21:00:12 <flaper87> #startmeeting Zaqar
21:00:12 <openstack> Meeting started Mon Jun  1 21:00:12 2015 UTC and is due to finish in 60 minutes.  The chair is flaper87. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:13 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:16 <openstack> The meeting name has been set to 'zaqar'
21:00:25 <flaper87> WHAT'S UUUUUUUUUUUUUUUUUUUUUUUUUUP PEOPLEEEEEEEEEEEEEE ????????????????????????????????????????????????????
21:00:31 * flaper87 super excited
21:00:36 <flaper87> vkmc: dynarro flwang ?
21:00:46 <dynarro> \o/
21:00:56 <flaper87> our agenda for today
21:00:58 <flaper87> #link https://wiki.openstack.org/wiki/Meetings/Zaqar#Agenda
21:01:02 <kragniz> sup!
21:01:04 <sriram> o/
21:01:07 <sigmavirus24> o/
21:01:10 <flaper87> helloooo there :)
21:01:15 * flaper87 overwhelmed
21:01:19 * sigmavirus24 lurks like kragniz
21:01:20 <flaper87> y'all welcome
21:01:28 <flaper87> sigmavirus24: we recruited kragniz
21:01:30 <kragniz> sigmavirus24: I lurk better than you lurk
21:01:32 <flaper87> he doesn't know yet
21:01:37 <sigmavirus24> flaper87: sounds legit
21:01:48 * sigmavirus24 waves to vkmc
21:01:51 <flaper87> #topic Summit Feedback
21:02:02 <flaper87> vkmc: want to start?
21:02:27 * flaper87 throws cold water on vkmc's face
21:02:33 <flaper87> wake up girl!
21:02:43 <vkmc> whaaaaaa
21:02:44 <vkmc> sorry
21:02:45 <flaper87> Now you know what we, Europeans, suffer with this meetings.
21:02:45 <vkmc> hi!
21:02:52 <flaper87> these*
21:03:04 * vkmc waves
21:03:10 <sigmavirus24> flaper87: don't be so dramatic. =P
21:03:15 <flaper87> vkmc: feedback from the summit? Anything you want to share to start with?
21:03:21 * flaper87 cries on sigmavirus24's shoulder
21:03:28 <vkmc> ha! it was a pretty interesting summit
21:03:35 <sigmavirus24> there there
21:03:51 <vkmc> we had great sessions with people working on Sahara and Heat
21:04:45 <ryansb> glad we could help :)
21:08:48 <ryansb> sooo, meeting?
21:08:50 <kragniz> rip our brave leaders
21:09:00 <ryansb> time for anarchy then
21:09:00 * kragniz eyes up freenode
21:10:38 <ryansb> ¯\_ツ_/¯
21:11:38 <vkmc> netsplit
21:11:59 <ryansb> welcome back
21:12:06 <vkmc> thx
21:12:38 <vkmc> flaper87, netsplit!
21:12:43 <flaper87> knock?
21:12:44 <flaper87> sorry, connection
21:12:45 <flaper87> :(
21:12:46 <flaper87> yeah
21:12:54 <flaper87> what was the last message you got ?
21:13:14 * flaper87 checks logs
21:13:39 <vkmc> I was updating folks about the summit, not sure if those reached the other end
21:13:45 <flaper87> #link http://lists.openstack.org/pipermail/openstack-dev/2015-May/064739.html
21:13:46 <ryansb> flaper87: feedback from summit? Anything you want to share to start with?"
21:13:49 <flaper87> for people reading logs, that's the discussion on the mailing list
21:13:54 <ryansb> last I saw
21:13:55 <flaper87> It was indeed a very interesting summit... As vkmc said, there were quite a few sessions with people from other projects
21:13:59 <flaper87> these sessions - as we wanted - ended up in action items that - unlike other summits - are actually being worked on
21:14:04 <flaper87> That's great news for the project and it means we'll keep spending time on it
21:14:08 <flaper87> if anyone tries to bail out on the project, I'll .... actually... you know what's going to happen
21:14:19 <flaper87> that's the overall summary from me
21:14:21 <vkmc> lol
21:14:26 <flaper87> I'm quite happy with the result
21:14:40 <flaper87> anyone wants to add something? otherwise we can get to the actual plans/items
21:15:22 <vkmc> plans/items \o/
21:15:31 <ryansb> +
21:15:33 <ryansb> +1
21:15:49 <sigmavirus24> ++1 does nothing ryansb
21:15:53 <sigmavirus24> at least not in C =P
21:16:03 <dynarro> let's go on!
