20:00:50 <shardy> #startmeeting heat
20:00:51 <openstack> Meeting started Wed Apr 10 20:00:50 2013 UTC.  The chair is shardy. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:52 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:00:54 <openstack> The meeting name has been set to 'heat'
20:01:08 <shardy> #topic rollcall
20:01:12 <zaneb> o/
20:01:14 <shardy> shardy here
20:01:14 <asalkeld> o/
20:01:25 <jpeeler> hi!
20:01:49 <stevebaker> w00t
20:01:50 <sdake> hi
20:03:02 <shardy> #topic Review last week's actions
20:03:17 <shardy> So I don't see any in the minutes, and I didn't make last weeks meeting
20:03:35 <asalkeld> yea, I can't remember any
20:03:38 <shardy> Anyone got anything to report related to last week's discussions?
20:03:42 <sdake> the minutes are in a link off the agenda page but don't recall any actions
20:03:54 <shardy> ok, can move on then
20:04:07 <shardy> #topic Summit session review
20:04:29 <shardy> So I guess we've reviewed the sessions, but are we happy with the sessions, and who's doing what?
20:04:54 <asalkeld> yea, I need to do some work on mine
20:04:57 <llirkaz> Hello
20:05:13 <shardy> I'm planning to lead the credentials session, but anyone have anything they need help with for the sessions they proposed?
20:05:38 <stevebaker> we should have time to prepare for the Heat Update session in person
20:05:51 <shardy> stevebaker: sounds good
20:05:53 <zaneb> not sure if SpamapS wants me to lead the parallel resource creation session?
20:06:08 <zaneb> also, not sure what that session should involve
20:06:11 <shardy> For those who missed it earlier, we're planning an informal Heat get-together on the SUnday night
20:06:18 <zaneb> "once upon a time, I already wrote all the code."
20:06:28 <asalkeld> anyone has any thoughts on scaling heat look here: https://etherpad.openstack.org/heat-multiple-engines
20:06:42 <shardy> zaneb: Live demo? :D
20:06:51 <zaneb> lol
20:07:05 <asalkeld> we should really put stuff in the etherpads so others can contibute
20:07:07 <zaneb> that assumes I can actually still get Openstack to install and run on my laptop
20:07:14 <barefoot> where will the information get together be?
20:07:19 <barefoot> informal*
20:07:25 <asalkeld> https://wiki.openstack.org/wiki/Summit/Havana/Etherpads#Heat
20:07:48 <asalkeld> barefoot, don't know
20:07:51 <shardy> barefoot: to be decided
20:07:58 <barefoot> kk thanks
20:08:19 <asalkeld> we will meetup at the sunday summit registration
20:08:38 <asalkeld> I think that is 3pm
20:09:17 <stevebaker> ok
20:09:23 <asalkeld> (see where to go from there)
20:09:35 <shardy> Anything else summit related?
20:09:52 <zaneb> what is the deal with the OpenShift event on Sunday?
20:10:17 <zaneb> will that cause scheduling issues?
20:10:24 <stevebaker> I've signed up for it but I may just drift in an out
20:10:28 <asalkeld> not sure, but if you can go that would good
20:10:40 <hanney> o/
20:10:42 <zaneb> stevebaker: link?
20:11:07 <stevebaker> http://openshiftorigincommunityday.eventbrite.com/
20:11:11 <stevebaker> 3 tickets left!
20:11:32 <zaneb> btw, would any of the RAX folks who might be here like to introduce themselves?
20:12:18 <sandywalsh> o/
20:12:37 <asalkeld> hi sandy
20:12:53 <oubiwann> zaneb: hey man, I'm Duncan McGreggor
20:13:14 <zaneb> so, there's two teams involved here from RAX, right?
20:13:14 <barefoot> zaneb: hey, I'm Shawn Ashlee
20:13:31 <shardy> welcome guys :)
20:13:32 <oubiwann> some other RAX folks are at lunch, but they should be here soon
20:13:35 <oubiwann> shardy: thanks!
20:13:41 <zaneb> indeed, welcome
20:13:46 <barefoot> thanks
20:13:52 <oubiwann> wirehead_ <-- Racker too
20:13:56 <wirehead_> hello
20:14:05 <zaneb> would y'all like to give us a rundown on which areas you're working in?
