15:01:48 <eglynn-lunch> #startmeeting ceilometer
15:01:49 <openstack> Meeting started Thu Mar 26 15:01:48 2015 UTC and is due to finish in 60 minutes.  The chair is eglynn-lunch. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:50 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:52 <openstack> The meeting name has been set to 'ceilometer'
15:02:00 <cdent> o/
15:02:05 <fabiog> hi
15:02:10 <idegtiarov> o/
15:02:12 <ityaptin> o/
15:02:17 <_nadya_> o/
15:02:23 <eglynn> howdy all
15:02:26 <fabiog> eglynn: you got a long lunch :-)
15:02:50 <llu-laptop> o/
15:02:51 <eglynn> fabiog: I'm a hungry man :)
15:03:24 <eglynn> #topic kilo-rc1
15:03:27 <eglynn> #link https://launchpad.net/ceilometer/+milestone/kilo-rc1
15:03:44 <eglynn> kio-3 is in the bag so the focus now falls on rc1
15:03:50 <eglynn> *kilo-3
15:03:54 <gentux> o/
15:03:57 <prad> o/
15:04:02 <eglynn> the idea to minimize the number RCs we need to cut
15:04:22 <eglynn> as any RCn with n>1 requires a very solid justification
15:04:36 <eglynn> #link https://wiki.openstack.org/wiki/Kilo_Release_Schedule
15:04:56 <eglynn> so the approx start of the RC season is the week of April 9th
15:05:13 <eglynn> but nothing to stop us going early if no further blockers
15:05:45 <eglynn> let's a quick sanitize on the kilo-rc1 list
15:06:15 <gordc> o/
15:06:21 <eglynn> cdent: this one has been punted from milestone to milestone ... https://bugs.launchpad.net/bugs/1348818 (PYTHONHASHSEED)
15:06:22 <openstack> Launchpad bug 1348818 in Cinder "Unittests do not succeed with random PYTHONHASHSEED value" [Medium,In progress] - Assigned to Eric Harney (eharney)
15:06:23 <sileht> o/
15:06:34 <eglynn> should we just kill it at this stage?
15:06:45 <jd__> o/
15:06:48 <cdent> yeah, nobody knows how to fix it because nobody really knows why it is happening
15:06:53 <cdent> it's not in ceilo itself
15:07:21 <eglynn> cdent: close as invalid, or punt into the great blue yonder?
15:07:21 <cdent> its somewhere in wsme extension if I recall
15:07:48 <gordc> cdent: yeah... or that's what i think
15:08:06 <eglynn> OK, punting to next
15:08:09 <cdent> It might be that the right thing is to let the hashseed be random for all the tests but the docs test. I'll do that.
15:08:11 <gordc> the error is in the doc check, that's all i know
15:08:32 <eglynn> I'm inclined to punt this also ... https://bugs.launchpad.net/ceilometer/+bug/1376915
15:08:34 <openstack> Launchpad bug 1376915 in Ceilometer "Access to sensitive audit data is not properly restricted" [High,In progress] - Assigned to Matthew Edmonds (edmondsw)
15:08:34 <gordc> cdent: that works for me. (if it works in gate)
15:08:39 <cdent> as there are ordering errors leaking into the other tests
15:09:00 <eglynn> cdent: thanks ... pls re-target if it makes sense to do so
15:09:23 <eglynn> any objections to punt LP#1376915?
15:09:45 <eglynn> ... disagreement on the bug definition, lacking traction, already punted from various milestones
15:10:06 <llu-laptop> eglynn: let's punt that
15:10:10 <eglynn> k, re-tagetted to "next" also
15:10:16 <gordc> works for me. i think there's a viable workaround. i'll comment on bug
15:11:05 <eglynn> the rest of https://launchpad.net/ceilometer/+milestone/kilo-rc1 is either landed already or seems like a kosher thing we'd want to have fixed in the final kilo release
15:11:40 <llu-laptop> gordc: is your https://bugs.launchpad.net/ceilometer/+bug/1436077 landing in rc1?
15:11:41 <openstack> Launchpad bug 1436077 in Ceilometer "minimise hmac calls to avoid performance degradation" [Medium,In progress] - Assigned to gordon chung (chungg)
15:12:06 <eglynn> gordc: what do reckon of the importancre field for https://bugs.launchpad.net/ceilometer/+bug/1434711 ?
15:12:07 <openstack> Launchpad bug 1434711 in Ceilometer "middleware and profiler notifications generate non-metric meters" [Undecided,In progress] - Assigned to gordon chung (chungg)
15:12:49 <gordc> eglynn: it'd be good to have... i missed them initially.
15:13:06 <gordc> llu-laptop: that'd be good to have as well... i haven't thought of solution to your comment just yet though.
15:13:13 <ildikov> o/
15:13:26 <eglynn> gordc: k, sounds like a "medium"
15:14:01 <eglynn> I'm gonna say "medium" also for the mongo 3.0 bug, since we don't explicitly support this version or test against it in upstream CI
15:14:01 <gordc> eglynn: cool cool. works for me. i think most of my items are 'nice to have's
15:14:21 <eglynn> coolness
15:15:33 <eglynn> so RC1 looks pretty quiet so far, so I guess that could explode once the distros start testing initial packages based on kilo-3
15:15:51 <eglynn> RDO kilo-3 packages are iminent IIUC
15:16:14 <eglynn> dunno about trusty, anyone tracking those packages?