21:16:45 <ryansb> sigmavirus24: works in gerrit
21:17:23 <vkmc> haha
21:18:13 <vkmc> flaper87, ?
21:20:02 <dynarro> netsplit again, I guess
21:20:14 * kragniz eyes up freenode again
21:21:25 <flaper87> fuck this shit
21:21:27 <flaper87> #topic Liberty specs and plans
21:21:37 <flaper87> no idea what's going on
21:21:41 <flaper87> but really, fuck this shit
21:21:44 <flaper87> #link https://review.openstack.org/#/q/status:open+project:openstack/zaqar-specs,n,z
21:21:49 <flaper87> those are our specs so far... There are some missing but I believe their priority is lower
21:21:55 <flaper87> flwang is not around :(
21:21:59 <flaper87> anyway, lets start with https://review.openstack.org/#/c/185951/
21:22:05 <flaper87> dynarro: anything you want to say/add ?
21:22:20 <flaper87> I think it's a no-brainer and we should let her work on that (will do proper introductions at the end)
21:22:27 <flaper87> vkmc: any comments on that one?
21:22:35 <flaper87> I suggested moving websockets out of there
21:22:52 <flaper87> since that's not an API change but rather a complete different protocol
21:23:08 <vkmc> yeah, that would make sense
21:23:20 <flaper87> s/would make/makes/
21:23:22 <flaper87> :P
21:23:25 <vkmc> haha
21:23:27 <dynarro> yeah, flaper87 I've changed those things you told me
21:23:53 <vkmc> the specs looks good to me as is now
21:23:57 <flaper87> ok, since netty netty is shitty shitty lets move on to the next one before I get kicked out again
21:23:59 <flaper87> vkmc: agreed
21:24:06 <flaper87> vkmc: pls, +2 and I'll approve later
21:24:26 <vkmc> certainly the client needs some love :)
21:24:27 <flaper87> #link https://review.openstack.org/#/c/186346/
21:24:30 <vkmc> thanks dynarro for writting the spec!
21:24:34 <flaper87> vkmc: oh yeah!
21:24:37 <flaper87> dynarro: danke
21:24:45 <dynarro> np!
21:24:49 <dynarro> :)
21:24:58 <flaper87> So, that one is simple too. Just a backlog dir so that people not interested in working on features can still propose them
21:25:14 <flaper87> it should be an easier way to get proposals from other folks
21:25:33 <flaper87> that's more like a heads up :P
21:25:36 <flaper87> #link https://review.openstack.org/#/c/186345/
21:25:47 <flaper87> ok, that's one of the juicy ones for this cycle
21:25:54 <flaper87> It's for pre-signed URLs
21:26:29 <flaper87> In a gist, pre-signed URLs will grant access to external - therefore unauthenticated - users to some resources in Zaqar
21:26:33 <ryansb> flaper87: I had a thought on that: would project-scoped tempurls make any sense?
21:26:50 <ryansb> to grant access to, say, a group of queues?
21:27:06 <flaper87> ryansb: mmh, in that case, I'd rather have a shared user with a limited role
21:27:27 <ryansb> I see; just a thought
21:27:34 <flaper87> not a crazy one, though.
21:27:51 <flaper87> It'd probably be easier once we have policy and the bases of pre-signed URLs
21:28:21 <ryansb> fajr
21:28:26 * flaper87 just noticed ryansb and therve have commented on the spec
21:28:26 <ryansb> *fair enough
21:28:36 <flaper87> ryansb: thanks for taking the time
21:28:44 <flaper87> I'll read the comments tomorrow and address them
21:29:09 <flaper87> That said, I'm curious to know if vkmc, dynarro and/or flwang (who's not here) have questions regarding the feature
21:29:22 <flaper87> (or ryansb, or anyone)
21:29:24 <flaper87> :P
21:29:37 <vkmc> I don't have any question regarding the feature, I should reread the spec
21:29:55 <flaper87> The main usecase is to allow these users/services - say a guest-agent - to access those resources without too much magic
21:30:01 <vkmc> I'd like to read, though, how is the Heat use case specifically
21:30:08 <vkmc> if its somewhere :)
21:30:16 <ryansb> vkmc: for the signed urls?
