08:00:13 <therve> #startmeeting heat
08:00:14 <openstack> Meeting started Wed Apr 20 08:00:13 2016 UTC and is due to finish in 60 minutes.  The chair is therve. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:00:15 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
08:00:17 <openstack> The meeting name has been set to 'heat'
08:00:21 <therve> #topic Roll call
08:00:25 <therve> Hola!
08:00:27 <shardy> o/
08:00:28 <skraynev> \o/
08:00:31 <ricolin> o/
08:00:50 <ramishra> hi
08:01:01 <stevebaker> \o
08:01:11 <therve> stevebaker too late?
08:01:22 <stevebaker> therve: nah, 8pm
08:01:26 <therve> Hooray :)
08:01:38 <therve> #topic Adding items to agenda
08:01:47 <therve> #link https://wiki.openstack.org/wiki/Meetings/HeatAgenda#Agenda_.282016-04-20_0800_UTC.29
08:02:05 <therve> #topic Summit schedule
08:02:22 <therve> Okay, let's sync on that
08:02:23 <stevebaker> therve: I assume you've seen my and zane's emails
08:02:34 <therve> stevebaker, Yeah, responded and made a change
08:02:41 <therve> I hope it will fit most people
08:02:45 <shardy> So the changes proposed seem OK to me, sorry that we didn't spot the conflicts sooner
08:02:49 <therve> shardy, I guess it'll guet tough for you
08:03:03 <shardy> We will need to get someone (at least jdob) to attend the validation session
08:03:12 <therve> Yeah previous PTLs didn't work well :p
08:03:18 <skraynev> therve: hey ?!
08:03:19 <shardy> therve: Yeah, I'll obviously prioritize the TripleO session this summit
08:03:22 <ochuprykov> hi
08:03:35 <shardy> but it sounds like overall we'll have the topic coverage we need
08:03:44 <skraynev> therve: I just missed this topic about overlaps in sessions...
08:03:50 <shardy> happy to sync with folks on the beer-track regarding the heat sessions :)
08:04:00 <stevebaker> beer track!
08:04:05 <therve> :)
08:04:17 <ricolin> :)
08:04:19 <skraynev> shardy: I planed to attend on validation session
08:04:39 <skraynev> stevebaker: do we have special slot in schedule for beer ?
08:04:44 <skraynev> nice :)
08:05:06 <shardy> skraynev: yup, we need some folks from the TripleO team there, and in particular from the tripleo-ui team, as we've got some requirements related to improving the current nested validation stuff
08:05:17 <shardy> I'm pretty sure jdob will be able to cover it, but will confirm later
08:05:25 <stevebaker> skraynev: it exists in the spaces inbetween
08:05:40 <stevebaker> therve: ok, my schedule now has no clashes, thanks
08:05:48 <therve> Cool
08:06:00 <skraynev> stevebaker: and at the evening I suppose ;)
08:06:04 <therve> Well moving on if no one has complains
08:06:12 <skraynev> therve: +
08:06:27 <therve> #topic Skip meeting for the next week?
08:06:33 <therve> Yes
08:06:34 <shardy> +1
08:06:34 <skraynev> therve: it's mine
08:06:45 <skraynev> just want to raise this question ;)
08:06:57 <therve> #topic Summit evening meetup in Austin
08:06:57 <skraynev> probably we need one mail about it
08:07:01 <skraynev> for community
08:07:15 <therve> skraynev, Hum okay, I'll send it
08:07:29 <skraynev> therve: good
08:07:38 <skraynev> what's about evening meetup ?
08:07:39 <ricolin> Any free time from all of you to have a beer during summit?
08:07:51 <ricolin> :)
08:07:53 <stevebaker> I think Monday might be my only spare slot
08:08:06 <stevebaker> ...or thursday - I don't know yet :(
08:08:49 <therve> Friday night is the easiest for me, after the meetup
08:08:58 <therve> Not sure how many people are still around
08:09:22 <stevebaker> therve: I'll be dancing in a field
08:09:30 <skraynev> ricolin: Friday is ok for me.
08:09:36 <therve> stevebaker, Is that a metaphor? :)
08:09:40 <stevebaker> nope http://levitation-austin.com/
08:09:58 <ricolin> or we join stevebaker for dance?
08:10:02 <ricolin> lol
08:10:14 <stevebaker> probably standing groaning in a field actually
08:10:27 <stevebaker> or sitting
08:10:29 <therve> :D
08:10:33 <shardy> That looks like fun, sadly I'll be on a flight home late Friday
08:11:05 <skraynev> sounds like a thursday maybe a compromise
08:11:32 <stevebaker> skraynev: a chunk of the redhatters are most likely at a team dinner
08:11:52 <ricolin> try to avoid redhatters dinner
08:11:57 <stevebaker> how about early Monday evening, and those going to the ops party can continue on?
08:12:06 <ricolin> on monday or thrusday maybe
08:12:07 <stevebaker> is that the only clash on Monday?
08:12:30 <shardy> There's the booth crawl and an Intel thing some folks may need to get to
08:12:40 <skraynev> stevebaker: I am ok with Monday, because I lated with registration on Monday's party ;)
08:12:45 <skraynev> so I don't mind
08:12:50 <shardy> I guess there's clashes whenever you go for it though, so I say go with the majority :)
08:13:08 <stevebaker> therve: do you want to post a doodle poll?
08:13:27 <therve> stevebaker, I'm sure some community member would be happy to do it
08:13:30 <therve> wink wink
08:13:56 <ricolin> hahaha
08:14:00 <skraynev> stevebaker, shardy: Friday morning ? or we may combine evening meetup + beer
08:14:06 <stevebaker> its like I've got a reputation or something :o
08:14:08 <ricolin> I would love to do it!
