15:00:20 <gordc> #startmeeting ceilometer
15:00:20 <openstack> Meeting started Thu Apr 23 15:00:20 2015 UTC and is due to finish in 60 minutes.  The chair is gordc. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:21 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:24 <cdent> o/
15:00:24 <openstack> The meeting name has been set to 'ceilometer'
15:00:26 <ityaptin> o/
15:00:31 <idegtiarov_> o/
15:00:36 <eglynn> o/
15:01:17 <sileht> o/
15:02:48 * gordc taps finger... waits for 3minutes past
15:03:28 <cdent> irony
15:03:42 <gordc> that's what the 'x' does
15:03:47 <gordc> #topic kilo-rc requests/bugs
15:04:07 <gordc> so we have 4 items tracked for rc2
15:04:20 <gordc> i think we're just waiting on cdents patch for time.sleep issue
15:04:33 <gordc> anyone else have patches we want to get into rc2?
15:04:38 <ildikov> o/
15:04:50 <gordc> there's this new bug: https://bugs.launchpad.net/ceilometer/+bug/1447647
15:04:50 <openstack> Launchpad bug 1447647 in Ceilometer "Unhandled SqlAchemy exception indexing trait with large text" [Undecided,New] - Assigned to Rohit Jaiswal (rohit-jaiswal-3)
15:04:51 <cdent> gordc: are you counting the eventlet one that I've started?
15:05:03 <gordc> cdent: yep.
15:05:10 <cdent> cooooooool
15:05:34 <gordc> cdent: https://bugs.launchpad.net/ceilometer/+bug/1447599 ?
15:05:35 <openstack> Launchpad bug 1447599 in Ceilometer "mongodb retry logic uses unpatched time.sleep to wait" [High,In progress] - Assigned to Chris Dent (chdent)
15:05:39 <cdent> yessir
15:05:53 <gordc> cool cool, yeah that's the last one i'm tracking
15:06:29 <gordc> the sqlalchemy exception bug above i don't think is a regression (but it might be) i don't know if we want that targetting for kilo
15:07:41 <gordc> thoughts? i'll push it to liberty (unless it's a regression from my event 'fixing' work)
15:08:00 <cdent> gordc: seems worth checking if it is a regression and if it is not, don't worry
15:08:08 <cdent> bugs happen
15:08:17 <gordc> cdent: agreed.
15:08:38 <gordc> anyone other blocking issues?
15:08:48 <gordc> any*
15:09:27 <gordc> skipping to ceilometerclient then
15:09:47 <gordc> #topic recurring: ceilometerclient status
15:10:10 <gordc> 1.2 is release for liberty.
15:10:31 <gordc> i'm currently waiting on dhellmann and release team for a 1.0.14 release for kilo
15:11:02 <gordc> i've backported most of 1.1.0 patches to 1.0.14 but please let me know if there's any other blockers
15:11:38 <gordc> no? no. cool. and moving on. :)
15:11:46 <gordc> #topic summit session proposals
15:12:10 <gordc> we have a few entries already https://docs.google.com/spreadsheets/d/1vlEe75uZc5faytpvIMPrQdViOOsRLnfeThk_Rp_lIvI/edit#gid=0
15:12:28 <gordc> just a reminder: April 30 will be our 'aggressive' cut off date
15:13:02 <gordc> i don't know if anyone wants to give a quick elevator pitch on what they're proposing (considering we have so much time)
15:13:13 <cdent> I've attached an etherpad to the one I'm proposing: https://etherpad.openstack.org/p/ceilo-multi-identity
15:14:00 <cdent> I think it would be pretty useful if people read that and added to it because I think there are a lot of issues underlying the stuff I've written about that might warrant rather a lot of talking/planning, including some before summit, because time at summit will be a bit limited and those topics are pretty wide ranging
15:14:20 <gordc> i'd probably suggest people look at that because it's pretty broad and we should probably do as much work pre-summit for it.
15:14:50 <cdent> we could get away with doing nothing, but that would be a bit sad
15:15:06 <cdent> we need to kick things up a notch
15:15:27 <gordc> noooo!!! why didn't you do this under eglynn's watch.
