20:03:27 <mordred> #startmeeting tc
20:03:28 <openstack> Meeting started Tue Feb 26 20:03:27 2013 UTC.  The chair is mordred. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:03:29 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:03:31 <openstack> The meeting name has been set to 'tc'
20:03:33 <bcwaldon> vishy is on his way
20:03:42 <vishy> hi
20:03:49 <mordred> #link https://wiki.openstack.org/wiki/Governance/TechnicalCommittee
20:03:56 <mordred> agenda is above
20:04:13 <mordred> #topic End-of-cycle graduation review: Ceilometer - Scope complementarity
20:04:14 <jaypipes> o/
20:04:40 <mordred> next up is talking about Ceilometer scope
20:04:58 <mordred> nijaba: you wanna kick that off?
20:05:03 <nijaba> sure
20:05:16 <nijaba> so, who has read the wiki page we prepared?
20:05:24 <heckj> read it last week - was it changed?
20:05:25 <nijaba> or do you want me to cover some summary?
20:05:37 <nijaba> heckj: yes, quite a bit, lots of schemas
20:05:38 <eglynn> heckj: yep, it's been updated today
20:05:43 <mordred> link?
20:05:54 <dhellmann> #link https://wiki.openstack.org/wiki/Ceilometer/Graduation
20:05:57 <mordred> thanks
20:06:00 <heckj> thanks
20:06:00 <nijaba> #link https://wiki.openstack.org/wiki/Ceilometer/Graduation
20:06:42 <mordred> sepcifically, this section is to discuss "desirability of integrate Ceilometer in the common OpenStack Havana release" (to steal ttx's words)
20:06:43 <heckj> the biggest question in my mind is where do you currently draw the line on "Not Ceilometer"?
20:06:44 <mordred> Is the project complementary in scope, or overlapping with others ?
20:06:56 <mordred> which makes heckj's question a good one :)
20:07:03 <heckj> :-)
20:07:12 <nijaba> I think this is covered in https://wiki.openstack.org/wiki/Ceilometer/Graduation#Detailed_scoping_of_future_monitoring_support
20:07:50 <heckj> so not only data collection and aggregattion, but some evaluation and post-process means as well?
20:07:52 <nijaba> so in short, we collect metrics, we do metering, and we do the part of alerting that makes sense to hae close toe the data collection
20:08:25 <notmyname> would you consider swift's statsd integration (for hundreds of metrics) to be overlapping with seilometer?
20:08:38 <notmyname> s/seilometer/ceilometer
20:08:51 <nijaba> notmyname: no, it should be a potential source
20:09:26 <eglynn> notmyname: ... and ceilometer would be the more general mechanism, covering openstack services other than swift
20:09:43 <notmyname> eglynn: well, sure, but so it's statsd ;-)
20:10:18 <dhellmann> notmyname: in your analogy we're not statsd, we're the tool that sends data to statsd
20:11:01 <swifterdarrell> dhellmann: I don't think it was an analogy--Swift emits StatsD-formatted UDP packets (if configured)
20:11:15 <eglynn> notmyname: note the dotted line in this schema also ... https://wiki.openstack.org/w/images/5/50/Ceilometer-multi-publish2.jpg
20:11:21 <notmyname> there's no analogy. perhaps bad phrasing on my part. do you envision that the metrics that swift currently has using statsd will be replaced by ceilometer metrics
20:11:28 <swifterdarrell> dhellmann: I think notmyname's point was that if ceilometer consumes that, it would have to at least implement the receiving half of the StatsD protocol?
20:11:33 <markmc> <notmyname> eglynn: well, sure, but so it's statsd ;-)
20:11:35 <sandywalsh> (imho, in the swift case we would be pulling aggregated data from statsd or logs ... otherwise the volume would be too great for CM to relay)
20:11:39 * markmc assumes s/it's/is/ there
20:11:48 <nijaba> notmyname: why would we replace?  We are just a conduit that people can chose to use
20:12:21 <dhellmann> right, I think it would be more useful to talk about what stats ceilometer could collect for other projects that aren't collecting them, to avoid having every project do their own thing
20:12:22 <eglynn> markmc: ok, that would make more sense
20:12:53 <notmyname> that's all I'm asking. my day-to-day involves a certain metrics-gathering system. does that need to change (that's a general question that can apply to all projects)
20:12:54 <heckj> nijaba: eglynn for the pieces that interact with Heat - the alarming and such, when (which release cycle) are you thinking you'll expand into that space? You talk about phases, but I'm not sure how they match to releases.
20:13:13 <eglynn> heckj: Havan cycle
20:13:18 <nijaba> heckj: havanah for sure
20:13:21 <eglynn> s/Havan/Havana/
20:13:24 * heckj nods
20:13:32 <mordred> word
20:13:46 <nijaba> mordred: word?
20:13:55 <dhellmann> notmyname: if you want all of your data going to statsd, then ceilometer will support that (when we write that publisher)
20:14:10 <dhellmann> so I think that's a "no"
20:14:11 <mordred> nijaba: old slang. nevermind. I was nodding
20:14:19 <nijaba> np
20:14:36 <eglynn> heckj: in https://wiki.openstack.org/w/images/8/8d/Ceilometer-monitoring-scope.jpeg ... blue = G, orange = H, yellow = Never
20:14:39 <heckj> (omg, it's "old" slang?)