21:30:22 <ryansb> that one isn't written down afaik
21:30:44 <ryansb> it's more or less "send a tempurl to a client so we can send it messages"
21:30:59 <vkmc> ryansb, to manage Heat resources?
21:31:34 <ryansb> no, to have clients (like cloud-init/heat-cfntools) on nodes be able to have messages sent to them without polling heat directly
21:32:00 <ryansb> so when heat creates a server, it can fire the data off to the zaqar queue for that client, and the client picks it up when it boots
21:32:01 <vkmc> oh cool
21:32:13 <flaper87> In addition to that, Sahara and Murano are looking forward to use this feature for their guest-agents
21:32:31 <ryansb> there's the other use case for heat, which is sending events like "your stack is done" or "this resource blew up" to users without polling heat
21:32:42 <ryansb> tl;dr: we want to stop having people poll us so much
21:32:47 <vkmc> that makes sense
21:32:59 <vkmc> thanks ryansb
21:33:01 * flaper87 read to day a proposal for "Efficient polling"
21:33:10 * flaper87 -> head -> desk
21:33:10 <vkmc> haha
21:33:14 <ryansb> lol
21:33:27 <flaper87> anyway... moving on
21:33:29 <ryansb> at least they didn't say "low-latency polling"
21:33:31 <flaper87> #link https://review.openstack.org/#/c/179673/ (Policy)
21:33:40 <flaper87> flwang proposed this before the summit
21:33:52 <flaper87> It follows pretty much what other openstack projects do w.r.t policies
21:34:03 <flaper87> It proposes adopting oslo.policy and adding a policy.json
21:34:27 <flaper87> I'd be happy to have this for several reasons. Besides having policies, I've been also thinking how to extend them in a per-queue basis
21:34:45 <flaper87> call me crazy but even in a per-message basis
21:34:52 <flaper87> but one step at a time
21:34:54 <vkmc> hm
21:35:02 <flaper87> I'll be writing those thoughts down this week
21:35:05 <vkmc> seems you like policies
21:35:09 <flaper87> so we can start brainstorming over a spec
21:35:10 <ryansb> per-message sounds a little crazy
21:35:18 <ryansb> but per queue/project would be great
21:35:34 <flaper87> vkmc: no, I just don't like people reading my messages
21:35:36 <vkmc> ryansb, +1
21:35:37 <flaper87> :P
21:35:43 <vkmc> flaper87, lol
21:35:47 <flaper87> ryansb: yeah, that'd be the first step
21:36:10 <flaper87> I think I have a good proposal
21:36:15 <flaper87> anyway
21:36:26 <flaper87> does anyone have objections/thoughts on that spec ?
21:36:37 <flaper87> I nitpicked on it today
21:36:47 <flaper87> but other than that, it seems pretty straight-forward
21:36:47 <vkmc> no right now, will review tomorrow
21:37:05 <flaper87> vkmc: why didn't you read the specs *before* the meeting? ah? ah? ah?
21:37:20 * flaper87 sent a meeting invitation a week ago
21:37:29 <vkmc> flaper87, I got lost on debugging other stuff
21:37:37 <vkmc> the day just... happened
21:37:41 <vkmc> and I didn't notice
21:37:44 <flaper87> vkmc: it's aaaaaaaaaaaaaaaaaaaaaaaaaalways the same excuse
21:37:45 <flaper87> :P
21:37:49 * flaper87 hugs vkmc
21:37:52 <vkmc> hahaha
21:37:56 <dynarro> lol
21:37:58 <vkmc> I'll review tomorrow, I promise
21:38:14 <flaper87> yeah right... Send me some gummybears while you are at it
21:38:26 <vkmc> sure sure
21:38:35 <flaper87> anyway, that's all spec wise
21:38:50 <flaper87> actually, I lied
21:39:05 <flaper87> vkmc: I'll get to your websocket review soon
21:39:16 <flaper87> however, It'd be amazing to complete that spec during liberty
21:39:16 <vkmc> flaper87, sounds good!