15:15:36 <eglynn> gordc: LOL :)
15:15:50 <cdent> is jd__ around? some of that stuff is related to things he's talked about
15:16:01 <cdent> gordc I blame eglynn for being a quitter
15:16:03 <jd__> o/
15:16:15 <cdent> if he wasn't it would be under his watch
15:16:19 <gordc> cdent: lol
15:16:57 <jd__> sorry reading the backlog
15:17:45 <jd__> I like this Etherpad from what I can glance
15:18:04 <jd__> I don't know if there are a session to talk about that but I'd be happy to
15:18:10 <eglynn> gordc: I missed the conclusion on https://bugs.launchpad.net/ceilometer/+bug/1447647 ... kilo-rc2 also?
15:18:10 <openstack> Launchpad bug 1447647 in Ceilometer "Unhandled SqlAchemy exception indexing trait with large text" [Undecided,New] - Assigned to Rohit Jaiswal (rohit-jaiswal-3)
15:18:21 <cdent> I've linked it to a session jd__, but my hope is that more people will write on the etherpad before then
15:18:34 <cdent> so that more juices are flowing
15:18:45 <gordc> eglynn: i think we'll defer it (if it was always liked that)
15:18:55 <jd__> I have a pretty clear vision of what I want to achieve in this regard I'll push it, through our dear PTL I hope, and probably with cdent too!
15:18:57 <eglynn> gordc: ok
15:19:02 <gordc> i need to check quickly after meeting.
15:19:52 <gordc> jd__: cdent: i guess we can pull out 1 or 2 things and narrow topic down (assuming we nominate it for summit)
15:20:01 <msly> o/
15:20:09 <eglynn> https://etherpad.openstack.org/p/ceilo-multi-identity sounds like a session that could set the tone for the ceilo track at summit
15:20:34 <eglynn> so ideally a fishbowl session at or close to the start of the track?
15:20:47 <gordc> eglynn: works for me.
15:20:47 <jd__> gordc: as you wish, but this is something that _is_ going to happen during Liberty :)
15:21:16 * gordc drags his feet.
15:21:31 <gordc> if no one has any other elevator pitches for their proposals...
15:21:48 <cdent> I just want to point out that I have a crush on prad's proposal
15:22:12 <eglynn> cdent: "declarative" carches your eye? :)
15:22:23 <gordc> cdent: same... but with less weird wording.
15:22:23 * cdent is so predictable
15:23:24 <gordc> before switching, here's the latest summit schedule: https://docs.google.com/spreadsheets/d/1VsFdRYGbX5eCde81XDV7TrPBfEC7cgtOFikruYmqbPY/edit#gid=569963128
15:23:41 <gordc> it looks like we've got an extra fishbowl
15:24:06 <gordc> and we need to run between working session and fishbowl rooms on wed
15:24:15 <jd__> cdent: what's prad proposal?
15:24:38 <cdent> jd__: declarative notification handling: stop writing code to process a notification, write yaml or json
15:25:07 <jd__> cdent: yay, huge work to do in this area, and so little has been done unfortunately
15:25:32 <jd__> I think that's an area that had a lot of noises (e.g. schema) for years and still nothing came out AFAIK
15:25:44 <cdent> I worry, though, that all this cool tech will be for naught if we don't "keep" the users, whatever that means.
15:25:44 <jd__> so let's make more noise? with the hope or result maybe? :)
15:26:02 <cdent> thus the "kick it up a notch" comment earlier
15:26:12 <cdent> we need to make ceilo not just cool but also relevant
15:26:18 * gordc gave up on schema... too many opinions.
15:26:23 <gordc> openstack maded me jaded
15:26:59 <jd__> yeah, too much bikeshedding as usual
15:27:12 <jd__> not enough active push of visions
15:27:32 <jd__> but we'll manage to get there sometimes :)
15:27:49 * gordc is silently pushing cadf from dark corners
15:28:26 <gordc> neva! :)
15:28:33 <gordc> anything else on summit?
15:28:55 <gordc> #topic recurring: Gnocchi status
15:29:09 <gordc> jd__: sileht: cdent: anything?