20:14:07 <asalkeld> are you guys attending other summit dev sessions?
20:14:17 <oubiwann> asalkeld: yes
20:14:29 <asalkeld> or could we do some more meetups on tue/wed
20:14:51 <asalkeld> one of those days seemed a bit "light"
20:14:52 <oubiwann> still working out the logistics on who exactly will be there, but we'll be in most of the Heat discussions at the summit
20:15:01 <oubiwann> asalkeld: +1!
20:15:37 <asalkeld> well maybe we can grab an unconf. room
20:17:05 <fsargent> I'm Felix Sargent, Hi! <-- Racker
20:17:35 <asalkeld> hi
20:17:55 <oubiwann> zaneb: oh, your suggestion -- yes, we can talk about the autoscale portion
20:18:06 <oubiwann> not sure if the other team is here right now or not?
20:18:36 <zaneb> ok cool, thanks for that guys
20:18:53 <oubiwann> we're super curious to see where folks want to take autoscaling
20:18:57 <asalkeld> you guys interested in autoscale must also go to the thursday ceilometer talks
20:19:07 <fsargent> I'll be there.
20:19:09 <asalkeld> planing alarming
20:19:22 <oubiwann> whether that's to split some portion out for a user-consumable API with a standalone REST API service or whether everything happens in Heat
20:19:59 <oubiwann> whichever one ends up being the preferred approach, we want to start hacking on contributions that make "that" possible
20:20:07 <oubiwann> (whatever "that" ends up entailing)
20:20:14 <asalkeld> oubiwann, we could have a "user-consumable API with a standalone REST API service" in heat
20:20:21 <oubiwann> in other words, no agenda, just devs ready and waiting to jump in!
20:20:25 <asalkeld> that would also be easy
20:20:31 <asalkeld> cool
20:20:37 <oubiwann> asalkeld: that sounds like a great start, then!
20:21:06 <shardy> Yeah, agree, easiest first step is to leave the AS core logic in heat-engine and add a new API if that's what you need
20:21:24 <asalkeld> so one problem is it is difficult to depend on a non-core project
20:21:30 <zaneb> +1
20:21:38 <SpamapS> o/
20:21:43 <SpamapS> sorry.. got stuck in traffic
20:21:56 <oubiwann> SpamapS: wassup!
20:23:00 <asalkeld> wow we are going to have a pile of devs from now on
20:23:06 <oubiwann> :-)
20:23:27 <popstar> :)
20:24:11 <shardy> Do we want to continue this AS discussion next week, and firm up plans potentially for some additional informal/meetup time during the week?
20:24:29 <oubiwann> asalkeld: sorry, working on parsing intent of that comment… maybe this?: meaning that keeping it in core makes things a lot easier, and if there ever was a need to split things, it would likely be very obvious and Heat would generate a new project (that presumably would also be accepted as core)?
20:24:36 <SpamapS> re Sunday night, thats a nogo for me, I don't arrive until nearly midnight
20:24:53 <asalkeld> :(
20:25:09 <asalkeld> oubiwann, well we would have to have a copy in Heat
20:25:17 <oubiwann> ah, got it
20:25:18 <oubiwann> yeah
20:25:22 <oubiwann> awkward
20:25:22 <asalkeld> so if we really want to split
20:25:32 <asalkeld> the code will need to be sync'd
20:25:40 <SpamapS> regarding autoscaling, I'd suggest that people do some brain dumps to mailing list today/tomorrow so we can all do our homework over the weekend and come prepared with solid questions.
20:25:41 <asalkeld> for a while anyways
20:25:42 <lifeless> oubiwann: o/
20:25:43 * oubiwann nods
20:25:55 <oubiwann> SpamapS: +1
20:26:00 <oubiwann> lifeless: !
20:26:03 <zaneb> it's also a lot easier to incubate a new project inside an existing core project than to try and get it accepted from scratch
20:26:18 <stevebaker> nova->glance
20:26:27 <SpamapS> I'd also like people to think on what Heat really is and why it should encompass autoscaling.
