46:59 guys, SAD is 20th, right? It seems that we still have quite specs to review. 15:47:06 #link https://review.openstack.org/#/q/status:open+project:openstack/ceilometer-specs,n,z 15:47:12 llu-laptop: good point 15:47:33 llu-laptop: I think we'll have to treat that freeze as more a "slush" 15:48:06 but yes, please do review and let's get any specs that are close over the line if possible 15:48:23 e.g. the llu-laptop's SNMP spec just needs a +1 15:50:03 some small number of those specs are hung up on proposers on holiday 15:50:29 cdent: good point 15:50:41 fabiog flagged that at mid-cycle 15:50:53 he'll be back next week IIRC 15:51:05 indeed 15:52:09 so I indicated to fabiog that we'll be somewhat flexible with the spec acceptance deadline because of that 15:53:12 who else lost money on kilo? ;) 15:53:25 was that the decision? 15:53:33 where did it happen? 15:53:34 yeap 15:53:41 * eglynn bet the farm on kepler 15:53:52 kilo is booooooooooooring 15:53:54 cdent: on the openstack (non dev) ML 15:54:08 ah, i'm less frequent on that one 15:54:30 cdent:understandable given the sheer volume. :) 15:54:43 cdent: http://lists.openstack.org/pipermail/openstack/2014-July/008447.html 15:55:28 meh 15:55:29 eglynn, I was voting for Kyoto :) 15:55:40 that came in second 15:55:52 ... there were worse options I guess 15:55:53 looks like no more follow the city/steet name convention 15:56:13 llu-laptop:+1 :( 15:56:20 is kyoto a japan city? 15:56:21 llu-laptop: yeah, it all went to hell when we allowed grizzly ;) 15:56:32 eglynn, :D 15:56:43 llu-laptop: yeah, but there was a tenuous connection to Paris also I think 15:57:18 ... {Rue|Place|Avenue|Metro} de Kyoto ;) 15:57:54 ok, we're up against the shot clock as usual 15:58:07 anyone got anything else they want to raise up before we close? 15:58:30 no, not from me 15:58:42 k, let's call it a wrap 15:58:55 thanks for the productive meeting! 15:59:00 thanks! 15:59:07 #endmeeting