08:14:38 <ricolin> Friday morning sounds good
08:15:24 <stevebaker> skraynev: yes, it doesn't have to be evening and alcohol based
08:15:24 <ricolin> push this to mailing list later:)
08:15:44 <shardy> +1 I'd be glad to sync with folks over coffee Friday am, as the TripleO/Heat afternoon meetups clash
08:15:54 <therve> Yep sounds good
08:15:57 <shardy> informal pre-meetup-meetup :)
08:16:27 <ricolin> nice!
08:16:42 <skraynev> therve: great. I hope, that now I fixed my fault with clashed sessions ;)
08:16:56 <therve> skraynev, I was completely joking
08:17:07 <skraynev> therve: I know :P
08:17:28 <therve> Alright
08:17:34 <therve> #topic Open discussion
08:18:03 <therve> liberty jobs are stuck for now because of some weird kilo reqs, hopefully it will resume soon
08:18:34 <shardy> Does anyone feel like taking https://bugs.launchpad.net/heat/+bug/1466694 off my hands?
08:18:35 <openstack> Launchpad bug 1466694 in heat "admin role can't delete other tenants stacks" [Medium,Triaged] - Assigned to Steven Hardy (shardy)
08:18:45 <ricolin> mitaka gate still require this fix
08:18:46 <ricolin> https://review.openstack.org/#/c/303890/
08:18:56 <stevebaker> oh, I've had a thought which I wanted to discuss
08:19:04 <shardy> It's a long-standing operator complaint, and I started a ML thread where ayoung pointed to a (now landed AIUI) keystone feature we could/should use
08:19:28 <shardy> It's looking less likely I'll have time in the near future to do that, so if anyone wants it I'm happy to pass it on ;)
08:19:48 <shardy> http://lists.openstack.org/pipermail/openstack-dev/2015-November/079006.html
08:20:44 <ricolin> can help on survey:)
08:20:47 <shardy> keystone patch: https://review.openstack.org/#/c/240719/
08:20:56 <stevebaker> shardy: you could target it to n-1 and make it unassigned - see who picks it up
08:21:18 <shardy> stevebaker: sure, I just wanted to mention it so it doesn't look like I just dropped it completely :)
08:21:40 <ramishra> shardy: may be I can give it a go
08:21:50 <stevebaker> my topic: our client plugins each create a client object with a shared token, what we should be doing is using a shared session...
08:22:07 <shardy> ramishra: Ok, thanks - we can chat about it after
08:22:27 <ramishra> shardy: sure
08:22:30 <stevebaker> but instead of doing that why don't we adopt the shade project and just use its client references - it already does all the right things with sessions
08:22:41 <therve> -1
08:22:55 <stevebaker> it may be completely inappropriate considering its linked with os_client_config
08:23:00 <shardy> +1 on sessions, but do we really need another layer of client abstraction?
08:23:24 <stevebaker> shardy: its just to get the client object refs, its otherwise ignored. except....
08:23:27 <therve> Yeah I don't want to get down that hole
08:23:51 <shardy> Like, isn't python-openstacksdk supposed to be the unified client?
08:23:58 <shardy> how does that relate to shade?
08:24:09 <stevebaker> shade has this task manager construct, which handles retries on rate limited failures - we have completely failed to address that in our api calls thus far
08:24:32 <stevebaker> not completely failed, but failed to address it completely
08:25:45 <shardy> Having a way to share common client related code seems great, but it's weird that we have shade under openstack-infra and openstacksdk under the "user" openstack org
08:25:57 <therve> stevebaker, So what problem using sessions solve?
08:26:01 <stevebaker> shardy: different opinions in sdk lib styles, and shade *might* be more battle hardened since its used by infra
08:26:13 <stevebaker> therve: token refreshing for free I think
08:26:36 <therve> stevebaker, That sound suspicious?
08:26:46 <therve> You need the password for that, no?
08:26:50 <stevebaker> shardy: shade is used by the ansible modules too, there is an argument for taking it out of the infra namespace but I haven't raised that yet
08:26:57 <shardy> stevebaker: Yeah, I guess that's my point, it's arguably safer to stick to using the python-fooclients then we're not taking a side in the sdk duplication
08:27:18 <stevebaker> shardy: shade is a wrapper over the fooclients
08:27:32 <therve> shade shouldn't exist :/
08:27:56 <stevebaker> anyhoo, if nothing else we should look at it for how it initialises clients
08:29:39 <therve> I'm all for improving that. I'm not sure what's the point of using session except "that's recommended"
08:30:01 <shardy> stevebaker: +1 on looking at it, but it'd be interesting to follow up on the org-change debate and understand why it's an infra specific tool atm
08:30:37 <stevebaker> shardy: it was just started by infra folk, thats all
08:31:09 <shardy> Yeah, I guess my point is that to have a worthwhile common client abstraction, there should be only one
08:32:28 <shardy> therve: I think the main advantage of sessions is you're using common code for all the auth and http session interaction
08:32:44 <shardy> e.g less duplication between clients in the long run, and potentially slightly less runtime overhead
08:33:11 <shardy> we're already sharing auth plugins, so it's a minor additional step to share the session (which references the auth plugin)
08:33:52 <therve> shardy, Yeah, I'm intrigued if in Heat's case it may not be dangerous to share http sessions
08:34:20 <stevebaker> not in the scope of a context I assume
08:34:55 <shardy> Yeah, if you actually used it concurrently it might not work, but I don't think that's what's being proposed, given the scope of an engine worker?
08:37:14 <therve> OK... Need to discuss it more next week :)
08:37:24 <therve> Anything else?
08:37:34 <stevebaker> i got nothin
08:38:05 <therve> Oki doki
08:38:08 <therve> #endmeeting