14:59:44 <eglynn> #startmeeting ceilometer
14:59:45 <openstack> Meeting started Thu Dec  4 14:59:44 2014 UTC and is due to finish in 60 minutes.  The chair is eglynn. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:59:47 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:59:49 <openstack> The meeting name has been set to 'ceilometer'
14:59:52 <eglynn> hey y'all!
14:59:56 <fabiog> hey
14:59:59 <nealph> hello!
15:00:18 <eglynn> #topic Kilo-1 blueprints
15:00:20 <llu-laptop> o/
15:00:26 <_nadya_> o/
15:00:27 <_elena_> o/
15:00:32 <eglynn> #link https://launchpad.net/ceilometer/+milestone/kilo-1
15:00:33 <DinaBelova> o/
15:00:45 <eglynn> not looking too bad at this stage
15:00:51 <cdent> o/
15:01:11 <idegtiarov> o/
15:01:12 <eglynn> it would be great to get the two blocked BPs unblocked by EoW
15:01:20 <eglynn> expecially ready-ceilometer-rbac-keystone-v3
15:01:37 <fabiog> eglynn: I'll do my best ... caught in a lot of "internal" things ...
15:01:54 <eglynn> fabiog: coolness, thank you sir!
15:02:08 <sileht> o/
15:02:28 <gordc> o/
15:02:33 <jd__> o/
15:02:54 <fabiog> eglynn: can we merge the spec for RBAC? I noticed it did not go in yet ...
15:03:06 <eglynn> fabiog: when I was reviewing https://review.openstack.org/115717 I tried to figure out if other any services were "domain-aware" as yet
15:03:17 <eglynn> fabiog: seemed only heat was using the concept in anger
15:03:25 <eglynn> fabiog: is that correct, do you know?
15:03:37 <ityaptin> o/
15:03:40 <fabiog> eglynn: I think V3 support is not really there for services yet
15:03:56 <IvanBerezovskiy> o/
15:04:00 <fabiog> eglynn: I will ask Morgan about V2 deprecation timeframe.
15:04:04 <eglynn> fabiog: (and yes, I think we can probably merge the spec with the understanding that it'll be a phased thing, and the domain support will come later)
15:04:11 <eglynn> fabiog: yeah, that's a good point
15:04:27 <eglynn> fabiog: I'd imaging the deprecation path will have to long, since so many services depend on it
15:04:33 <eglynn> I'd *imagine
15:05:25 <fabiog> eglynn: yeah, also there were discussions event to skip V3 in favour of V4, hierarchical projects ... but I haven't seem too much action around it
15:05:26 <eglynn> slightly related to releasing, here's the slate for the first release off stable/juno
15:05:29 <eglynn> #link https://launchpad.net/ceilometer/+milestone/2014.2.1
15:05:54 <eglynn> fabiog: yeah, I'd imagine that'll be Lemming timeframe (as opposed to Kilo)?
15:06:31 <fabiog> eglynn: yeah they will probably start in Lemming (I didn't know that is the name :-))
15:07:22 <eglynn> fabiog: cool ... not the official name, but seems to be the standard joke when refering to the L* cycle :)
15:07:42 <eglynn> BTW on that stable/juno release, if anyone is interested in getting more involved in stable-maint, let me know
15:07:47 <eglynn> backports are fun! :)
15:07:57 <eglynn> ... no really, they are
15:08:24 <eglynn> k, moving on from releases
15:08:26 <eglynn> #topic Packaging of the merged compute/central agents
15:08:45 <eglynn> IvanBerezovskiy wanted to discuss this
15:09:18 <IvanBerezovskiy> we have now two packages for both agent. How it should be after merging Dina's patches?
15:09:31 <IvanBerezovskiy> Do we need in two packages for backward compatibility or not?
15:09:35 <eglynn> IvanBerezovskiy: I made some comments in the specs review https://review.openstack.org/#/c/119077/11/specs/kilo/merge-compute-and-central-agents.rst
15:09:47 <DinaBelova> #link https://review.openstack.org/#/q/status:open+project:openstack/ceilometer+branch:master+topic:merge-compute-and-central-agents,n,z
15:10:34 <DinaBelova> eglynn, am I right that the main idea is something like this: let's keep two packages for compute/central agents
15:10:41 <DinaBelova> going to the new code
15:10:50 <DinaBelova> using the namespaces option
15:11:04 <DinaBelova> and create new package with merged agent?
