15:00:41 #startmeeting ceilometer 15:00:41 Meeting started Thu Sep 4 15:00:41 2014 UTC and is due to finish in 60 minutes. The chair is eglynn. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:43 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:46 The meeting name has been set to 'ceilometer' 15:00:46 o/ 15:00:51 o/ 15:00:51 o/ 15:00:54 o/ 15:00:54 o/ 15:01:08 hey folks! 15:01:12 #topic Juno-3 status and FFEs 15:01:23 #link https://launchpad.net/ceilometer/+milestone/juno-3 15:01:37 so kudos to everyone who got their patches landed 15:01:55 the gate rush was pretty extreme this time round 15:02:06 o/ 15:02:19 <_nadya_> o/ 15:02:19 eglynn, as usual.. 15:02:26 _nadya_, good evening :) 15:02:31 that effectively pushing the de facto milestone deadline a day or two earlier 15:02:37 o/ 15:02:42 o/ 15:03:11 so I discussed FFEs with our release manager earlier 15:03:24 here's the gist of what we agreed ... 15:03:51 nealph_: paas-event-format-for-ceilometer gets a pass by default as it documentation only 15:04:30 nealph_: do you feel this can be closed out and landed by next week? 15:04:47 eglynn: new patches up this morning (US) 15:04:53 so, yes. 15:05:10 Barring any resistence from the PaaS service folks. 15:05:21 nealph_: cool, let's try to get that landed by early/mid- next week so 15:05:56 eglynn: will do! 15:06:01 nealph_: thanks! 15:06:19 o/ 15:06:31 moving onto the next FFE 15:06:40 gordc's bigger-data-sql has been granted one week of grace-time to reconcile the different performance test results seen on different mysql versions etc 15:06:56 ... and to combine patches to minimize migrations etc. 15:07:04 gordc: does that sounds sufficient? 15:07:17 eglynn: cool. i'll keep an eye on it. 15:07:25 eglynn: just hoping gate holds up 15:07:42 gordc: (context is there was resistence to granting a longer extension given the possiblity for introducing regressions) 15:08:13 gordc: yeah I hope that the gate latency will go down a lot, now that we're over the juno-3 hump 15:08:52 gordc: presumably we're including in the sql-a ORM->core switch as well as the schema changes in that BP? 15:09:21 eglynn: i think it might help given ilya's results... 15:09:52 yeah, the test results reported by DinaBelova and Ilya do sound promising 15:09:52 eglynn: we can re-evaluate if necessary... ie. worse case is a 10% performance bump 15:10:03 gordc, eglynn - yeah - Ilya had to go afk right now 15:10:09 but he started mysql upgrade 15:10:12 cool 15:10:16 on the nodes of the lab 15:10:41 so he'll continue the tests 15:10:50 so this one is extra important because it's on the TC radar 15:10:57 (via the gapa analysis) 15:11:12 eglynn, yes, we keeep track on this for sure 15:11:20 so bigger-data-sql is definitely the highest priority RC1 task for the team 15:11:23 cool, thanks! 15:11:57 k, moving on to the last FFE 15:12:34 dunno if Edwin is around, but his impi-support BP will also get a week or so in grace-time 15:12:49 Edwin is not here, but I can pass the info to him tomorrow 15:13:09 so we can sort out the issues around aggregation of sensor-specific and system-wide aggregated data 15:13:13 I saw he just posted all the patches https://review.openstack.org/#/q/project:openstack/ceilometer+topic:bp/ipmi-support,n,z 15:13:51 and also get clarity on the behavior on non-Intel/nodemanager hosts 15:13:56 llu-laptop: cool, thanks! 15:14:33 llu-laptop: BTW do you know Edwin's IRC nick? ... I don't think I've seen him on the channel 15:15:05 sorry, don't know either :) 15:15:11 k 15:15:14 i'll ask him tomorrow 15:15:19 cool, thanks! 15:15:36 eglynn: i've just been emailing.. been pretty responsive (aside from time zone diff) 15:15:57 gordc: yeah +1, I've done the same 15:16:04 other than that, we still have a lot of bugs on the RC1 slate 15:16:10 #link https://launchpad.net/ceilometer/+milestone/juno-rc1 15:16:52 so I'll need to cull that bug list early next week to remove anything that isn't getting review traction 15:17:26 anything else on juno-3? 