02:04:41 <flwang> #startmeeting zaqar
02:04:43 <openstack> Meeting started Tue Jun 27 02:04:41 2017 UTC and is due to finish in 60 minutes.  The chair is flwang. Information about MeetBot at http://wiki.debian.org/MeetBot.
02:04:44 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
02:04:46 <openstack> The meeting name has been set to 'zaqar'
02:05:02 <flwang> #topic roll call
02:05:04 <flwang> o/
02:06:23 <gecong_> hi
02:06:33 <gengchc> hi guys
02:07:08 <flwang> gecong_: thanks for joining
02:07:26 <gecong_> thanks
02:08:09 <wanghao> hi
02:08:14 <flwang> wanghao: hey
02:09:46 <flwang> wanghao: gecong_: here is the agenda https://etherpad.openstack.org/p/zaqar-meeting-agenda
02:10:14 <flwang> #topic service queue spec
02:10:33 <gecong_> ok
02:10:40 <flwang> I have asked theve, zaneb and flaper87 to review this spec
02:10:48 <flwang> somebody did, but somebody didn't yet
02:11:14 <flwang> we have 1 month before the milestone of pike-3
02:11:25 <wanghao> okay,  I talked lei-zh yesterday about the POC patch,  and there is a requirement to Zaqar.
02:11:38 <flwang> wanghao: cool, can you talk more?
02:12:03 <wanghao> searchlight want Zaqar can support to send messages to many queues one time.
02:12:07 <wanghao> like boradcast
02:12:46 <flwang> how?
02:12:47 <wanghao> it will send a list of projects in X-SERVICE-DELEGATED-PROJECT-ID.
02:13:18 <wanghao> and Zaqar need to post messages to those projects one by one.
02:13:20 <flwang> and then zaqar do the magic?
02:13:25 <flwang> NO
02:13:36 <flwang> that totally out of the scope of zaqar
02:13:56 <wanghao> emm, I think so
02:14:28 <flwang> yep, I would say "sorry, searchlight, Zaqar can't do that"
02:14:42 <flwang> it will  make the API dirty IMHO
02:14:57 <wanghao> technically， we can do it, but we shouldn't
02:15:14 <gecong_> I agree
02:15:22 <wanghao> okay,  I will talk about this with lei-zh
02:15:28 <flwang> I won't say I don't care the feature of service queue, but if we need to introduce such a ugly feature into zaqar just to support service queue, i will totally reject the service queue spec
02:16:09 <flwang> so in short, no, we can't support that
02:16:18 <wanghao> got it.
02:16:36 <gengchc> I agree too.
02:17:04 <flwang> I will push zaneb and therve to revisit the spec and hopefully we can get an agreement before next meeting
02:17:12 <wanghao> bascially, the case is when admin create a flavor,  searchlight need to notify every project.
02:17:14 <wanghao> okay
02:17:36 <gecong_> got it
02:17:44 <flwang> and another bad thing for now is, we still don't receive any feedback from horizon team, those interns working on this horizon feature
02:18:07 <flwang> so wanghao, could you please reply that thread and push horizon team?
02:18:22 <flwang> if they don't really care about this feature, please let us know
02:18:26 <wanghao> flwang: okay
02:18:31 <flwang> wanghao: thanks
02:18:58 <flwang> I got strong interest from the Horizon PTL when we met at Boston
02:19:13 <flwang> hence why I push this feature after that
02:19:41 <flwang> but I can't see much support from them now, so I'd like to confirm if they're still interested in this
02:19:52 <wanghao> yeah
02:20:43 <flwang> anything else for the service queue we need to talk?
02:21:04 <wanghao> not from me
02:21:30 <gecong_> I haven't any question now
02:22:31 <flwang> #topic notification policy
02:22:59 <wanghao> okay,  emm,  this I have committed a patch without ut yet.
02:23:23 <gecong_> I will review it later
02:23:31 <wanghao> thanks
02:23:47 <flwang> wanghao: I didn't go through the spec yet, but I will do it today for sure
02:23:51 <flwang> sorry again for the delay
02:23:59 <wanghao> flwang: np,  never mind
02:24:23 <wanghao> and I will test it to ensure it work well
02:24:35 <flwang> the only concern like I mentioned is we're adding more and more reserved keys into the queue metadata
02:25:06 <wanghao> yes, now we add a lot of things in metadata
02:25:10 <flwang> we may need to review all the reserved metadata keys and decide the long term strategy for this
02:25:39 <gecong_> yes
02:26:05 <wanghao> maybe we can consider to introduce some new resource for queue
02:26:19 <wanghao> that we can extract some metadata from queue.