15:16:24 <gordc> rc1 is what day again?
15:16:44 <gordc> nm, i read history.
15:16:49 <jd__> 9th April IIRC
15:16:54 <gordc> cool cool
15:17:43 <eglynn> well, seems it's a "moveable feast"
15:17:46 <eglynn> like Easter :)
15:18:45 <eglynn> we can cut RC1 early if we like, or possibly delay it slightly if some critical issue explodes the week of April 9th
15:19:19 <eglynn> trick seems to be identify the potential release blockers before w/c April 9th to avoid a needless RC2
15:20:10 <eglynn> anything else on RC1?
15:20:31 <eglynn> #topic gnocchi
15:21:09 <jd__> "This is exciting" – The Wall Street Journal
15:21:15 <sileht> ahahah
15:21:25 <jd__> "You'll never regret pastas" – Times
15:21:30 <jd__> anything else? :D
15:21:42 <eglynn> ROFL :)
15:21:48 <jd__> so we release our first alpha aka 1.0.0a1
15:21:59 <eglynn> jd__: you mentioned also the trusy packaging work is ongoing?
15:22:01 <jd__> I think nobody tested it so NO BUG!
15:22:12 <jd__> yeah zigo_ is going to take of packaging for Debian
15:22:19 <jd__> I don't know about Ubuntu though
15:22:30 <_nadya_> congrats! :D
15:22:40 <jd__> other than that the project is pretty good shape and we make good progress every week
15:22:46 <jd__> it should be now more than usable
15:22:52 <jd__> the API is getting very sexy
15:23:02 <eglynn> jd__: excuse my ignorance, but are the ubuntu packages usually just forks of the deb work?
15:23:04 <jd__> we are still waiting on infra to rename the repository to openstack/gnocchi
15:23:13 <eglynn> jd__: ... or totally independent?
15:23:19 <jd__> eglynn: more or less yeah, that depends
15:23:24 <zigo_> Yeah, will do.
15:23:32 <zigo_> jd__: As always, Ubuntu guys will wait until I do it... :)
15:23:38 <jd__> hehe
15:24:00 <jd__> RC1 milestone is at https://launchpad.net/gnocchi/+milestone/1.0.0rc1 for whose who want to track
15:24:09 <eglynn> cool, we have something similarish with the Fedora->RDO->RHEL packaging foodchain
15:24:17 <jd__> we have a few patches that need reviews if anyone has time
15:24:25 <idegtiarov> hi! we are interested in adding other backend to indexer, we think about mongodb, is it make sense for you?
15:24:34 <jd__> idegtiarov: yes that makes sense
15:24:47 <idegtiarov> do we need to write a spec?
15:24:56 <jd__> idegtiarov: no, just open a blueprint to track your work
15:25:14 <idegtiarov> jd__, thank will do
15:25:19 <idegtiarov> *thanks
15:25:20 <jd__> if they are details you want to discuss we can do it on the mailing list
15:25:26 <jd__> but it should be pretty straight forward
15:25:42 <jd__> for people following at home I started to write an InfluxDB driver
15:25:51 <eglynn> cool
15:25:56 <jd__> but InfluxDB if far from being complete for our needs
15:26:10 <jd__> and 0.9 is still in rc stage and breaks/improves a lot of things
15:26:19 <jd__> so I'll follow that, but it's not going to be in 1.0
15:27:29 <cdent> I think they are too busy getting all silly valley up in the shiz
15:27:31 <jd__> also sileht patches for our new aggregation API has been merged into Ceilo
15:27:45 <sileht> and the heat resources too
15:27:52 <jd__> anything else you want to add sileht cdent ?
15:28:03 <cdent> I got the functional tests finally green
15:28:17 <eglynn> jd__: influx driver won't be in gnocchi 1.0, or support for that specific version of influx won't be?
15:28:28 <idegtiarov> jd__, thank you sir! We were going to start discussion in irc or it would be better on mailing list?
15:28:29 <cdent> Which means there will be pretty good live coverage of swift, ceph and postgresql and mysql in the gate
15:28:35 <sileht> I just started to create the history stuffs for resource attributes
15:28:44 <jd__> eglynn: influxdb won't be, there's no way to build a full driver currently with InfluxDB 0.8, and probably not yet with 0.9
15:28:54 <eglynn> jd__: cool, got it
15:29:10 <jd__> idegtiarov: in general if it's long I prefer ML, if it's just a few questions feel free to ping on IRC
15:29:26 <idegtiarov> jd__, got it thanks.
15:29:32 <jd__> yeah what's sileht is working on is exciting – come on be excited
15:29:46 <gordc> yay?
15:29:47 <jd__> we'll have history of resources states, think about flavor changes in VM
15:29:54 <jd__> isn't that awesome?
15:30:06 <gordc> yay!
15:30:14 <jd__> thanks gordc, pass by my office for your check
15:30:36 * gordc looks for flights.