20:26:34 <asalkeld> I am not sure of the value in seperating - personally, there is not much logic in autoscaling
20:26:38 <SpamapS> TO me, it is "the thing which calls the other APIs"
20:26:50 <asalkeld> if the monitoring can do alarms
20:26:56 <SpamapS> Which is why AS in heat makes a lot of sense.
20:27:19 <asalkeld> basically what is left is cooldown
20:27:32 <zaneb> it makes sense that people who are not using templates still want to use autoscaling though
20:27:35 <rax_randallb> I can envision wanting to use AS outside of heat, though.
20:28:04 <asalkeld> you guys are getting confused with project view and deployment view
20:28:32 <SpamapS> zaneb: right, which is why I suggest Heat not be "the thing that parses templates into API"
20:28:38 <asalkeld> what project some code lives in does not determine how you deploy
20:28:40 <rax_randallb> asalkeld: right. sorry, kinda new to the party here.
20:29:11 <asalkeld> as long as we have a good seperate rest api for people to use
20:29:20 <asalkeld> (still open to tho')
20:29:26 <asalkeld> (still open to it tho')
20:29:26 <zaneb> that's true, as long as it really can be deployed independently
20:29:34 <SpamapS> The template API is just the declarative entry point. Imperative calls "make me a stack. make an instance in the stack. make a scaling group in the stack." should be added to appease the "no templates for autoscaling" use case.
20:29:35 <asalkeld> yip
20:29:56 <oubiwann> nice
20:30:21 <zaneb> our friend from the ML will not be happy
20:30:29 * oubiwann chortles
20:30:31 <oubiwann> but!
20:30:45 <SpamapS> zaneb: I hope that he will, once we are all in a room and chatting about it.
20:30:47 <oubiwann> it sounded like he just needed to have some good conversations with folks…
20:30:53 <oubiwann> SpamapS: bingo
20:30:54 <zaneb> hopefully
20:31:15 <zaneb> in any event, even if we want to make it separate, giving it a separate api is the first step
20:31:21 <SpamapS> This is why taking out the language from the wiki page "Heat is [basically CloudFormation]..." is so important. :)
20:31:27 <zaneb> so I think we can all support doing that
20:31:38 <SpamapS> "One aspect of Heat is compatibility with CloudFormation"
20:32:18 <asalkeld> well it has always been the project that uses everything
20:32:34 <asalkeld> so we end up having to impl. missing stuff
20:32:39 <asalkeld> like lb
20:32:47 <asalkeld> autoscale/cw
20:33:02 <asalkeld> and it makes sense to split some out
20:33:14 <asalkeld> but maybe not everything
20:33:15 <oubiwann> how far along is Quantum LBaaS, does anyone know?
20:33:16 <SpamapS> asalkeld: right, that pragmatism can continue for sure.. but autoscaling feels like it will happily live just fine in heat.
20:33:34 <SpamapS> oubiwann: PoC from what I understand, but rapidly evolving.
20:33:39 <SpamapS> as in, they have proved the concept
20:33:57 <oubiwann> rhm
20:33:59 <oubiwann> okay
20:33:59 <asalkeld> oubiwann, honestly not sure, but probably more evolved than ours
20:34:02 <zaneb> SpamapS: we couldn't split it out altogether without signoff from the TC anyway
20:34:20 <asalkeld> zaneb, I think they have an api
20:34:26 <asalkeld> (stable)
20:34:37 <asalkeld> that multiple projects implement
20:34:46 <SpamapS> we have a blueprint to make use of that?
20:35:06 <asalkeld> yea there is one
20:35:14 * zaneb has lost track of what we're talking about
20:35:21 <oubiwann> I think we're still on LB
20:35:21 <asalkeld> lbaas
20:35:28 <shardy> asalkeld: there is?
20:35:28 <zaneb> thanks ;)
20:35:35 <asalkeld> looking
20:36:17 <asalkeld> maybe not
20:36:20 <SpamapS> we're talking about how Heat has implemented things internally that exist in other incubated or even core projects (reddwarf and quantum lbaas)
20:37:05 <shardy> SpamapS: Yup, we should raise BPs for cases where there are viable projects we could use instead of our nested implementations
20:37:14 <asalkeld> so we need to use lbaas and ceilometer for alarming (when that is ready)
20:37:17 <zaneb> yeah, and we should scrap those internal implementations as soon as there's an integrated project for it
20:37:39 <oubiwann> I love you guys
20:37:40 <asalkeld> and there is moniker too
20:37:51 <asalkeld> yikes settle down
20:37:54 <asalkeld> ;)
20:38:06 <zaneb> rofl
20:38:12 * oubiwann puts the group hug on hold
20:38:18 <shardy> Are we veering OT here - there's blueprint review on the agenda, shall we move on?
