12:00:10 <asalkeld> #startmeeting heat
12:00:10 <openstack> Meeting started Wed Jan  7 12:00:10 2015 UTC and is due to finish in 60 minutes.  The chair is asalkeld. Information about MeetBot at http://wiki.debian.org/MeetBot.
12:00:12 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
12:00:14 <openstack> The meeting name has been set to 'heat'
12:00:32 <asalkeld> #topic rollcall
12:00:52 <tspatzier> hi all
12:00:58 <ryansb> \o
12:01:19 <asalkeld> might be quiet
12:01:53 <zaneb> hello peeps
12:01:59 <ryansb> might make unanimous votes easy ;)
12:01:59 <asalkeld> hi zane
12:02:06 <asalkeld> har
12:02:13 * zaneb finally makes it to one of these 7am meetings
12:02:41 <asalkeld> #link https://wiki.openstack.org/wiki/Meetings/HeatAgenda
12:02:54 <asalkeld> let's wait a little
12:03:06 <asalkeld> see if any more folks pitch
12:03:18 <asalkeld> mean time thing of agenda
12:04:04 <asalkeld> #topic new agenda items
12:05:35 <asalkeld> hi ckmvishnu  unmeshg
12:05:41 <asalkeld> and ananta
12:05:49 <ananta> Hi asalkeld
12:05:55 <ckmvishnu> Hi asalkeld
12:05:57 <asalkeld> well lets progress
12:06:02 <unmeshg> hi
12:06:06 <zaneb> hey guys, happy new year :)
12:06:27 <asalkeld> #topic kilo-2
12:06:34 <asalkeld> #link https://launchpad.net/heat/+milestone/kilo-2
12:06:57 <asalkeld> so ttx says we should have our bp sorted by the end of the week
12:07:01 <asalkeld> (for k2)
12:07:28 <asalkeld> so let's review and approve what looks like it can make it
12:07:55 <asalkeld> shardy suggested doing some priority setting :-O
12:08:08 <asalkeld> speak of the devil :)
12:08:17 <shardy> sorry, got held up..
12:08:19 <asalkeld> shardy suggested doing some priority setting :-O
12:08:41 <shardy> Yea, seems like we have a lot of stuff in-flight and/or proposed for k2
12:09:01 <asalkeld> well i set k2 to things with patches up
12:09:09 <asalkeld> (bugs in progress)
12:09:18 <asalkeld> can change them later
12:09:29 <asalkeld> but the bp's are more important
12:09:36 <asalkeld> to set priorites
12:10:21 <asalkeld> https://blueprints.launchpad.net/heat/+spec/barbican-as-secret-backend
12:10:28 <asalkeld> not sure where that one is
12:10:37 <asalkeld> therve ...
12:11:13 <shardy> he's not here, I guess we can ping him in #heat after
12:11:19 <asalkeld> ok
12:11:39 <asalkeld> there is also the stack-tags bp (not approved)
12:11:48 <asalkeld> and the external resource thing
12:12:09 <asalkeld> but i can focus on the decouple-nested
12:12:33 <shardy> asalkeld: can you clarify, do we have to fix the adopt code to safely enable the external resource (with multi-step adopt..) feature?
12:13:07 <shardy> Originally it was an immutable external reference, but I see now you can do a series of updates to, effectively, adopt things
12:13:08 <asalkeld> shardy: how do you think it needs fixing?
12:13:33 <shardy> asalkeld: the various adopt bugs related to validation
12:13:57 <asalkeld> shardy: we do a check() at the time we take the resource
12:14:15 <shardy> Aha, and if that fails, we fail?
12:14:17 <asalkeld> to me that will say that it exists and the user can access it
12:14:24 <asalkeld> yeah
12:14:30 <shardy> Ok, sounds good, thanks
12:14:48 <asalkeld> tho' dependant on check() been implementde
12:14:57 <asalkeld> hi therve
12:15:09 <therve> Hi, sorry for being late
12:15:16 <asalkeld> therve: where is your barbican bp?
12:15:23 <asalkeld> good for k2 or k3
12:15:35 <asalkeld> i have set it to k2, but can change it
12:15:47 <asalkeld> https://blueprints.launchpad.net/heat/+spec/barbican-as-secret-backend
12:16:12 <therve> I have a spike somewhere, but not progressing much
12:16:30 <therve> k3 sounds more reasonable
12:16:34 <asalkeld> ok
12:16:56 <asalkeld> changed
12:17:19 <asalkeld> shardy: any other things you wanted to cover?