21:39:24 <flaper87> you'll have to propose it for liberty
21:39:35 <flaper87> I guess moving it and adding a history should be enough
21:39:38 <vkmc> yeah, I'd like to have up to claims in liberty 1
21:39:40 <flaper87> I've seen Nova folks doing so
21:40:08 <vkmc> we need to discuss what to do with pools, flavors
21:40:08 <flaper87> Also, another missing spec (shaifali said she's going to work on that) is the tests refactor
21:40:26 <flaper87> I need to triple-check what's her time commitment
21:40:42 <flaper87> vkmc: agreed, lets get the data plane out of the way
21:40:49 <flaper87> .. first
21:40:56 <vkmc> sounds good
21:41:07 <vkmc> so I'll propose the spec for liberty then
21:41:13 <vkmc> and rebase the messages patch for websockets
21:41:34 <flaper87> vkmc: just make sure you don't endup debugging rst for a day
21:41:37 <flaper87> ok ?
21:41:38 <flaper87> :P
21:41:48 <vkmc> I cannot promise that
21:41:52 * flaper87 hides from vkmc punches
21:41:55 <vkmc> haha
21:42:10 <flaper87> ok, now, that's really it spec wise
21:42:17 <flaper87> #topic clean ups
21:42:27 <flaper87> There are several clean ups to do in the codebase
21:42:39 <flaper87> For instance, we still have old sqlalchemy models around
21:43:01 <flaper87> I'll be dedicating time to some of those clean ups but I'll sure file bugs when I find them
21:43:15 <flaper87> if someone shows up looking for something to do, you know where to look
21:43:33 <flaper87> We also need to add an env var for zigo so he can run tests
21:43:41 <flaper87> It's not as straightforward as I thought
21:43:50 <flaper87> since we're using config files for our tests
21:44:06 <ryansb> shouldn't env vars override config files?
21:44:12 <flaper87> I proposed him to use sed but I never heard back from him. I think  he just rage-quited and ran away
21:44:40 <flaper87> ryansb: yeah, it's just the way these tests are structured that makes it hard
21:44:56 <flaper87> unfortunately, we ended up having mongdb:// urls everywhere
21:45:11 <flaper87> because we used to have sqlite://:memory: there
21:45:24 <ryansb> ah :/
21:45:25 <flaper87> That was a mistake that we need to fix somehow
21:45:35 <vkmc> maybe we could use mongo mock for that?
21:45:36 <vkmc> in the long term
21:45:38 <flaper87> especially requesting mongodb for unittests, which is not ideal
21:45:55 <flaper87> I looked into mongomock and I'm afraid that it may not support all the features we need
21:45:58 <flaper87> :/
21:46:06 <flaper87> We use many things from mongo
21:46:11 <vkmc> oh hmm
21:46:13 <flaper87> here's my thought re-tests
21:46:49 <flaper87> Once they are merged, we should just let devs choose what they want to test against (redis/mongodb) and then have our gates as we have them
21:47:00 <flaper87> (by merged I mean the work shaifali is going to do)
21:47:34 <flaper87> And move all tests requesting running dbs to `functional` where they belong
21:47:48 <flaper87> In other words, Shaifali has a lot of work
21:48:49 <flaper87> anyway, 10 mins left
21:48:54 <vkmc> we probably can divide some work on that side
21:49:00 <flaper87> vkmc: yeah
21:49:12 <flaper87> I think we'll have to wait for the merge to happen, though
21:49:24 <flaper87> just to have a clearer view of how the tests structure will look like
21:49:28 <vkmc> yeah, certainly
21:49:32 <flaper87> at the very least, we should wait for the spec
21:49:38 <flaper87> that said, I agree
21:49:42 <flaper87> we should split that
21:50:15 <flaper87> We can ask dynarro to do 99.9 % of the work and we do the remaining 0.1%
21:50:28 <dynarro> whaaaat!!!
21:50:29 <vkmc> +1 flaper87
21:50:33 <vkmc> oh hi dynarro
21:50:42 <dynarro> haha
21:50:43 <flaper87> dynarro: oh you're here? oooooooops ?
21:50:50 <flaper87> oh well
21:50:57 <flaper87> #topic Open Discussion
21:51:09 <flaper87> As usual, I'd like to welcome our new mentee
21:51:26 <flaper87> We've a record, we've been mentoring Outreachy mentees every cycle
21:51:31 <flaper87> and I'm proud of us
21:51:40 <flaper87> In this cycle, dynarro will be our mentee
21:51:44 <dynarro> \o/
21:51:46 <vkmc> welcome dynarro!
21:51:52 <flaper87> and I'd like to thank vkmc for stepping up as a mentor in this cycle
21:51:53 * angvp claps
21:51:58 <vkmc> happy to have you on board :)
21:51:59 <ryansb> welcome!