20:38:20 <SpamapS> zaneb: I think those should live-on as deprecated for a while. There will be times that compatiblity is not 100% and people need a release-cycle to fix things.
20:38:40 <zaneb> SpamapS: contrib/plugins?
20:38:46 <SpamapS> zaneb: yeah I like that
20:39:06 <zaneb> so glad we implemented that whole plugin thing
20:39:14 <barefoot> haha
20:39:14 <asalkeld> +1
20:39:15 <zaneb> solves so many arguments :)
20:39:57 <asalkeld> there seems to be a new aas every week
20:40:23 <lifeless> asalkeld: I was saying to a colleague the other day
20:40:24 <shardy> #topic Blueprint review for Havana
20:40:29 <lifeless> asalkeld: we need aasaas
20:40:43 <asalkeld> other thing when you see one is to encourage them to use heat for org
20:40:52 <asalkeld> haha
20:41:07 <shardy> Looks like we got to this one:
20:41:13 <shardy> #link https://blueprints.launchpad.net/heat/+spec/auth-token-only
20:41:29 <shardy> stevebaker: you're working on this now right?
20:41:54 <stevebaker> Yep, just need some reviews on what is there
20:42:13 * SpamapS will hit the review queue post meeting
20:42:17 * asalkeld not an auth guru
20:42:28 <shardy> stevebaker: Ok, cool, I'll bump the status then
20:42:29 <stevebaker> It looks like there is potential to pass the service catalog from api->engine...
20:42:56 <asalkeld> is there anyone interested in multi-cloud?
20:42:58 <stevebaker> so that can change anything that fetches auth_url from the context
20:43:15 <stevebaker> or an endpoint from heat-engine.conf
20:43:16 <asalkeld> like regersting multiple openstack clouds
20:43:31 <stevebaker> I had a play with multicloud against HP
20:43:50 <asalkeld> stevebaker, you could think about how we could support that
20:44:02 <stevebaker> I think we can get there, but we'll either need to patch or subclass keystoneclient auth_token
20:44:14 <shardy> #link https://blueprints.launchpad.net/heat/+spec/bash-environment-function
20:44:16 <stevebaker> yes, but would be easier to explain with a beer in one hand
20:44:32 <asalkeld> :)
20:44:33 <shardy> zaneb: you happy to take this one?
20:44:36 <SpamapS> asalkeld: openstack-infra is interested in multi-cloud, but dunno if they can get to it this cycle.
20:44:45 <zaneb> sure
20:44:48 <asalkeld> sure
20:44:55 <stevebaker> zaneb: or choose a template format?
20:44:58 <zaneb> although, there was a competing proposal too I thought
20:45:14 <shardy> #link https://blueprints.launchpad.net/heat/+spec/cfn-create-aws-symlinks
20:45:19 <stevebaker> https://blueprints.launchpad.net/heat/+spec/template-string-function
20:45:25 <asalkeld> isn't that done?
20:45:31 <shardy> So I think this one is being handled via heat-cfntools?
20:45:46 <asalkeld> aws-symlinks
20:45:47 <zaneb> stevebaker: yeah, that one. I like that one better
20:45:48 <asalkeld> ya
20:45:49 <stevebaker> that should probably be transferred to heat-cfntools
20:46:14 <shardy> stevebaker: k, I thought this was already done, we can move or obsolete it
20:46:33 <shardy> https://blueprints.launchpad.net/heat/+spec/cloudwatch-update-stack
20:46:45 <shardy> I'm doing this one, should be simple
20:46:46 <asalkeld> that is going to be never ending
20:47:09 <asalkeld> o, that one
20:47:15 <stevebaker> shardy: I've obsoleted it
20:47:24 <asalkeld> didn't see the cloudwatch bit
20:47:41 * shardy has been forgetting \# link
20:47:44 <shardy> oops
20:48:01 <shardy> #link https://blueprints.launchpad.net/heat/+spec/configurable-loadbalancer
20:48:19 <SpamapS> heh
20:48:29 <SpamapS> that one is obsoleted by lbaas
20:48:38 <asalkeld> I got quite far with that one
20:48:40 <zaneb> lets implement an LBaaS resource type instead
20:48:53 <asalkeld> not quite, it raised interesting issues
20:48:59 <asalkeld> so it was good
20:49:01 <shardy> ok, sounds good, we can raise a LBaaS BP and obsolete this one
20:49:01 <stevebaker> native *and* aws please?