12:17:58 <asalkeld> it would be great to have some convergence bp there
12:18:00 <shardy> asalkeld: No, looks good, from the review and spec backlog, I thought there would be more BPs to cover..
12:18:12 <shardy> it's going to be another giant M3 isn't it?
12:18:13 <asalkeld> zaneb: ananta ^
12:18:30 <asalkeld> any thing we can do now to help out
12:18:37 <ananta> hi zane
12:18:54 <zaneb> asalkeld: not right now, but I hope soon
12:19:10 <zaneb> actually there is one thing
12:19:28 <ananta> we came after the vacation and I am looking for discussion
12:20:00 <asalkeld> ananta: sure
12:20:04 <zaneb> add a field to the Stack table to specify ld-style vs. new style stacks and a config option for creating new ones in the new style
12:20:17 <zaneb> (won't do anything yet, obvs)
12:20:22 <zaneb> but that is step 0
12:20:29 <asalkeld> ld-style?
12:20:35 <zaneb> old-style
12:20:37 <asalkeld> old?
12:20:38 <zaneb> typo, sorry
12:20:59 <shardy> zaneb: are we yet in a position to start raising BPs for the various specs and getting folks implementing them?
12:21:07 <shardy> s/specs/steps
12:21:12 <zaneb> not yet
12:21:16 <shardy> but specs too, I guess ;)
12:21:51 <zaneb> I think we have a separate agenda item to go over this?
12:21:58 <asalkeld> ok, well the next 2 agenda items might help
12:22:03 <shardy> zaneb: Ok, I guess I'm wondering how realistic landing this for Kilo is, if we're nearly at M2 and the tasks aren't specified yet
12:22:08 <asalkeld> (yeah, lets move on)
12:22:24 <asalkeld> #topic midcycle meetup
12:22:25 <zaneb> shardy: yes, it's a worry
12:22:28 <shardy> (no criticism, I know it's a hard problem)
12:22:37 <asalkeld> Planing the midcycle meetup online (brought forward from last time)
12:22:46 <asalkeld> #link https://etherpad.openstack.org/p/kilo-heat-midcycle-meetup
12:23:11 <asalkeld> maybe we can get this moved on in the meetup
12:23:28 <asalkeld> any ideas on how to do an online meetup?
12:24:03 <asalkeld> come up with some topics, and choose times and software
12:24:27 <asalkeld> tumble weeds ....
12:24:49 <shardy> asalkeld: agree, get an agenda with topics, then figure out tz's so the key folks can attend
12:25:07 <tspatzier> I guess agenda and some time slots for topic make sense, so people know when to dial in
12:25:15 <asalkeld> anyone keen to plan this?
12:25:24 <asalkeld> (or i can)
12:25:36 <tspatzier> ML for collecting topic?
12:25:42 <asalkeld> ok, lets start with topics
12:25:48 <shardy> +1
12:25:58 <asalkeld> convergence
12:26:04 <asalkeld> <eof>
12:26:08 <zaneb> lol
12:26:08 <shardy> lol
12:26:43 <asalkeld> now on the really hard part getting people to choose video software
12:27:02 <ryansb> I don't recall there being gripes about Skype
12:27:08 <ryansb> I do recall gripes about hangouts
12:27:09 <asalkeld> +1
12:27:19 <ryansb> soooo Skype?
12:27:28 <ananta> convergence has to be broken into many  to have any effective discussion
12:27:35 <ananta> Hangout
12:27:52 <asalkeld> ananta: i was being a bit tounge in cheek there
12:28:05 * zaneb was the chief griper about hangouts
12:28:15 <ananta> i am fine with skype though i had last time with zane
12:28:17 <asalkeld> sure we can come up with a better agenda
12:28:30 <zaneb> Skype is fine when it works... which is not as often as you would hope
12:28:34 <ananta> asalkeld: sure
12:28:45 <tspatzier> hangout burns all of my cores on the Mac -> so I am in favour of skype
12:28:46 <asalkeld> ananta: just edit that etherpad
12:28:50 <shardy> Ultimately we can fall back to a conference call if needed
12:28:59 <zaneb> tbh I'm not sure that video is really required here
12:29:09 <asalkeld> shardy: yeah just shut down the video
12:29:11 <zaneb> audio + maybe screen sharing seems more helpful
12:29:15 <shardy> We don't *have* to look at each other to talk :)
12:29:16 <tspatzier> zaneb: +1
12:29:22 <asalkeld> +1
12:29:31 <ananta> +1
12:29:35 <unmeshg> +1
12:30:17 <asalkeld> ananta: / zaneb you want to break the convergence topics up a bit
12:30:32 <asalkeld> and maybe max 1 hour a day?