15:17:31 eglynn, about "Central agent pollsters not using discovery mechanism" bug 15:17:42 I took a look on the final version this morning 15:17:51 #link https://bugs.launchpad.net/bugs/1364352 15:17:52 Launchpad bug 1364352 in ceilometer "Central agent pollsters not using discovery mechanism" [High,In progress] 15:18:03 ir seems to work okay with multinode devstack, multiple central agents, etc 15:18:26 so /me hopes to see it soon in the repo : 15:18:27 :) 15:18:41 DinaBelova: cool, good to have that confirmation 15:19:04 it's missed juno-3 at this stage obviously, but definitely need to land it for RC1 15:19:17 yeah, for sure 15:19:24 testing it now myself 15:19:24 hehe, i see -1 by fabiog_ :) 15:20:17 well, anyway, this change is quite important :) 15:20:27 absolutely, agreed 15:20:58 but we've a bit of time before RC1 for objections such as fabiog_'s to be aired and addressed 15:21:34 shall we move on? 15:21:38 i'm digging into a ceilometerclient bug as well... see any tempest ceilometer-alarm-evaluator log 15:21:46 http://logs.openstack.org/53/96653/9/check/check-tempest-dsvm-full/d8edccb/logs/screen-ceilometer-alarm-evaluator.txt.gz?level=TRACE 15:21:59 if someone knows ceilometerclient really well, feel free to take this over from me. 15:22:05 gordc: yeah, I saw that mentioned on IRC last night, didn't get a chance to look 15:22:32 gordc: do you have a sense of where the problem lies? 15:22:33 eglynn: yeah. i can't see it locally so i'm sort of blindly patching right now 15:22:40 gordc, frankly speaking this does not look like ceilo client error for me >_< 15:23:29 eglynn: i think it's how we're using keystone discovery... i noticed there's a fallback action the other clients use so i just ported taht over 15:23:37 i'm not exactly sure if it fixes this case specifically tough 15:23:39 though* 15:24:00 i'll keep digging for now 15:24:10 hmmm, that log signature rings a bell, I think ildikov saw a similar issue 15:24:59 gordc: IIRC she solved it simply by switching the port numbers 15:25:01 gordc: if you don't make any headway on that today and want to punt it to me I can pick it up tomorrow and after. 15:25:19 i.e. s/35357/5000/ 15:25:34 eglynn: I recall that but wasn't there some debate about the correctness of doing so? 15:25:54 eglynn: yeah... could be a devstack patch as well... just wanted to bring up it's broken. :) 15:26:01 cdent: yeah, I don't recall all the details TBH, discussed about a month ago on IRC 15:26:09 visibly broken 15:26:21 https://bugs.launchpad.net/python-ceilometerclient/+bug/1351841 15:26:22 Launchpad bug 1351841 in python-ceilometerclient "python-ceilometerclient does not works without v3 keystone endpoint" [High,Triaged] 15:26:39 eglynn: that's probably the bug ildiko was looking at. 15:27:08 gordc: k, that's definitely high priority so we'll need to line up a python-ceilometerclient release with the final juno tag 15:27:50 ... I'll follow up with ildikov when she gets back from her work trip 15:28:41 k, moving on ... 15:28:44 #topic Proposed design summit changes 15:29:06 just a heads up on a few summit changes that being discussed currently 15:29:35 first, a move towards more open/collaborative scheduling of the design session tracks 15:30:25 avoiding using the webapp for proposals, instead just using etherpads etc. to agree the content of the tracks 15:30:47 doesn't sound a million miles from the way we did this for the Juno track 15:31:09 except with s/etherpad/googledoc/ in that case 15:31:32 any opinions on how folks would like to approach that? 15:32:06 eglynn: would you create a kilo or summit folder in ceilometer specs too? 15:32:20 fabiog_: yep, good point 15:32:24 so people can start populate with potential ideas 15:32:52 fabiog_, good point, yeah 15:32:53 fabiog_: I'll create a kilo directory and any outstanding BP proposals can be rebased onto that 15:33:09 eglynn: thanks that helps a lot 15:33:37 are we still have the group table with whiteboard? 15:33:54 fabiog_, do you mean project pods? 15:33:55 just one thing to note though about ceilo-specs, some of the other projects are -2'ing all kilo BP proposal to avoid distracting reviewers from juno-3 and RC1 15:34:11 DinaBelova: yes, I didn't remember the name 15:34:12 personally I think that's a bit OTT 15:34:32 fabiog_: did you see that design summit reloaded thread on the ML? 15:34:52 #link http://lists.openstack.org/pipermail/openstack-dev/2014-August/044169.html 15:34:58 eglynn, ++ - that's weird a while - especially keeping in mind the fact we ahave not so mane specs 15:35:08 eglynn: I will give it a look and shut up ;-) 15:35:24 fabiog_: that was the other summit change I wanted to discuss ... 