20:49:23 <zaneb> for the nested stack resources we have...
20:49:30 <asalkeld> stevebaker, goes native
20:49:30 <shardy> stevebaker: agree, same goes for lots of other resource too tho ;)
20:49:33 <zaneb> it might be good to load the template from a file
20:49:38 <zaneb> instead of putting it inline
20:49:56 <SpamapS> 11 minutes... perhaps we should remain at periscope depth for the remainder?
20:49:57 <stevebaker> or reference a running stack?
20:50:13 <stevebaker> I'd like to talk heat-cfntools security fix release
20:50:25 <SpamapS> indeed
20:50:32 <shardy> Ok, lets move back to open discussion for the last 10mins and talk bps more next week
20:50:42 <shardy> I guess there will be lots more after summit anyway ;)
20:50:46 <SpamapS> next week == summit ;)
20:50:50 <shardy> #topic open discussion
20:50:59 <stevebaker> https://bugs.launchpad.net/heat-cfntools/+bug/1164756
20:51:01 <stevebaker> https://bugs.launchpad.net/heat-cfntools/+bug/1166323
20:51:03 <shardy> So two things I wanted to mention:
20:51:41 <SpamapS> stevebaker: I think once the commits are solid we should draft an announcement, and just commit/release/announce all at once.
20:51:42 <shardy> 1 - I've offered to step in for sdake doing PTL duties, then we'll have a proper vote on it (and anyone else can step up) after summit
20:52:23 <asalkeld> sounds good
20:52:25 <shardy> 2 - There a pile of VPC/Quantum related bugs in New state - stevebaker and/or jpeeler, can you triage and take some of those?
20:52:53 <stevebaker> yep, they should already be assigned, mostly to jpeeler ;)
20:53:03 <shardy> haha, nice :)
20:53:13 <shardy> Ok, anyone else got anything to discuss?
20:53:41 <asalkeld> try sleep on your flight
20:53:53 <wirehead_> I have zero timezone changes on my flight.
20:53:55 <stevebaker> SpamapS has some /tmp security fixes for heat-cfntools. I just wondered if there is anything different we should do once they are ready for release
20:54:30 <asalkeld> maybe ask the os sec folks
20:54:39 <SpamapS> stevebaker: we don't need a CVE or anything, as heat-cfntools is pretty nascent and may not even be distributed just yet..
20:54:45 <stevebaker> and Sources tarball extraction isn't working for me, which is hindering testing https://bugs.launchpad.net/heat-cfntools/+bug/1166323
20:54:56 <SpamapS> asalkeld: I did, they said incubated projects don't fall under their umbrella, and suggest an informal release.
20:55:10 <asalkeld> maybe just release fast
20:55:17 <SpamapS> yeah thats what I think we should do
20:55:20 <asalkeld> so people get the fixes
20:55:24 <stevebaker> SpamapS: OK, I guess you should just raise normal gerrit changes then
20:55:44 <SpamapS> ok, can I get a commitment for reviews in the next 2 hours?
20:55:57 <SpamapS> actually
20:56:00 <SpamapS> lets coordinate in #heat
20:56:15 <stevebaker> yep, but as i said #1166323 doesn't work for me atm
20:56:16 <SpamapS> don't need everybody for this
20:56:23 <SpamapS> stevebaker: right I have a fix for that
20:56:28 <stevebaker> sweet
20:56:38 * SpamapS is spinning a lot of non-heat plates this week unfortunately
20:57:06 <shardy> Ok, nearly out of time, anything else?
20:57:18 <shardy> Ok, thanks all
20:57:21 <shardy> #endmeeting