15:30:37 * jd__ . o O (plane so expense he'll never do it hihi)
15:30:45 <jd__> s/expense/expensive/
15:30:48 <gordc> jokes on you
15:31:32 <eglynn> sileht: so this is to allow reconstructing the resource state timeline from old attribute values?
15:31:42 <fabiog> jd__: are you creating a new resource when the flavor changes and link it to the previous one?
15:31:47 <eglynn> sileht: awesome :)
15:32:11 <sileht> eglynn, yes
15:32:13 <fabiog> jd__: or you adding a history inside the same resource?
15:32:14 <jd__> fabiog: that's our idea indeed
15:32:37 <sileht> eglynn, I plan to duplicate the resource
15:32:42 <fabiog> jd__: ok, that makes sense. The second wont :)
15:32:59 <jd__> fabiog: cool 👍
15:33:24 <fabiog> sileht: yes, you want that. Also if I remember correctly nova may give it a different resource id
15:33:50 <eglynn> sileht: coolness, so the events persistence won't be needed now for that purpose?
15:34:41 <sileht> eglynn, if the gnocchi resource track the attribut I guess yes
15:35:04 <sileht> eglynn, but ceilometer event is still more flexible for that
15:37:15 <eglynn> yeah, say looking at the raw resize events, rather than say querying the resource attribute history to find all instance resizes that happened in the last month
15:39:43 <eglynn> sileht: ^^^ reason I ask is that this very usecase has come up before with customers doing chargeback
15:40:02 <eglynn> ... I've generally recommended using events instead of a classic ceilo samples query
15:40:22 <eglynn> ... the trade-off might be the other way though for the new gnocchi resource attribute history
15:40:34 <eglynn> anyhoo
15:41:15 <eglynn> anything else pasta-related?
15:41:31 <jd__> I think that's all
15:41:33 <sileht> eglynn, got it, I just started to store the history, the API should allow that, I hope.
15:41:50 <eglynn> sileht: coolness, thanks
15:41:58 <eglynn> #topic open-discussion
15:42:52 <sileht> the bot is sick
15:43:05 <cdent> it's because eglynn came on as eglynn-lunch
15:43:09 <cdent> and then switched
15:43:15 <eglynn> a-ha
15:43:25 * eglynn goes back to lunch :)
15:43:34 * cdent is jealous and hungry
15:44:08 <eglynn-lunch> #topic the-topic-that-missed-all-the-discussion
15:44:32 <cdent> When is the right time to start noodling about summit discussions?
15:45:21 <eglynn-lunch> cdent: usually in the ~3 weeks preceeding, we make the discisions on the track agenda
15:45:33 <cdent> cool, thanks
15:45:39 <fabiog> eglynn-lunch: cdent I think last time we had a spreadsheet to populate, that was very useful
15:45:43 <eglynn-lunch> cdent: for Paris, we dispensed with al lthe formal CFP business
15:46:07 <eglynn-lunch> fabiog: yep, let's spin up something similar for Vancouver
15:46:13 <gordc> fabiog: yeah, spreadsheet worked pretty well
15:47:03 <eglynn-lunch> cdent: part of the reason it's so late is that there's often horse-trading around the slots-per-track and the distribution of tracks during the week
15:47:14 <cdent> yeah, I remember that
15:47:41 <eglynn-lunch> cool
15:48:09 <cdent> oh that reminds me: I convinced gordc, jokingly, that there's going to be a bowling night and somehow it evolved into a bowl off between projects
15:48:10 <eglynn-lunch> ... but never too early to start cooking up ideas for sessions
15:48:40 <eglynn-lunch> cdent: a-ha, cool ... ceilo versus who?
15:48:47 <cdent> THE WHOLE WORLD
15:48:52 <eglynn-lunch> LOL :)
15:48:55 <cdent> dunno :)
15:49:15 <cdent> I think we'd be better off going for a nice stroll in stanley park, actually.
15:49:52 <gordc> cdent: you're the afterhours planner.
15:49:54 <sigmavirus24> cdent: glance accepts your challenge =P
15:49:56 * sigmavirus24 kids
15:50:13 <cdent> gordc is our ringer
15:50:42 <cdent> If I'm the afterhours planner we're eating sushi for hours every night on someone else's tab.
15:50:54 <gordc> cdent: sigmavirus24: i got fantastic technique... just as much power as possible.
15:51:14 <sigmavirus24> gordc: that's how you loose an arm in the gutter
15:51:23 <sigmavirus24> goes flying right off with teh ball when you least expect it
15:51:43 <gordc> it's a risk admittedly.
15:52:08 <sigmavirus24> On the other hand, if the arm doesn't force the ball into the gutter, it'll take out all the pins
15:53:36 <cdent> I think that means we're done.
15:54:06 <fabiog> cdent: I think so too :)
15:54:12 <eglynn-lunch> cdent: I concur ... arms in the gutter is the signal to call it a wrap :)
15:54:21 <gordc> lol! back to testing.
15:54:24 <eglynn-lunch> thanking y'all for your time
15:54:30 <eglynn-lunch> #endmeeting ceilometer