12:30:42 <asalkeld> or longer?
12:30:58 <ananta> asalkeld: sure we can break
12:31:12 <ryansb> probably 2 1-hour or 45-min sessions is a sweet spot
12:31:24 <ananta> 1 hour looks good after  for smaller topics
12:31:34 <asalkeld> ryansb: ok, we can trial differient things
12:31:59 <asalkeld> ryansb: can you put some thoughts about format on the etherpad?
12:32:42 <ryansb> sure
12:32:48 <asalkeld> cool
12:33:02 <shardy> Ok, dumb question alert, can someone summarise the key advantage of the convergence model for scalability, vs just decoupling per-stack via RPC?
12:33:17 <asalkeld> shardy: big stacks
12:33:19 <shardy> Given all the effort going into this, I just wanted a reminder/refocus of why
12:33:39 <shardy> asalkeld: the main use-case for that IIRC was TripleO, which is now moving to lots of small stacks..
12:33:47 <asalkeld> lol
12:33:54 <shardy> hence my question ;)
12:34:06 <asalkeld> ok let's just ditch it :-O (joke)
12:34:18 <asalkeld> but good question
12:34:34 <asalkeld> lets focus on getting decouple-nested in
12:34:36 <asalkeld> and test
12:34:47 <asalkeld> while people work on convergence
12:34:48 <asalkeld> ?
12:34:52 <shardy> I'm playing devils advocate a bit here, I just want to focus on the use-case, e.g who actually needs this
12:35:09 <ananta> convergence is not just scalability
12:35:20 <asalkeld> well the continouse observer thing
12:35:24 <shardy> asalkeld: Yeah, I guess I'm wondering if decouple-nested is a sufficient scalability stop-gap for Kilo, if we fail to get Convergence in
12:35:25 <zaneb> I think it's recognised that phase 2 is the bigger win
12:35:57 <zaneb> but we need to do phase 1 to get to phase 2, and it does have some benefits of its own
12:36:17 <asalkeld> shardy: it might be a stop gap
12:36:32 <ananta> in phase 1 atleast we fix the issues we hit with backup stack etc and form the basis for phase 2
12:36:47 <ananta> and heat engine restart?
12:37:08 <shardy> ananta: aha, yes that's a good one, resiliency to heat-engine dying
12:37:14 <shardy> thanks
12:37:15 <zaneb> yeah, there are many potential latent issues with the backup stack that phase 1 fixes
12:37:16 <asalkeld> ananta: yeah lots of semi related things that need fixing are getting fixed
12:37:53 <zaneb> and this architecture will make the whole thing more robust and scalable
12:38:01 <ananta> asalkeld: yeah and sometimes i worry because of that we keep forgetting why we need convergence :)
12:38:13 <ananta> zaneb:  +1
12:38:26 <asalkeld> ananta: it's always good to ask basic questions
12:38:35 <ananta> asalkeld: i agree
12:38:41 <shardy> zaneb: I don't doubt that, I'm just trying to clarify the user-visible benefits, given the massive effort involved
12:38:42 <asalkeld> and remind ourselves why we are doing this
12:38:56 * shardy is good at asking basic questions ;)
12:39:01 <asalkeld> lol
12:39:19 <asalkeld> so would it help bringing the meetup earlier?
12:39:24 <zaneb> shardy: in many ways the benefits are to us in the future, more than to users
12:39:26 <asalkeld> so we can get on with this?
12:39:39 <zaneb> shardy: but the backup stack stuff is a real problem waiting to happen
12:39:43 <asalkeld> (like next week)
12:40:13 <shardy> zaneb: Ok, cool, so it can, if needed, become a more gradual rework, given the time-pressure re kilo?
12:40:21 <zaneb> asalkeld: this = meetup or convergence?