20:14:47 <heckj> eglynn: ah - thank you
20:15:45 <gabrielhurley> heckj, mordred: nah, us youngsters still say "word"
20:16:03 <mordred> ok. so any more on that topic before we move on to general final Q&A ?
20:16:16 <annegentle> so by the H release you intend to have APIs that people can use? and for G release people can use this publisher?
20:16:38 <dhellmann> annegentle: we have a web api now
20:16:47 <mordred> #topic End-of-cycle graduation review: Ceilometer - Final Q&A
20:16:54 <eglynn> annegentle: we have a metering API in G, will have alarming APIs in H
20:17:02 <nijaba> annegentle: for G we have the multi publisher architecture complete, which then allows to add cloudwatch api in H
20:17:15 <annegentle> ok but the wiki is the only location for docs so far? -- how do people deploy?
20:17:29 <dhellmann> #link http://docs.openstack.org/developer/ceilometer/
20:17:39 <nijaba> annegentle: wiki is not the only pace for doc...
20:17:45 <nijaba> thanks dhellmann
20:17:58 <dhellmann> we have some (slightly out of date) installation instructions there covering devstack and manual installation
20:18:18 * jaypipes looking forward to seeing integration test suite (tempest) coverage of ceilometer
20:18:18 <eglynn> #link http://docs.openstack.org/developer/ceilometer/install.html#installing-and-running-the-development-version
20:18:21 <nijaba> there are also puppet and chef modules too
20:18:29 <dhellmann> jaypipes: +1
20:18:34 <annegentle> ok got those
20:18:34 <mordred> jaypipes: +1
20:19:04 <dhellmann> annegentle: the docs site holds our formal documentation. anything in the wiki is likely to be an in-process or older design doc
20:19:06 <annegentle> do you envision this integrating with the admin view on the Horizon dashboard?
20:19:22 <jaypipes> annegentle: yes, there is a note about horizon integration.
20:19:31 <nijaba> annegentle: no, just provide a sample plugin at some point
20:20:05 <nijaba> annegentle: to be fully usefull on a public cloud deployement, we would need to know about cost, which is out of scope
20:20:10 <jaypipes> nijaba: sorry, saw "plugin for horizon underway" in the wiki
20:20:25 <eglynn> annegentle: watch this space ... https://blueprints.launchpad.net/ceilometer/+spec/horizon-plugin
20:20:39 <nijaba> jaypipes: yes, but I consider it a sample, mainly usefull for private clouds
20:20:51 <jaypipes> nijaba: cool, good to know, thx!
20:20:53 <mordred> nijaba: but could your sample plugin be something that could be integrated with horizon directly?
20:20:54 <gabrielhurley> we'll work more on that in H
20:20:56 <jd__> still we may provide some default visualisation or something
20:21:10 <nijaba> mordred: I do not see why not, we ar open to dicussion
20:23:06 <mordred> ok. so, any last questions before we vote on this bad-boy?
20:23:23 <annegentle> sorry one more
20:23:40 <nijaba> annegentle: please ask away!
20:23:41 <annegentle> can you explain what "in terms of addressing the Heat requirement" means for elements 6 and 7 of your scope
20:23:43 <mordred> go for it! we've got all day
20:24:02 * annegentle is stealing time from the Book Sprint :)
20:24:04 <eglynn> annegentle: heat have a requirement for clouwatch -like functionality
20:24:05 <nijaba> annegentle: in other word, the neeed for cloudwatch to exist
20:24:23 <eglynn> annegentle: they currently have a rudimentary internal implementation
20:24:35 <eglynn> annegentle: but wish to replace with something provided by ceilometer
20:24:40 <mordred> by cloudwatch you mean heat wants to be able to take actions based on health of services right?
20:24:46 <mordred> like, autoscaling?
20:24:49 <eglynn> exactly
20:24:58 <eglynn> (it already does)
20:25:06 <mordred> (some of us don't follow amazon features and have to translate from time to time)
20:25:08 <annegentle> ok so this is natural re-factoring
20:25:17 <annegentle> mordred: for sure
20:25:26 <nijaba> annegentle: yes
20:25:26 <eglynn> but the internal Heat CW implementation is simple and not general purpose
20:25:53 <eglynn> (e.g. relies on a script run within instances to report metrics)
20:26:50 <eglynn> (we're on the same page with the Heat folks on this, so we'd view it as complementarity as opposed to overlap ...)
20:26:55 <annegentle> ok
20:27:54 <sdake_> eglynn agree
20:27:58 <notmyname> what about the scope question of not "what is in scope for ceilometer" but "what is in scope for openstack"? does this project fit in with the overall goals and mission of openstack?
20:28:20 <nijaba> notmyname: I believe it does
20:28:36 <mordred> I think it does - especially given the integration with other projects such as nova and heat for intelligent decision making
20:28:38 <markmc> I think we covered that in an earlier part of the review?