15:11:04 <eglynn> yeah I'd expect packagers to make the transition by adding a net-new openstack-ceilometer-polling package
15:11:08 <eglynn> ... but to maintain support for the existing openstack-ceilometer-central & openstack-ceilometer-compure packages also
15:11:23 <gordc> did we ever figure out how to handle scenario where we have one agent on compute and one agent elsewhere but both in same group... wouldn't compute tasks go to foreign agent and fail?
15:11:28 <eglynn> for a allinone PoC => install penstack-ceilometer-polling
15:11:44 <gordc> disclaimer: i haven't been following this spec.
15:11:59 <llu-laptop> that means openstack-ceilometer-central & openstack-ceilometer-compure packages depends on openstack-ceilometer-polling, right?
15:12:30 <eglynn> gordc: hmmm, the merged agent would have to join multiple partition groups surely?
15:12:31 <IvanBerezovskiy> they should depend in this scenario, I think
15:12:38 <DinaBelova> gordc, they won't be in one group :)
15:12:54 <DinaBelova> the group-prefix depends on the namespaces
15:12:57 <eglynn> gordc: (one for the central partitioning, the other for the compute partitioning)
15:13:00 <DinaBelova> for now
15:13:14 <DinaBelova> if it will be other type of setting - via pollster-list
15:13:18 <gordc> cool cool. it'd be a doc thing.
15:13:33 <gordc> carry on. :)
15:13:35 <DinaBelova> I suppose to have all of them being in groups defined by this list as well
15:13:39 <DinaBelova> a-ha, cool
15:13:41 <eglynn> llu-laptop: for the Fedora packaging, all the openstack-ceilometer-central & openstack-ceilometer-compure packages really provide is the upstart/systemd config for the service
15:13:53 <eglynn> llu-laptop: i.e. not actual python code
15:14:05 <llu-laptop> eglynn: got that, thx
15:14:19 <IvanBerezovskiy> eglynn, if we'll have 3 packages (for compute, central and common-polling), what about puppets? we need in update to use third package
15:14:20 <eglynn> llu-laptop: all those packages then depend on a common package which contains all the python code
15:14:20 <llu-laptop> so there will be a ceilometer-common where all those depends on
15:14:40 <eglynn> IvanBerezovskiy: again I mentioned that also in the review comment linked above
15:15:19 <eglynn> TL;DR: we'll need a new puppet manifest for the combined agent
15:15:21 <DinaBelova> eglynn, what's the process of contributing to the packages?
15:15:32 <eglynn> DinaBelova: for Fedora?
15:15:54 <DinaBelova> eglynn, well....
15:16:01 <DinaBelova> I don't know process at all >_<
15:16:16 <DinaBelova> so I'm basically interested in Fedora, CentOS and Ubuntu :D
15:16:26 <DinaBelova> hehe
15:16:28 <eglynn> DinaBelova: here's the info on the Fedora package https://admin.fedoraproject.org/pkgdb/package/openstack-ceilometer/
15:16:51 <eglynn> DinaBelova: from which the CentOS and RHEL packaging (RDO and RHOS) is also derived
15:17:00 <DinaBelova> a-ha, cool
15:17:04 <llu-laptop> DinaBelova: I guess the process will be different, at least from my colleagues experience of packaging for FC & Ubuntu
15:17:13 <DinaBelova> so basically that's contrubution process
15:17:16 <DinaBelova> llu-laptop, a-ha
15:17:20 <DinaBelova> do you have any details?
15:17:33 <eglynn> DinaBelova: becoming a Fedora packager involves gradually building up karma, commit rights etc.
15:17:46 <eglynn> DinaBelova: I can dig out some links on that later
15:17:46 <llu-laptop> DinaBelova: sorry I don't, just heard from him
15:17:51 <DinaBelova> eglynn, thanks!
