15:02:39 <kgriffs> #startmeeting zaqar
15:02:40 <openstack> Meeting started Mon Aug 11 15:02:39 2014 UTC and is due to finish in 60 minutes.  The chair is kgriffs. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:02:41 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:02:43 <openstack> The meeting name has been set to 'zaqar'
15:02:48 <malini> o/
15:02:54 <kgriffs> o/
15:03:09 <kgriffs> who else is around?
15:03:15 <vkmc> o/
15:03:20 <prashanthr_> kgriffs: o/
15:04:35 <kgriffs> ok, now that we know who the *real* team members are...
15:04:39 <kgriffs> ;)
15:04:46 <malini> :D
15:05:02 <kgriffs> I just glanced at the agenda - didn't look like flaper87 had updated it (for shame!)
15:05:05 <kgriffs> so...
15:05:12 <kgriffs> I'm going to shoot from the hip here
15:05:23 <kgriffs> #topic docs
15:05:31 <kgriffs> vkmc: how goes the battle?
15:05:53 * kgriffs wishes everyone to remember the pen is mightier than the sword
15:07:00 <vkmc> kgriffs, is it going great, right now there are some patchs waiting for reviews
15:07:04 <kgriffs> for my part, I saw Catherine commented on my patch - I will get those fixed right after this
15:07:22 <vkmc> I think the dev guide is looking good :)
15:07:34 <vkmc> I also started working on the configuration docs
15:07:48 <kgriffs> great. thanks everyone who has been helping with docs!
15:08:20 <kgriffs> I will keep working on them this week along with vkmc. Love to get some more help.
15:08:43 <vkmc> Oz is giving me a hand with that... I asked him some tips on how to tune Zaqar for different configurations so we provide some useful examples
15:09:09 <kgriffs> cool. we owe him some poptartz
15:09:14 <vkmc> yeah :D
15:09:22 <malini> he is a wizard..can make his own ;)
15:09:43 * kgriffs thinks that is an awesome super power to have
15:10:56 <kgriffs> btw, I spoke with a Heat core reviewer last friday. I'll share more about that later, but he did mention it wasn't clear how to do point-to-point RPC-style communication.
15:11:50 <kgriffs> he didn't realize queues are duplex by default, and once I explained how that worked and what the echo param was, he was cool. so anyway, I'd like to help write up some how-to's/tutorials for different things like that
15:12:00 <kgriffs> malini: we can base it off of cloud backup and RSE
15:12:47 <malini> kgriffs: does tht need to go into user guide?
15:13:10 <vkmc> +1 malini
15:13:13 <kgriffs> yeah
15:13:18 <kgriffs> should it also be in the dev guide?
15:13:44 <vkmc> hm I don't think so... it's about using Zaqar, not developing it
15:13:54 <kgriffs> ok
15:14:03 <kgriffs> I'm looking at your outline
15:14:17 <malini> vkmc: tht's what I think too
15:14:39 <kgriffs> where would be a good place to walk the user through implementing 3-4 messaging patterns with Zaqar?
15:15:10 <malini> let's chk what other projects have done
15:15:20 <vkmc> kgriffs, what do you mean with implementing a message pattern?
15:15:25 <kgriffs> vkmc: like
15:15:52 <kgriffs> here is how you would do point-to-point, and how you would get a response back on the same queue
15:16:35 <kgriffs> "suppose i wanted to communicate with a thing over there in this particular way... here are the steps to do that"
15:16:56 <kgriffs> does that make sense?
15:17:03 <vkmc> yeah it does
15:17:07 <malini> do we already talk abt pub-sub, producer-consumer somewhere?
15:17:42 <vkmc> IMO it's more related to the deployment rather than the development
15:18:08 <malini> hmmm..now I think its a develpment thing
15:18:29 <kgriffs> wait...
15:18:31 <vkmc> (maybe I misunderstood)
15:18:34 <kgriffs> yeah
15:18:51 <kgriffs> so this is how an application developer would use marconi to connect components of their distributed system
15:19:16 <kgriffs> like, how would heat use Zaqar to talk to 3rd-party config management tools?
15:19:22 <AAzza> for me it is user thing, and quite important
15:19:40 <AAzza> wouldn't mind to read it by myself)
15:19:53 <malini> But this is more abt how you wud write your application to use queues in a specific way
15:20:06 <prashanthr_> kgriffs: I feel that must be very close to the introduction as that would help understand if it is relevant to them immediately.