15:29:20 <jd__> yeah so we released 1.0.0rc1
15:29:28 <jd__> https://pypi.python.org/pypi/gnocchi
15:29:34 <gordc> \o/
15:29:36 <sileht> congrats !
15:29:40 <jd__> https://launchpad.net/gnocchi/1.0/1.0.0rc1
15:29:59 <cdent> that long description needs some work
15:29:59 <jd__> I cut stable/1.0 branch and now we are in trouble because the devstack-gate does not like the branch name
15:30:04 * cdent makes a bug as a reminder
15:30:18 <jd__> so we are trying to fix that with cdent and sileht
15:30:34 <jd__> and it seems we made some progress thanks to cdent so that's cool
15:30:42 <jd__> that's a blocker to release rc2 and the final 1.0.0
15:30:47 <jd__> so I hope we fix that ASAP
15:31:11 <sileht> jd__, because you want to release 1.0.0 with oslo.db 1.8 ?
15:31:12 <jd__> other than that I had a few people contacting me because they are trying and are interested into Gnocchi, even without OpenStack
15:31:16 <jd__> so that's pretty exciting
15:31:28 <jd__> sileht: no
15:31:31 <gordc> cool!
15:31:49 <gordc> jd__: i posted your blog to ops-list. hopefully people read it... and bug you.
15:31:51 <jd__> sileht: but we'll still need the VENV hack from cdent for e.h. stable/1.1 I think
15:32:19 <jd__> someone even read my Gnocchi blog post and sent me a mail in the case I could hire him to work on it lol
15:33:03 <gordc> jd__: you hear the story of the person who outsourced his job? he lived a good life.lol
15:33:24 <jd__> haha
15:34:16 <gordc> looks like everything is being managed in gnocchi territory...
15:34:25 <jd__> yep
15:34:26 <gordc> #topic Open discussion
15:36:05 <gordc> this is some proper silence... i'll let it linger for a minute
15:36:06 <cdent> Does anyone have any pointers to "cool hacks done with ceilometer". Something that actually does something. I keep getting people asking me "what's ceilo for" and the answers never satisfy
15:36:43 <cdent> "you can get the stddev on your cpu usage" is insufficient
15:36:54 <gordc> i think the cool hacks are trade secrets
15:37:41 <gordc> i know one person is using the file dispatcher to process data in json rather than database... interesting? mildly interesting?
15:39:03 <cdent> gordc: I mention that sort of thing on the etherpad I linked before: people dispatch elsewhere to do cool stuff. What does that say?
15:39:33 <gordc> flexibility?
15:40:01 <cdent> that we should kill all work on the destination side of the dispatcher
15:40:07 <cdent> and let people do that part themselves
15:40:38 <cdent> yet weirdly that is where we put most of our efforts
15:41:05 <gordc> i don't get it... isn't that what is happening already? people have their own use case. they pushed it up stream for others?
15:44:17 <cdent> gordc my point was that we spend a lot of energy on the storage layer
15:44:45 <cdent> but the point that actually allows people to do stuff happens well before that
15:46:15 <gordc> true. i guess there's many 'exit points' in ceilometer. that said, for a complete solution (if one has the use case for it), the storage part is pretty important.
15:46:51 <gordc> i guess the ops meeting will give us a pretty good idea of what is important parts of ceilometer
15:46:59 <cdent> ++
15:47:03 <gordc> but based on feedback, it's the storage and polling agents
15:47:16 <gordc> feedback==complaints
15:47:27 <cdent> I think we need to be careful to distinguish parts that cause trouble vs parts that are important
15:47:50 <cdent> people complain rightly that the storage doesn't work well, but that doesn't make it important, just required
15:47:59 <cdent> that's a subtle but important difference
15:49:02 <gordc> fair enough. i don't really have a response to that.lol
15:49:36 <cdent> I buy the car because I want it to get me places. It does, however, need brakes.
15:49:57 * cdent proposes we end the meeting ten minutes early so we can shut me up
15:50:11 * gordc obliges
15:50:13 <gordc> anything else?
15:50:51 <gordc> nope? have a good day/evening/sleep folks
15:50:56 <gordc> #endmeeting