15:35:57 TL;DR: keep Friday for contributor meetups in a format similar to the project pods in ATL 15:36:57 fabiog_: TBH I'm not sure why we don't just do the project pods in the same way again as ATL, as it would avoid losing 33% of the design session slots 15:37:23 eglynn: I guess there is not enough space like in Atlanta 15:37:44 I guess price per sqft in Paris is higher than in Atlanta ;-) 15:37:51 fabiog_: yeah that could be it ... in any case, I raised that idea here: http://lists.openstack.org/pipermail/openstack-dev/2014-September/044914.html 15:38:38 eglynn: do you think we can ask Hyatt to set-up pods for the different projects? 15:39:00 eglynn: being one of the conference hotels, they may agree on that, just a thought 15:39:38 a-ha, k, yeah ... though it would obviously add to the cost 15:39:59 ... I'll ask though to get some clarity, as I think the pod was really useful last time round 15:40:11 eglynn: maybe not, we are closing the hotel anyway for Openstack ;-) 15:41:34 * gordc thought pod was useful as well. 15:42:45 cool, moving on ... 15:42:53 #topic Tempest 15:43:17 DinaBelova: I saw your patch to unskip that nova notification test 15:43:18 eglynn, as for the Tempest - here it is tests unskip https://review.openstack.org/#/c/115212/ 15:43:22 yeah :D 15:43:37 and Matthew said he'll add his +2 after gate will feel better 15:43:47 so I just need finally to find Sean 15:43:58 he's kind of ghost :) 15:44:17 DinaBelova: we can probably leave it for now and bring it up when gate calms. 15:44:29 gordc, yeah, for sure 15:44:37 DinaBelova: cool, sounds reasonable 15:44:43 cdent has a special kind of skill in shaking +2s out of Sean ;) 15:44:46 so after it'll be merged, it'll be possible to remove -2s from other tetss 15:44:52 eglynn :D 15:44:57 It's bizarre but useful. 15:45:08 cdent, quite agree here :) 15:45:08 cdent: lol 15:45:25 cdent, do you have talent to remove -2's from Sean? 15:45:27 DinaBelova: "remove -2s from other tests" ... that's the other stalled notification tests? 15:45:35 eglynn. yeah 15:45:41 not for nova, but for other services 15:45:43 I haven't yet had the opportunity 15:45:50 cdent, you'll have one ;) 15:45:53 DinaBelova: ... that would be great! 15:45:54 I promise 15:46:05 I'm guess the skill comes from the fact that I have more commits on devstack, grenade and tempest than I do on ceilo 15:46:08 eglynn, yeah, I really hope to remove these -2s.... 15:46:21 cdent, yeah, I guess so 15:46:41 * eglynn thinking again about the TC gap analysis, it would be good to able to declare victory on the tempest task as it's very nearly over the line 15:47:15 eglynn - indeed, especially keeping in mind the fact, that now all the jobs are feeling ok for notification tests 15:47:33 on any backend we have job now 15:47:40 both on Mongo and MySQL 15:48:16 so my mongo experimental jobs are just ok for the notifications, although MySQL feels okay as well now 15:48:30 gordc and I decided it was kind of temporary issue 15:48:30 DinaBelova: cool ... BTW I've curated the gap analysis wiki a bit but feel free to fix anything I've mispresented 15:48:34 with the gate itself 15:48:36 #link https://wiki.openstack.org/wiki/Governance/TechnicalCommittee/Ceilometer_Gap_Coverage 15:48:47 eglynn, ok, I'll take a look, thanks 15:49:09 anything else relevant to Tempest? 15:49:18 eglynn, I guess no 15:49:30 cool 15:49:43 BTW TSDaaS/gnocchi status punted to the next meeting as jd__ is on vacation 15:50:01 #topic open discussion 15:50:34 eglynn: when do you estimate the "juno3 busy" gate queue be ready for accepting patches for RC1? I saw ceilometer rc3 tarball is already cut? 15:52:02 llu-laptop: yep, juno-3 is tagged and bagged :) 15:52:08 llu-laptop: but I think we can start accepting RC1-oriented patches onto master right away 15:52:32 llu-laptop: IIUC then these must be backported to milestone-proposed in order to get into RC1 15:53:05 llu-laptop: ... so it's the backport, not the landing on master, that represents the acceptance into the RC 15:53:16 ... at least that was the way we did for past releases 15:54:06 eglynn: got that 15:54:35 cool 15:56:03 llu-laptop: actually I tell a lie ... the branch has changed a little for juno 15:56:07 IIRC, it'll be the "proposed/juno" branch you backport to 15:56:16 not "milestone-proposed" 15:57:06 the old "milestone-proposed" branch naming has been dropped because we've stopped backporting in the pre-milestone 1-2 day period, as we did before Juno 15:57:50 k, we're up against the shot-clock as per usual ... 15:58:00 thanks all for a productive meeting! 15:58:08 \o 15:58:16 bye! 15:58:26 bye 15:58:26 #endmeeting ceilometer