15:20:07 <kgriffs> basically, we need to take this and turn it into a more how-to style with curl examples and stuff: https://wiki.openstack.org/wiki/Use_Cases_(Marconi)
15:20:54 <kgriffs> prashanthr_: good thought. It would be nice to summarize some of the things you can do with Zaqar in the intro in the user guide, maybe also dev guide?
15:21:32 <prashanthr_> kgriffs: Yes most of the times we choose a software / third party library if the use case kind of matches our need :)
15:21:44 <vkmc> is true that is about how would  you write your application, but still you are *using* Zaqar
15:22:29 <kgriffs> I think it does belong in the user guide for the most part
15:23:38 <vkmc> although yeah... is confusing because it's for a developer user
15:23:54 <malini> :D
15:24:01 <kgriffs> right. developer guide is for contributors to the project, right?
15:24:15 <kgriffs> while user guide is for app developers building on top of openstack components?
15:24:16 <vkmc> but.. if they are going to use Zaqar as a black box, then I think it's for the user guide
15:24:18 <vkmc> exactly
15:24:43 <prashanthr_> ahh okay.
15:24:46 <kgriffs> ok. Can you add that to the outline somewhere? It isn't a grad requirement, so we can wait and do it last
15:24:51 <kgriffs> but I don't want to forget
15:25:04 <vkmc> here is an example of user guide http://docs.openstack.org/api/openstack-compute/2/content/, Nova's user guide
15:25:31 <vkmc> and it's conformed by general API information, API operations and extensions
15:25:40 <vkmc> and I'd add, use cases
15:26:03 <vkmc> to include how to implement a message pattern using Zaqar
15:26:30 <vkmc> would that work?
15:27:43 <kgriffs> hmm, the nova guide looks pretty bare - more like a reference. but I don't know of any other place to document how-to's since the dev guide is for contributors, not end users
15:28:33 <vkmc> yeah
15:29:03 <kgriffs> maybe under "examples"
15:29:14 <kgriffs> swift has some stuff
15:29:14 <kgriffs> http://docs.openstack.org/api/openstack-object-storage/1.0/content/examples.html
15:29:19 <vkmc> well, we could ask the doc team to give us some feedback about that :)
15:29:21 <kgriffs> like, showing how to page through results
15:29:50 <kgriffs> that looks closer to what we were talking about
15:30:45 <kgriffs> anyway, yeah I guess we could ask
15:31:43 <kgriffs> ok, moving on?
15:32:08 <malini> yes!
15:32:14 <vkmc> we can add it to the user guide outline and ask when the moment comes, but I think it is ok :)
15:33:09 <kgriffs> kk
15:33:10 <kgriffs> thanks
15:33:46 <vkmc> np
15:33:53 <kgriffs> #topic benchmark results
15:34:02 <kgriffs> so... sriram isn't here i guess
15:34:22 <kgriffs> i was hoping to do an initial test today
15:35:00 <kgriffs> I've been working on a bare metal cluster as well
15:35:27 <kgriffs> so we can compare against vms and also should make tests a little more consistent because we won't have noisy neighbor
15:35:38 <kgriffs> I hope to have that cluster set up by wed
15:35:47 <kgriffs> malini: are you in the office today?
15:35:50 <malini> yes
15:36:00 <kgriffs> is sriram around?
15:36:36 <malini> he is busy on another high priority stuff..But will chk with him
15:37:54 <kgriffs> ok. I can probably run the tests but I need the info on the statsd dashboard... URL and login if he could email me
15:38:12 <malini> I'll ping him for tht
15:38:16 <kgriffs> kk, thanks!
15:38:50 <kgriffs> ok, looking at other graduation work items...
15:38:55 <kgriffs> #topic redis
15:39:09 <kgriffs> prashanthr_: I noticed you submitted a new patchset?
15:39:27 <prashanthr_> kgriffs: Yes i did submit one today
15:39:33 <prashanthr_> was able to pass all unit tests
15:39:37 <prashanthr_> on py27 and py33
15:39:47 <prashanthr_> along with all the comments on the previous patchset
15:40:16 <prashanthr_> https://review.openstack.org/#/c/97178/14
15:41:18 <AAzza> prashanthr_: i'm reviewing it now, have some concens about string/bytes, especially in claims controller.
15:41:23 <kgriffs> awesome, thanks!