21:52:15 <flaper87> angvp: oh look showed up just when we finished splitting the work
21:52:18 <flaper87> ....
21:52:23 <dynarro> haha
21:52:32 <flaper87> so, really. Welcome, dynarro
21:52:49 <flaper87> we're all excited to have you and you better do a great job or we'll never like you again
21:52:50 <flaper87> EVER!
21:52:54 <flaper87> >.>
21:52:55 <dynarro> thanks! I'm SO happy to be part of it!
21:53:05 <dynarro> hahaha
21:53:28 <vkmc> dynarro, you know you are doomed, right? you won't be able to leave Zaqar, ever!
21:53:49 <dynarro> hahaha....I know...
21:54:01 <flaper87> I'd like to also thank all the folks that participated in sessions at the summit.... If you were here, I'd mention you but I guess ryansb will take all the credits this time
21:54:02 <flaper87> :)
21:54:11 <ryansb> woo
21:54:20 <flaper87> ryansb: thanks for all your amazing feedback... We look forward to see tons of patches landing everywhere
21:54:28 <ryansb> Also, during the summit we talked a bit about using swift as a backend for zaqar. So far, I have the basic (post/get/delete) flow working and I'll probably post a review later in the week
21:54:33 <flaper87> (patches related to Zaqar, we don't care about the rest)
21:54:44 <angvp> flaper87 vkmc ok, i can help dynarro with 0.1% of the work too :P
21:54:46 <vkmc> ryansb, that's great!
21:54:46 <ryansb> heh, yeah. Hoping to get some zaqar stuff into L1-heat
21:54:49 <flaper87> ryansb: wow, sweet, niiiiiiiiiiiiiice!
21:54:57 <flaper87> I'd love to see that backend
21:55:05 <dynarro> angvp: hahaha thanks
21:55:14 <ryansb> currently I sort of hacked it, I'm still using redis for the management store
21:55:38 <ryansb> but my eventual hope is to do sqlalchemy for management and swift for all the queues/messages
21:55:49 <flaper87> ryansb: that's ideal
21:55:55 <flaper87> any reason why you're not using sqla already ?
21:56:10 <ryansb> because I tried it briefly and got an indecipherable error message
21:56:22 <flaper87> Redis will work until you try to create a pool in it (since it's not supported :P)
21:56:36 <flaper87> ryansb: mmh, ok, I'll give it a try and fix whatever is broken there
21:56:47 <flaper87> angvp: actually, you do it
21:56:51 <flaper87> :P
21:56:51 <ryansb> it may have been due to operator (me) error
21:57:02 <flaper87> ohhh, it's always ops fault
21:57:05 <flaper87> :P
21:57:13 <ryansb> only when I'm the operator ;)
21:57:16 <ryansb> in any case, pools shouldn't be needed for swift
21:57:36 <flaper87> ryansb: btw, are you developing it in the code base? or as an external plugin ?
21:57:40 <ryansb> since the scaling/storage is all handled in swift, no need for us to wrap an API
21:57:45 <flaper87> you know you can have it as a third-party plugin, right ?
21:58:05 <ryansb> I'm doing it in the code base, I didn't feel like messing around to make it a plugin
21:58:24 <flaper87> ryansb: but you may want to have swift+redis in the data-plane and you'll need pools in that case
21:58:36 <ryansb> hrm. maybe
21:58:39 <flaper87> pools are not just to scale the same storage type
21:58:44 <flaper87> that's what I wanted to say
21:58:49 <flaper87> not that you really need that combination
21:58:56 <ryansb> yeah, I see now
21:58:56 <flaper87> ryansb: Thanks for hacking on that
21:59:05 <flaper87> ryansb: has it been hard to implement ?
21:59:14 <flaper87> (I mean form a Zaqar's perspective)
21:59:24 <ryansb> not so far, but there are a lot of things I haven't done yet (claims)
21:59:33 <flaper87> gotcha, those might be messy
21:59:42 <flaper87> ok, we ran out of time :)
21:59:54 <ryansb> ah. Well I ran out of implementation :)
22:00:02 * ryansb moves back to #zaqar
22:00:03 <flaper87> Next week we'll have the meeting at this same time and the week after we'll be back to our alternate
22:00:09 <flaper87> .. schedule
22:00:11 <vkmc> cool!
22:00:12 <flaper87> THANKS ALL!
22:00:17 <flaper87> LONG LIVE ZAQAR!
22:00:19 <flaper87> #endmeeting