15:17:56 <DinaBelova> eglynn, it'll be really nice
15:18:06 <DinaBelova> llu-laptop, ok, I'll try to investigate :)
15:18:09 <eglynn> DinaBelova: for the .deb packaging, I don't know much about how it works, but I'm guess similar
15:18:21 <DinaBelova> eglynn, I suppose the common process should be the same
15:18:23 <eglynn> DinaBelova: Chuck Short (zul) would be a good source to ask about that
15:19:06 <DinaBelova> eglynn, zigo I guess is the nice point to contact :)
15:19:09 <DinaBelova> ok, cool
15:19:22 <DinaBelova> ok, let's move on I guess
15:19:27 <eglynn> DinaBelova: ah yes, of cource, zigo too :)
15:19:30 <DinaBelova> that's just an announcement :)
15:19:33 <eglynn> yep, let's move on
15:19:41 <eglynn> #topic "TSDaaS/gnocchi status"
15:20:14 <eglynn> #link https://review.openstack.org/#/q/project:stackforge/gnocchi,n,z
15:20:48 <eglynn> notable landings
15:20:53 <eglynn> ... sileht got the gnocchi dispatcher and related circularity patches in \o/ :)
15:21:25 <eglynn> ... jd__ landed some fine-grained RBAC policys for the entity CRUD API
15:21:49 <eglynn> also Entities are no longer Entities :)
15:22:03 <eglynn> now they're known as Metrics
15:22:06 <DinaBelova> eglynn, /me fighting with gnocchi and devstack :) finally won it -> some opentsdb progress is coming
15:22:41 <eglynn> DinaBelova: \o/ nice one! ... I'll steal your notes on the fun you had getting that working :)
15:22:51 <DinaBelova> :D
15:22:56 <DinaBelova> :)
15:23:02 <DinaBelova> np :)
15:23:19 <idegtiarov> metrics looks better for me then entities )))
15:23:24 <jd__> sileht is also working on aggregation
15:23:28 <eglynn> idegtiarov: +1
15:23:36 <DinaBelova> idegtiarov, at least more understandable about what's it in fact :)
15:23:38 <jd__> and I'm gonna continue policy implementation over archive policies and resources
15:23:46 <idegtiarov> DinaBelova:  for sure
15:23:48 <eglynn> jd__: coolness
15:23:53 <DinaBelova> jd__, got problems with policies btw
15:23:54 <jd__> I'm also working on a new documentation generator running real queries against Gnocchi to have real examples up to date
15:23:56 <DinaBelova> on the devstack + gnocchi
15:24:02 <eglynn> jd__: yeah, I haven't seen that aggregation patch update yet, looking forward to reviewing it
15:24:03 <jd__> DinaBelova: let's debug in #openstack-ceilometer then :)
15:24:14 <DinaBelova> ok, I found the workaround
15:24:18 <DinaBelova> but it's tricky
15:24:31 <sileht> I will try to land the devstack patch too
15:24:35 <DinaBelova> jd__, debug is coming :)
15:24:43 <jd__> otherwise things look pretty good
15:25:03 <eglynn> sileht: excellent
15:25:22 <eglynn> jd__: coolness, anything else on gnocchi?
15:25:41 <jd__> not yet :D
15:25:56 <llu-laptop> speaking of gnocchi, I saw 2 bps for ceilometer of adding new DB backend support, what's our general guide here? encourage them to add support into gnocchi instead of ceilometer?
15:26:09 <eglynn> llu-laptop: cassandra you mean?
15:26:22 <DinaBelova> llu-laptop, yes, good question!
15:26:28 <llu-laptop> I saw one cassandra, and maybe another
15:26:34 <eglynn> llu-laptop: yeah, I tried to gently nudge towards gnocchi
15:26:42 <_nadya_> and db2 woke up...
15:26:45 <DinaBelova> cassandra change has one more issue actually...
15:26:48 <eglynn> llu-laptop: but IIRC the contributor seemed some interested in "classic" ceilo
15:27:03 <DinaBelova> too much cassandra heplers in the ceilo code was proposed
15:27:17 <DinaBelova> much -> many, was -> were
15:27:30 <gordc> llu-laptop: i have no intend on making elasticsearch db for gnocchi... apologies.
15:27:30 <DinaBelova> imho no matter gnocchi or ceilo that should be separated lib
15:28:00 <gordc> DinaBelova: yeah, the cassandra patch has a bunch of code i think should really be in it's client... or maybe i'm too dumb to understand it... let's assume the former.
15:28:16 <DinaBelova> gordc, ++
15:28:36 <jd__> I think it's ok if it's for event support
15:28:53 <jd__> for samples, it seems like it's not a good idea anymore if we plan to drop the old API
15:28:56 <llu-laptop> gordc: right, elasticsearch
15:29:43 <eglynn> seems like the effort of adding a driver would be rewarded with more future "shelf-life" on the gnocchi side
15:29:53 <_nadya_> jd__: yep, but how to explain this to contributors? :)
15:30:17 <DinaBelova> _nadya_, -2s?