15:41:27 <kgriffs> let's all help review
15:41:41 <prashanthr_> AAzza : Sure. Waiting for comments :)
15:42:07 <prashanthr_> thank you kgriffs. I have also created a draft of the design document
15:42:12 <prashanthr_> we were speaking about last week
15:42:13 <prashanthr_> https://wiki.openstack.org/wiki/Redis_Storage_Driver_Design_(Zaqar)
15:42:47 <AAzza> prashanthr_: honestly don't know yet how to do it good, but will play with code and probably tomorrow report about resutls :)
15:43:02 <prashanthr_> AAzza: Sure.
15:43:41 <kgriffs> prashanthr_: ah, cool. looks like a good start to the doc
15:44:00 <kgriffs> that will be super helpful... we will port it to the dev/contributor guide eventually I imagine
15:44:03 <kgriffs> vkmc: ^^
15:44:11 <kgriffs> #link https://wiki.openstack.org/wiki/Redis_Storage_Driver_Design_(Zaqar)
15:44:13 <vkmc> +1
15:44:15 <prashanthr_> kgriffs: Sure. that would be useful
15:44:18 <malini> kgriffs: u should have the env info from sriram now
15:44:24 <kgriffs> malini: thanks!
15:44:39 <kgriffs> ok, prashanthr_ anything else?
15:45:09 <prashanthr_> kgriffs: Nope. that's all from my side.
15:45:23 <kgriffs> cool, thanks a ton! I know this has been a lot of work. hopefully it's been fun too. :)
15:45:37 <prashanthr_> kgriffs: Nope. that's all from my side.
15:45:41 <kgriffs> ok
15:45:53 <kgriffs> vkmc, malini: can you help review prashanthr_'s patch?
15:46:06 <malini> sure
15:46:32 <kgriffs> thanks - it is a big one, so if you don't have time to look at all of it, claims hasn't seen a lot of eyes yet afaik
15:47:02 <prashanthr_> kgriffs: True.
15:47:43 <vkmc> of course
15:48:03 <kgriffs> thanks!
15:48:05 <kgriffs> ok, moving on
15:48:55 <kgriffs> #topic heat feedback
15:49:00 <kgriffs> #link https://etherpad.openstack.org/p/marconi-scratch
15:49:03 <kgriffs> I put some notes there
15:49:56 <kgriffs> So, they have a couple important use cases, and will have to basically recreate Zaqar to do them; obviously, they would rather use Zaqar but they aren't going to wait forever for us to graduate
15:50:27 <kgriffs> do those use cases make sense? it is hard to convey them via text
15:52:03 <malini> the second one does
15:52:15 <malini> is the first dependent on us implementing notifications?
15:53:27 <kgriffs> no, actually
15:54:26 <kgriffs> they could publish an event to a queue and the third-party tool could just slurp from Zaqar API
15:54:48 <kgriffs> (anyway, that is what I understood from Randall)
15:54:59 <malini> cool..thx!
15:55:54 <kgriffs> ok
15:55:55 <kgriffs> training
15:55:59 <kgriffs> we touched on that already
15:56:15 <kgriffs> may be good to do some screencasts or something
15:56:25 <kgriffs> in addition to continuing to add to the docs
15:56:52 <kgriffs> performance
15:56:54 <kgriffs> that was suprising
15:57:05 <kgriffs> there seems to be some FUD going around on performance
15:57:30 <kgriffs> I don't know if someone else benchmarked and didn't tell us but started spreading rumors, or it is just a few people speculating
15:57:39 <kgriffs> which means we need to get the "truth" out there ASAP
15:57:41 <kgriffs> make sense?
15:57:55 <kgriffs> but in doing that...
15:58:12 <kgriffs> we need to be careful that people aren't expecting zeromq performance or something
15:59:10 <kgriffs> there are tons of use cases that work great at SQS performance levels
16:00:00 <kgriffs> finally, positioning... don't have time to touch on this, but randall was saying that the way we talk about Zaqar is important - when we say "messaging" they think oslo.messaging, AMQP, etc.
16:00:16 <kgriffs> which is a different set of use cases/goals from Zaqar
16:00:33 <kgriffs> he suggested talking more about "events" and such... I'm still thinking through that
16:02:02 <kgriffs> ok folks, looks like we are out of time
16:02:48 <kgriffs> #topic endmeeting
16:03:12 <kgriffs> I will try to move the heat feedback to the graduation pad
16:03:21 <kgriffs> #endmeeting