02:26:42 <gecong_> good idea
02:27:41 <gecong_> maybe I can do something
02:28:17 <flwang> you mean new attribute/field for queue data model?
02:28:18 <wanghao> nice!
02:28:32 <wanghao> flwang: kind of
02:28:38 <flwang> wanghao: could be
02:29:42 <flwang> we should keep this in mind for now and revisit it in Queen
02:29:51 <wanghao> emm, agree
02:29:56 <gecong_> got it
02:30:04 <flwang> because currently, all users just use zaqar client to access zaqar api
02:30:20 <flwang> and with distil client, user actually don't really care how the data is saved
02:30:25 <flwang> so not too bad
02:31:14 <flwang> and before we can see more strong requirements, let's keep an eye and revisit later
02:31:32 <gecong_> emm, ok
02:31:33 <wanghao> okay
02:31:38 <flwang> wanghao: back to the notification policy spec, i really want to get it in Pike
02:31:40 <gengchc> ok
02:31:51 <wanghao> flwang: yeah, me too
02:31:52 <flwang> so let's work on it and make it happen
02:31:57 <wanghao> cool
02:32:13 <flwang> anything elese?
02:32:31 <wanghao> we plan to support only one backoff function in Pike, so it's not hard.
02:33:06 <flwang> wanghao: ah, i see.
02:33:30 <flwang> so we will mainly implement the framework and then support policies one by one, right?
02:34:02 <wanghao> yes,
02:34:04 <flwang> wanghao: cool, i see.
02:34:06 <gecong_> yeah
02:34:29 <flwang> anything else?
02:34:38 <wanghao> no
02:34:40 <flwang> #topic dead letter queue
02:34:43 <gecong_> is there anything I can do fot that
02:34:54 <flwang> gecong_: if you're willing to help
02:35:03 <flwang> maybe you can help the dead letter queue :)
02:35:10 <gecong_> yes I will
02:35:14 <gecong_> oK
02:35:18 <flwang> sorry, wanghao, i'm grabbing gecong_ for my feature ;)
02:35:24 <wanghao> np
02:35:26 <wanghao> haha
02:35:37 <gecong_> np
02:35:47 <flwang> dead letter queue spec has been there for a long time
02:35:52 <flwang> due to many reasons
02:36:02 <gecong_> I am very happy
02:36:06 <flwang> i'd like to get it done it pike to suppor the mongo backend
02:36:17 <flwang> as you can see in the spec
02:36:40 <gecong_> yeah
02:36:45 <wanghao> I think spec is almost closed,  not big issue for me.
02:36:48 <flwang> the simple idea is just move the message from the original queue to the dead letter queue
02:36:56 <flwang> wanghao: thanks
02:37:14 <gecong_> I have reviewed it
02:37:29 <flwang> so that said, i'd like to merge it asap and focus on the code
02:37:39 <flwang> for tiny issues, we can address them during the code review
02:37:46 <wanghao> +1
02:37:50 <flwang> and update the spec later
02:37:51 <gecong_> +1
02:38:07 <flwang> you know, the way of spec is good, but it's not really efficient
02:38:18 <flwang> we can't get a perfect design
02:38:20 <wanghao> yeah, some details in code.
02:38:36 <flwang> we will make some mistakes, leave some bugs
02:38:50 <flwang> but IMHO, it's fine, if we can get feedbacks, we can fix them
02:38:58 <gecong_> if there is anything need for me to do ,please let me and gengchc to help you
02:39:04 <flwang> if we can't get feedback, them it's nothing, right ? ;)
02:39:06 <wanghao> :) that's always happend
02:39:31 <flwang> gecong_: gengchc: so you're different people?
02:39:37 <gecong_> yes
02:39:38 <flwang> shit, i think they're same ids :D
02:39:46 <flwang> fantastic
02:39:52 <gecong_> I would like to introduce
02:39:58 <wanghao> nice
02:39:59 <gecong_> to your gus
02:40:11 <flwang> you guys come from zte? pls remind me your company name
02:40:20 <flwang> gecong_: pls pls pls
02:40:23 <gecong_> yes , he is also from zte
02:40:34 <flwang> pls introduce yourselves
02:41:22 <flwang> I'm really sorry for missing gengchc
02:41:30 <flwang> gengchc: waves from NZ
02:41:34 <gengchc> I am gengchangcai from zte company,i am interesting in zaqar.