12:40:24 <asalkeld> given that's it's online, we can do this anytime
12:40:36 <asalkeld> (meetup about convergence)
12:40:39 <asalkeld> both
12:41:15 <asalkeld> well lets come up with the topics
12:41:22 <asalkeld> see what needs talking about
12:41:28 <zaneb> shardy: I think phase 1 needs to land completely and be enable-able in a specified release
12:41:55 <zaneb> I still hope that's kilo, at least for the brave
12:42:05 <shardy> zaneb: heh, Ok, thanks
12:42:49 <zaneb> but I am with you on the schedule concerns
12:43:13 <zaneb> I started looking at it again yesterday, and hope to make progress this week
12:43:16 <asalkeld> zaneb: we can all focus on review/testing
12:43:43 <asalkeld> to help out, who ever is driving the coding needs to ask for help
12:43:55 <asalkeld> lots of willing hands
12:44:24 <zaneb> yeah, we had lots of people volunteering to implement bps as well
12:44:43 <zaneb> but we need a plan first before we can set everyone loose
12:44:43 <asalkeld> #topic convergence
12:44:50 <asalkeld> (belated)
12:45:02 <asalkeld> zaneb: totally
12:45:25 <asalkeld> so zaneb ananta you want to tell where you guys are?
12:45:37 <ananta> zaneb: we should co-author the change we want to bring in the kilo
12:45:38 <asalkeld> is the basic design agreed on?
12:45:48 <zaneb> not yet
12:45:52 <ckmvishnu> nope
12:46:00 <zaneb> so I'll give an update of where I am at
12:46:25 <ananta> I was also looking at zane's poc and will give my update
12:47:20 <ananta> i do have a feeling that we are close to a design we want to but with some minor differences
12:47:33 <zaneb> before Xmas ckmvishnu and co raised some valid issues with my design and how it fails to address how to serialise operations on individual resources when multiple update passes are happening simultaneously
12:47:55 <zaneb> so I am looking at that this week, trying to come up with modifications that resolve that
12:48:13 <asalkeld> ok
12:48:28 <zaneb> unfortunately that's something that is explicitly out of scope for my framework to test
12:48:49 <zaneb> since it is event-driven rather than parallel
12:49:27 <asalkeld> is that going to be a problem to test?
12:49:44 <asalkeld> half re-write the framework
12:49:51 <asalkeld> to use real threads?
12:49:51 <zaneb> so I'll be flying a little bit blind, but hopefully I can write some code that gives the general idea and can be reviewed, even if it's not really tested in the simulator
12:51:01 <zaneb> asalkeld: don't think I will have time, but I am not against somebody contributing that
12:51:20 <zaneb> both sorts of tests are interesting
12:51:25 <asalkeld> ok
12:52:07 <asalkeld> i'll see if i get some time - maybe skype you about some ideas to save time
12:52:28 <zaneb> 'k
12:52:46 <asalkeld> zaneb: is your evening better (than now0
12:53:04 <zaneb> yes, lots of meetings in the mornings :(
12:53:22 <asalkeld> ga, this is the problem we are going to have with the online meetup
12:53:36 <asalkeld> i hope it's not me always in the moring
12:53:44 <zaneb> hmm, actually no, today is the opposite
12:53:50 <zaneb> now might be better
12:53:55 * asalkeld is not a morning person
12:54:11 * zaneb neither
12:54:19 <asalkeld> zaneb: after meeting?
12:54:23 <zaneb> sure
12:54:27 <asalkeld> k
12:54:31 <asalkeld> 5mins
12:54:44 <asalkeld> anything anyone wants to chip in?
12:55:33 <ryansb> where do we want to work on the agenda? Same etherpad, or a new one?
12:55:41 <ryansb> (meetup agenda)
12:55:47 <asalkeld> ryansb: i think so
12:55:52 <asalkeld> all in one place
12:56:01 <ryansb> Sounds good to me
12:56:11 <asalkeld> zaneb: ananta ckmvishnu : do we need to tidy the convergence specs up and resubmit?
12:56:19 <asalkeld> or is that too early
12:56:39 <zaneb> probably a bit early
12:56:53 <asalkeld> ok
12:56:54 <ananta> not yet...only after the we have common understanding of design
12:57:02 <ckmvishnu> yeah, but needs to broken down
12:57:07 <zaneb> but yeah, we will likely need to at some point
12:57:22 <asalkeld> k, makes sense
12:57:57 <asalkeld> ending a bit early ...
12:58:04 <asalkeld> #endmeeting