15:30:17 <jd__> _nadya_: copy paste what I wrote? ;)
15:30:18 <DinaBelova> :(
15:30:25 <_nadya_> jd__: hehe
15:31:20 <eglynn> _nadya_: well we can't dictate what they spend their effort on, the best we can do is "nudge" towards gnocchi
15:31:32 <_nadya_> DinaBelova: why -2? because "we refuse to accept new changes in drivers, we have new policy"?
15:31:49 <nealph> _nadya_: +1. aren't we telegraphing a bit if we don't allow them to add to "classic" cm? meaning, -2's is saying "you can't"
15:32:17 * eglynn prefers the carrot to the stick in this context
15:32:25 <DinaBelova> _nadya_, nope, but if just words won't work and we could not encourage people to move their code to gnocchi
15:32:28 <DinaBelova> what else can we do?
15:32:49 <DinaBelova> merge to ceilo?
15:33:03 <DinaBelova> that's also variant...
15:33:16 <_nadya_> eglynn: yep, agreed that we can't... but we don't wan't to have a lot of "unneeded helpers" in the code
15:33:18 <DinaBelova> but that'll be kind of useless efforts for contributors..
15:33:32 <eglynn> my hope would be that if we can explain the future benefits of gnocchi, that reasonable contributors will want to be a part of that
15:33:53 <llu-laptop> eglynn: agreed
15:34:00 <idegtiarov> eglynn: +1
15:34:15 <_nadya_> DinaBelova: yep, I just raising the question :) unfortunately I don't know how it should be resolved in opensource
15:34:24 <DinaBelova> _nadya_, yeah..
15:34:51 <eglynn> k, shall we move on?
15:35:11 <_nadya_> maybe we may add smth like "our wishes to contributors" on wiki :)
15:35:39 <idegtiarov> _nadya_:  good point
15:35:40 <eglynn> _nadya_: yeah, it would be good to have to "loose" policy/preferences written down that we could point to
15:35:43 <gordc> _nadya_: no one reads the wiki... (see most blog posts about ceilometer)
15:35:55 <_nadya_> eglynn: yep, sorry for interrupting, go on
15:35:59 <nealph> _nadya_: I think a general explanation of architectural direction would do the same....plus get more eyes.
15:36:03 <cdent> gordc: is that because there's nothing real there?
15:36:04 <eglynn> #topic "Final call on dates/location for mid-cycle"
15:36:16 <eglynn> #link https://etherpad.openstack.org/p/CeilometerKiloMidCycle
15:36:26 <gordc> cdent: but my fancy architecture diagram?!?!? lol
15:36:36 <cdent> it is pretty nice
15:36:41 <gordc> lol
15:36:49 <fabiog> eglynn: it seems that we are going for the week of 20 in Gallway
15:36:50 <cdent> only because of the rotation though ;)
15:37:03 <gordc> cdent: true true. :)
15:37:12 <eglynn> fabiog: yep, are you good on the organizational side of that?
15:37:32 <fabiog> eglynn: yes, if we confirm I can go ahead and start
15:38:00 <eglynn> fabiog: well it seems we have quorum in terms of folks signed up
15:38:03 <fabiog> DinaBelova: any news on your attendance (company approval)?
15:38:14 <DinaBelova> fabiog, not yet... to early...
15:38:16 <nealph> fabiog: eglynn:  but is there a list of compelling topics to cover?
15:38:27 <DinaBelova> too*
15:38:36 <DinaBelova> fabiog, sorry
15:38:38 <eglynn> nealph: my focus will be on gnocchi migration
15:38:46 <jd__> I'm a bit worried about the low attendance for that mid cycle
15:39:01 <jd__> not sure I've many things to say face to face to DinaBelova, eglynn and ildikov
15:39:05 <eglynn> jd__: 6 below quorum do you think?
15:39:10 <jd__> not that I don't miss you guys
15:39:15 <eglynn> jd__: LOL :)
15:39:23 <cdent> I don't yet know if I'd be able to go, but would like to
15:39:29 <fabiog> eglynn: how many we were in Paris?
15:39:46 <jd__> but we can also spend some afternoons in a Hangout and that might be equally OK to me I think
15:39:52 <fabiog> cdent: aren't you kind of local?
15:39:59 <eglynn> fabiog: hmmm, same ballpark I think, maybe one or two more?
15:40:02 <_nadya_> jd__: you may write the code, not only talking :)
15:40:03 * jd__ just thinking out loud
15:40:15 <jd__> _nadya_: right
15:40:36 <gordc> eglynn: is this being combined with other projects like last time?