02:41:35 <gecong_> please wait a minutes
02:41:46 <gengchc> We want to introduce zaqar in our project, but we do not know what aspects zaqar is  more suitable for.
02:42:26 <flwang> gengchc: as a service for your private cloud solution or for public cloud?
02:42:55 <gengchc> for private cloud
02:43:13 <flwang> gengchc: ok, cool
02:43:59 <flwang> now we're a small team, we do need more resources to make zaqar better
02:44:07 <flwang> so really appreciate for your joining
02:44:14 <wanghao> yes, welcome
02:44:22 <gecong_> welcome
02:44:33 <gengchc> thank you
02:44:34 <flwang> I'm Feilong Wang, and now I'm serving the PTL for 4th cycle
02:44:48 <flwang> I was working for IBM and moved to NZ 3 years ago
02:45:12 <flwang> now i'm working on a public cloud based in New Zealand as the dev manager
02:45:33 <flwang> and I also worked on Glance for the last 4 years
02:45:46 <flwang> wanghao: your turn
02:46:02 <gengchc> I have joined your wechat.
02:46:05 <wanghao> I'm wanghao, now working for Fiberhome,  and jumped from Huawei one years ago.
02:46:16 <gengchc> My wechat is 耿常才.
02:46:32 <flwang> gengchc: oh, cool, i will add you to our group
02:46:53 <wanghao> and I focus on Zaqar and Cinder for a long time.
02:47:16 <flwang> gengchc: i can't find your name :(
02:47:26 <gecong_> nice to meet you, feilong and wanghao
02:47:34 <flwang> cool, we can talk more offline
02:47:40 <flwang> now let's continue the meetint
02:47:52 <gecong_> ok
02:48:18 <flwang> gecong_: wanghao: pls revisit the dead letter queue spec and let's merge the spec soon
02:48:27 <flwang> and then i'm going to focus on the code for mongo backend
02:48:27 <wanghao> flwang: sure
02:48:34 <gengchc> I am sorry,my wechat is gcc5780437969
02:48:35 <gecong_> np
02:48:54 <flwang> and gecong_, you can work on the swift backend or redis, if you're interested in
02:49:02 <gecong_> np
02:49:34 <flwang> the last topic for today meeting is the Sydney summit topics
02:49:49 <flwang> we need to submit a topic
02:50:05 <flwang> #topic Sydney summit session
02:50:28 <flwang> i was thinking it would be nice if we can present the service queue feature with searchlight team and horizon team
02:50:37 <wanghao> yes, me too
02:50:42 <flwang> but I'm not too sure if we can achieve that in time
02:50:56 <wanghao> maybe just searchlight?
02:51:04 <flwang> so wanghao pls reply the email and push horizon guys to get feedback
02:51:14 <wanghao> em, sure
02:51:27 <flwang> without horizon support, it's not really attractive, TBH
02:51:41 <wanghao> emm, I see
02:51:52 <flwang> but meanwhile, we need to brainstorm if there is another topic we can propose
02:52:36 <flwang> gecong_: gengchc: do you have any customer requirement about zaqar?
02:52:58 <flwang> why ZTE are interested in zaqar?
02:53:28 <gecong_> for now ,not yet
02:54:03 <flwang> i see, because more of China customers are still in the stage of virtualization+, IMHO
02:54:19 <flwang> they just want to play with cloud, but they're not really sure how to use it
02:54:34 <flwang> and what's the workload they want to put on cloud
02:54:45 <gengchc> There is not yet, but we are studying how to use zaqar, the boss is very interested in it.
02:54:52 <flwang> and most of the workload just traditional legacy application
02:55:07 <flwang> not really distributed, cloud app
02:55:39 <flwang> yep, as you can see, all public cloud have queue/msg service
02:56:34 <flwang> anyway, gecong_, gengchc  I'm really happy to see ZTE it putting resource on zaqar and any contribution for the project will be really welcomed
02:56:48 <gengchc> is zaqar supported distributed nodes?
02:56:54 <flwang> all the people within the team are very friendly and positive
02:57:00 <flwang> feel free ask any question
02:57:07 <flwang> gengchc: sure
02:57:27 <flwang> gengchc: do you mean the distributed data storage of message?
02:57:41 <flwang> ok, we have 3 mins for open talk
02:57:47 <flwang> #topic open discussion
02:58:29 <flwang> gengchc: gecong_: any question?
02:59:30 <wanghao> okay time is up
02:59:40 <flwang> let's back to zaqar channel
02:59:42 <flwang> #endmeeting