20:28:47 * markmc thinks it does, very much
20:28:51 <nijaba> notmyname: it provides the one transversal point to do data collection and therefore support the goals of the project
20:28:55 <markmc> what use is an unmetered cloud?
20:29:09 <mordred> markmc: in soviet russia, cloud meters you?
20:29:25 <markmc> mordred, I've never been, you tell me?
20:30:41 <mordred> second failed pop culture reference by me today...
20:30:51 <creiht> lol
20:30:52 <mordred> ok. questions seem to have died down...
20:30:52 <jaypipes> we voting?
20:30:56 <jgriffith> mordred: I got it
20:31:22 <mordred> #startvote Approve graduation of Ceilometer (to be integrated in common Havana release)? yes, no, abstain
20:31:23 <openstack> Begin voting on: Approve graduation of Ceilometer (to be integrated in common Havana release)? Valid vote options are yes, no, abstain.
20:31:24 <openstack> Vote using '#vote OPTION'. Only your last vote counts.
20:31:39 <russellb> #vote yes
20:31:40 <heckj> #vote yes
20:31:40 <mordred> #vote yes
20:31:42 <gabrielhurley> #vote yes
20:31:47 <jaypipes> #vote yes
20:31:49 <markmc> #vote yes
20:31:53 <bcwaldon> #vote yes
20:31:55 <jgriffith> #vote yes
20:31:56 <markmc> mordred, I got it, I'm just messing with you
20:32:03 * mordred punches markmc
20:32:12 <annegentle> #vote yes
20:32:17 <vishy> #vote yes
20:32:33 <danwent> #vote yes
20:33:22 <mordred> notmyname: ping?
20:33:43 <notmyname> I honestly don't know what to vote :-)
20:33:55 <jaypipes> notmyname: your conscience?
20:34:11 <notmyname> I have reasons for voting all of the options
20:34:35 <notmyname> I think they've done a fantastic job working with openstack. so for that +1. I don't think their design is what I would have chosen +0. I don't think a "unified plugin framework" is a good design -1
20:34:48 <notmyname> with apologies for the wording
20:35:00 <heckj> notmyname: if you d
20:35:08 <heckj> don't know, I suggest you abstain
20:35:10 <notmyname> #vote abstain
20:35:42 <mordred> that's everybody
20:35:44 <mordred> #endvote
20:35:45 <openstack> Voted on "Approve graduation of Ceilometer (to be integrated in common Havana release)?" Results are
20:35:46 <openstack> yes (11): markmc, bcwaldon, vishy, annegentle, heckj, jaypipes, russellb, jgriffith, mordred, gabrielhurley, danwent
20:35:47 <openstack> abstain (1): notmyname
20:35:55 <nijaba> \o/
20:35:56 <markmc> congrats nijaba, dhellmann, jd__, eglynn, asalkeld et al. :)
20:36:00 <mordred> congratulatoins to ceilometer
20:36:03 <nijaba> thanks a lot tc members!
20:36:08 <sandywalsh> thanks & congrats!
20:36:15 <russellb> thanks for all your hard work so far
20:36:22 <eglynn> w00t!
20:36:43 <jd__> thanks!
20:37:03 <mordred> which brings us to the next topic...
20:37:16 <mordred> #topic Update on elections organization
20:37:26 <mordred> at least, I think I know what this is about ...
20:37:40 <mordred> it's time for PTL and general TC elections for those whose seats are up
20:37:56 <mordred> since ttx is up for election, I'll be organizing the elections (god help us all)
20:38:08 <mordred> which means I'll be sending a warning email today
20:38:31 <mordred> we'll be including ptl elections for ceilometer for sure
20:38:50 <mordred> we JUST did a heat election ... but I think we need to do another one because now heat is official?
20:38:59 <nijaba> and heat as well, I am sure
20:39:15 <nijaba> mordred: ah
20:39:32 <mordred> anybody have any concerns or questions or thoughts or views or opinions or poems they'd like to share about TC elections?
20:39:44 * nijaba closes his mouth for good...
20:39:47 <sdake_> yes another heat election
20:39:58 <sdake_> ttx was clear on this point in earlier conversations
20:40:05 <markmc> mordred, we're all doing a fine job, we don't need no elections
20:40:10 * markmc declares marshal law
20:40:20 <russellb> ha
20:40:46 <markmc> mordred, what's the timetable?
20:40:51 <markmc> i.e. when will they be held?
20:41:06 <mordred> March 1-7: Nominations for PTL
20:41:15 <mordred> March 8-14: Vote for PTL
20:41:21 <mordred> March 15-21: Nominations for direct seats
20:41:27 <mordred> March 22-28: Vote for direct seats
20:41:49 <mordred> sdake_: awesome - because the first election was so close
20:42:12 <sdake_> i think most heat devs dont want to sit in meetings .. ;)
20:42:36 <mordred> wow. you have smart devs...
20:42:39 <russellb> you could stand
20:43:36 <mordred> ok. well, if there's nothing else on that topic ... that's all for the agenda
20:43:43 <mordred> anybody got anything else?
20:44:38 <markmc> thanks mordred
20:44:44 <mordred> thanks everybody!
20:44:46 <mordred> #endmeeting