15:40:42 <eglynn> jd__: yep, I would like to mix up the format and not spend all the time on round table discussion as before
15:40:46 <jd__> could we move it south at least? Caribbeans anyone?
15:40:58 <eglynn> gordc: nope, just ceilo IIUC
15:41:20 <eglynn> fabiog: (unless the other HP folks in Galway want to pile on?)
15:41:22 <fabiog> eglynn: I think some discussion around "access" scope would be good
15:41:38 <eglynn> fabiog: "access" scope?
15:41:42 <fabiog> eglynn: I think it will be mainly Ceilo at this stage
15:41:43 <cdent> fabiog: yes, but budgeting is rather confused
15:42:00 <eglynn> fabiog: do you mean travel options to Galway?
15:42:17 <fabiog> eglynn: yes to decide what grants to give based on Domain or Project
15:42:30 <fabiog> eglynn: no is a topic for the meeting ..: -)
15:42:35 <eglynn> eglynn: a-ha, k, got it
15:43:10 <eglynn> right let's pullk together to quick etherpad with topic suggestions
15:43:12 <fabiog> Ok. So do we have consensus on the meeting?
15:43:42 <fabiog> or any objection?
15:43:55 <eglynn> fabiog: ... as agreed as we'll ever be I think
15:44:30 <fabiog> eglynn: well, I will consider the meeting to happen and start planning
15:44:34 <gordc> same as last meetup for me... will join via hangouts unless kickstarter comes through again.
15:44:51 <eglynn> gordc: drats! :(
15:45:03 <eglynn> I guess anyone who's interested in attending, but unsure of budget ... please chase approvals if you can so that we've got a solid headcount
15:45:23 <fabiog> I think the unsure are DinaBelova and cdent
15:45:30 <fabiog> and gordc at this stage
15:45:31 <eglynn> yeap
15:45:44 <cdent> I'll sail.
15:45:50 <cdent> (if I had a sailboat)
15:45:52 <eglynn> I'll cycle :)
15:45:56 <DinaBelova> fabiog, basically after the summit costs companies will send people only if really important quesiotns will be discussed, etc.... that's why budget approval stuff might work only if it'll be kind of specific agenda, etc.
15:46:07 <DinaBelova> cdent, what I should do????
15:46:15 * jd__ waits for the agenda too
15:46:17 <eglynn> DinaBelova: good point, let's work on firming up the agenda this week
15:46:30 <DinaBelova> eglynn, yeah, sorry to say this
15:46:48 <DinaBelova> I can't go to my manager and say "send me somewhere" :)
15:46:56 <DinaBelova> at least not so easy :D
15:47:06 <eglynn> DinaBelova: yep, I understand
15:47:08 <gordc> DinaBelova: we've seen the news... you guys are rich now :P
15:47:25 <DinaBelova> gordc, I did not see this money :D
15:47:47 <cdent> ha!
15:47:49 <_nadya_> gordc: hehe, yep
15:48:07 <jd__> next time I want this office http://www8.hp.com/lamerica_nsc_carib/en/contact-hp/office-locations.html
15:48:22 <DinaBelova> jd__, hehe
15:48:26 <DinaBelova> not bad
15:48:30 <fabiog> jd__: I offered Sunnyvale in CA
15:48:41 <jd__> fabiog: yeah I voted for that one :p
15:48:42 <fabiog> but you guys all signed up for Gallway ... :-)
15:48:43 <DinaBelova> ocean is cold there :)
15:49:02 <eglynn> could folks start throwing topic suggestions into this etherpad https://etherpad.openstack.org/p/galway-jan-2014-ceilometer-sprint
15:49:54 <eglynn> sorry folks, I'm gonna have to run shortly
15:50:03 <eglynn> I'm late for a company event starting at 1600UTC
15:50:11 <eglynn> OK if we close a few mins early today?
15:50:16 <nealph> jd__: I've been fighting for a meeting at the HP Sydney office....
15:50:16 <jd__> yep boss
15:50:22 <eglynn> jd__: thank you sir!
15:50:40 <eglynn> thanks for you time folks, 'fraid I'm gonna have to cut and run!
15:50:46 <DinaBelova> I guess all important moment are discussed :)
15:50:46 * cdent has 10 last minute things to keep eglynn
15:50:48 <DinaBelova> eglynn, thanks!
15:50:56 <eglynn> cdent: LOL :)
15:50:56 <fabiog> ok, thanks
15:51:12 <_nadya_> thank, guys!
15:51:14 <eglynn> #endmeeting ceilometer